1 2017-12-14 David Malcolm <dmalcolm@redhat.com>
3 PR tree-optimization/83312
4 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
5 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
6 "val" param, and to cope with arbitrary basic blocks.
7 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
8 handle NULL_TREE for "val", dropping "bb" param.
9 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
10 drop "bb" param. Handle NULL_TREE for "val".
11 (find_case_label_for_value): Make "switch_stmt" param const.
12 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
14 (vrp_prop::check_all_array_refs): Reimplement as...
15 (check_array_bounds_dom_walker::before_dom_children): ...this new
16 vfunc. Replace linear search through BB block list, excluding
17 those with non-executable in-edges via dominator walk.
19 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21 * config/arm/arm.opt (mverbose-cost-dump): New option.
22 * config/arm/arm.c (arm_rtx_costs): Use it.
24 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
27 * reload1.c (emit_input_reload_insns): Skip debug markers.
29 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
31 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
35 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
36 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
37 * config/sh/sh.c: Skip debug insns besides notes.
38 * config/sh/sh.md: Likewise.
39 * config/sh/sh_treg_combine.cc: Likewise.
40 * config/sh/sync.md: Likewise.
42 2017-12-14 Tom de Vries <tom@codesourcery.com>
44 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
45 Add item for weak_undefined.
47 2017-12-14 Richard Biener <rguenther@suse.de>
49 PR tree-optimization/67842
50 * sese.h (bb_in_region): Remove #if 0'ed code.
52 2017-12-14 Richard Biener <rguenther@suse.de>
54 PR tree-optimization/83326
55 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
56 may_be_zero parameter and handle it by not marking the first
57 peeled copy as not exiting the loop.
58 (try_peel_loop): Likewise.
59 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
60 to handle the case of constant or zero iterations and perform
61 loop header copying on-the-fly.
63 2017-12-14 Richard Biener <rguenther@suse.de>
65 PR tree-optimization/83418
66 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
67 Instead of asserting we don't get unfolded comparisons deal with
70 2017-12-14 Jakub Jelinek <jakub@redhat.com>
73 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
74 basic blocks. Assert debug bind insns don't appear outside of bbs,
75 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
78 PR tree-optimization/83198
79 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
80 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
81 value if arg is a REAL_CST with incompatible type.
83 2017-12-14 Sudakshina Das <sudi.das@arm.com>
84 Bin Cheng <bin.cheng@arm.com>
87 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
89 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
92 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
94 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
95 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
97 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
98 Alan Hayward <alan.hayward@arm.com>
99 David Sherwood <david.sherwood@arm.com>
101 * poly-int.h: New file.
102 * poly-int-types.h: Likewise.
103 * coretypes.h: Include them.
104 (POLY_INT_CONVERSION): Define.
105 * target.def (estimated_poly_value): New hook.
106 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
107 * doc/tm.texi: Regenerate.
108 * doc/poly-int.texi: New file.
109 * doc/gccint.texi: Include it.
110 * doc/rtl.texi: Describe restrictions on subreg modes.
111 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
112 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
113 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
114 * targhooks.h (default_estimated_poly_value): Declare.
115 * targhooks.c (default_estimated_poly_value): New function.
116 * target.h (estimated_poly_value): Likewise.
117 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
118 (wi::unary_traits): Delete.
119 (wi::binary_traits::signed_shift_result_type): Define for
120 offset_int << HOST_WIDE_INT, etc.
121 (generic_wide_int::operator <<=): Define for all types and use
122 wi::lshift instead of <<.
123 (wi::hwi_with_prec): Add a default constructor.
124 (wi::ints_for): New class.
125 (operator <<): Define for all wide-int types.
126 (operator /): New function.
127 (operator %): Likewise.
128 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
129 (ASSERT_MAYBE_NE_AT): New macros.
131 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
132 Dominik Vogt <vogt@linux.vnet.ibm.com>
135 * emit-rtl.c (init_emit): Remove ??? comment.
136 * explow.c (get_dynamic_stack_size): Take known alignment of stack
137 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
138 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
139 alignment of 3 virtual registers to BITS_PER_WORD.
141 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
143 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
145 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
146 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
148 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
151 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
153 2017-12-13 Jakub Jelinek <jakub@redhat.com>
155 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
156 with the exception of debug begin stmt markers appear before
160 * final.c (rest_of_handle_final): Call variable_tracking_main only
161 if !flag_var_tracking.
163 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
164 Jakub Jelinek <jakub@redhat.com>
168 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
169 labels debug stmts that can only appear after labels.
171 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
173 PR rtl-optimization/82398
174 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
175 EXPR_USEFULNESS in priority comparison.
177 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
179 PR rtl-optimization/83393
180 * combine.c (move_deaths): If reg_stat points to a too new insn in
181 last_death, do not use it: find the proper insn instead.
183 2017-12-12 Jeff Law <law@redhat.com>
185 PR tree-optimization/83298
186 PR tree-optimization/83362
187 PR tree-optimization/83383
188 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
189 push_value_range a public interface. Add new argument to
190 record_ranges_from_stmt.
191 * gimple-ssa-evrp-analyze.c
192 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
193 Update comments. Handle recording temporary equivalences.
194 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
195 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
196 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
197 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
198 gimple-ssa-evrp-analyze.h.
199 (record_temporary_equivalences_from_phis): Add new argument. When
200 the PHI arg is an SSA_NAME, set the result's range to the range
202 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
204 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
205 Pass it down to children as needed.
206 (thread_outgoing_edges): Likewise.
207 (thread_across_edge): Likewise. Push/pop range state as needed.
208 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
210 2017-12-12 Julia Koval <julia.koval@intel.com>
212 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
213 (PTA_CANNONLAKE): Remove PTA_CLWB.
215 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
218 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
219 (vcondv2div2df): Likewise.
220 (vconduv2dfv2di): Likewise.
222 2017-12-12 Jakub Jelinek <jakub@redhat.com>
224 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
225 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
226 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
227 ATTR_MATHFN_FPROUNDING_ERRNO.
229 2017-12-12 Richard Biener <rguenther@suse.de>
231 PR tree-optimization/83385
232 * graphite-scop-detection.c (get_order, order): Remove.
233 (bb_to_rpo): New global.
235 (build_scops): Sort pbbs in RPO order.
237 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
239 * combine.c (simplify_set): Do not transform subregs to zero_extends
240 if the destination mode is a vector mode.
242 2017-12-12 Jakub Jelinek <jakub@redhat.com>
244 PR tree-optimization/83359
245 * tree-cfg.h (fold_loop_internal_call): Declare.
246 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
247 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
248 (find_loop_dist_alias): New function.
249 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
250 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
253 PR tree-optimization/80631
254 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
255 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
256 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
257 hardcoding zero as the value if COND_EXPR is never true. For
258 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
259 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
260 hardcoding MAX_EXPR as the reduction operation.
261 (is_nonwrapping_integer_induction): Allow negative step.
262 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
263 vect_create_epilog_for_reduction, if no value is suitable, don't
264 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
266 2017-12-12 Richard Biener <rguenther@suse.de>
268 PR tree-optimization/81889
269 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
270 range info from the non-wrapping IV instead of just the range
273 2017-12-12 Julia Koval <julia.koval@intel.com>
275 * config.gcc: Add vaesintrin.h.
276 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
278 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
279 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
281 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
282 * config/i386/immintrin.h: Include vaesintrin.h.
283 * config/i386/sse.md (vaesdec_<mode>): New pattern.
284 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
285 _mm_aesdec_epi128): New intrinsics.
287 2017-12-12 Julia Koval <julia.koval@intel.com>
289 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
290 OPTION_MASK_ISA_VAES_UNSET): New.
291 (ix86_handle_option): Handle -mvaes.
292 * config/i386/cpuid.h: Define bit_VAES.
293 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
294 * config/i386/i386-c.c (__VAES__): New.
295 * config/i386/i386.c (ix86_target_string): Add -mvaes.
296 (ix86_valid_target_attribute_inner_p): Ditto.
297 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
298 * config/i386/i386.opt: Add -mvaes.
299 * doc/invoke.texi: Ditto.
301 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
303 * debug.h (gcc_debug_hooks): Add inline_entry.
304 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
305 * debug.c (do_nothing_debug_hooks): Likewise.
306 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
307 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
308 (dwarf2_lineno_debug_hooks): Likewise.
310 * common.opt (gstatement-frontiers): New, setting
311 debug_nonbind_markers_p.
312 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
313 * toplev.c (process_options): Autodetect value for debug statement
315 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
316 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
318 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
319 markers. Integrate source bind into debug stmt expand loop.
320 (pass_expand::execute): Check debug marker limit. Avoid deep
321 TER and expand debug locations for debug bind insns only.
322 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
324 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
325 * final.c (reemit_insn_block_notes): Take current block from
326 nonbind markers. Declare note where it's first set.
327 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
328 begin stmt markers if enabled.
329 (notice_source_line): Handle nonbind markers. Fail if their
330 location is unknown or that of builtins.
331 (rest_of_handle_final): Convert begin stmt markers to notes if
332 var-tracking didn't run.
333 (rest_of_clean_state): Skip begin stmt markers.
334 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
336 * function.c (allocate_struct_function): Set begin_stmt_markers.
337 * function.h (struct function): Add debug_marker_count counter
338 and debug_nonbind_markers flag.
339 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
340 * gimple-low.c (lower_function_body): Adjust
341 debug_nonbind_markers.
342 (lower_stmt): Drop or skip gimple debug stmts.
343 (lower_try_catch): Skip debug stmts.
344 * gimple.c (gimple_build_debug_begin_stmt): New.
345 (gimple_copy): Increment debug_marker_count if copying one.
346 * gimple.h (gimple_build_debug_begin_stmt): Declare.
347 * gimplify.c (rexpr_location): New.
348 (rexpr_has_location): New.
349 (warn_switch_unreachable_r): Handle gimple debug stmts.
350 (shortcut_cond_r): Call expr_location.
352 (find_goto_label): New.
353 (shortcut_cond_expr): Call expr_has_location, expr_location, and
355 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
357 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
358 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
359 (LANG_HOOKS_INITIALIZER): ... this.
360 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
361 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
363 (update_ebb_live_info): Skip debug insn markers.
364 * lra.c (debug_insn_static_data): Rename to...
365 (debug_bind_static_data): ... this.
366 (debug_marker_static_data): New.
367 (lra_set_insn_recog_data): Select one of the above depending
369 (lra_update_isn_regno_info): Don't assume debug insns have
371 (push_insns): Skip debug insns.
372 * lto-streamer-in.c (input_function): Drop debug stmts
373 depending on active options. Adjust debug_nonbind_markers.
374 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
375 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
376 begin stmt marker notes.
377 (print_insn): Likewise.
378 * recog.c (extract_insn): Recognize rtl for debug markers.
379 * rtl.def (DEBUG_MARKER): New.
380 * tree-inline.c: Include params.h.
381 (remap_gimple_stmt): Handle nonbind markers.
382 (maybe_move_debug_stmts_to_successors): Likewise.
383 (copy_debug_stmt): Likewise.
384 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
385 markers regardless of no side effects.
386 (tsi_link_before): Don't update container's side effects when adding
388 (tsi_link_after): Likewise.
389 (expr_first): Skip begin stmt markers.
390 (expr_last): Likewise.
391 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
392 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
393 Disregard nonbind markers.
394 * tree.c (make_node_stat): Don't set side effects for begin stmt
396 (build1_stat): Likewise.
397 * tree.def (DEBUG_BEGIN_STMT): New.
398 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
399 * var-tracking.c (delete_debug_insns): Renamed to...
400 (delete_vta_debug_insns): ... this.
401 (reemit_marker_as_note): New.
402 (vt_initialize): Reemit markers.
403 (delete_vta_debug_insns): Likewise.
404 (vt_debug_insns_local): Reemit or delete markers.
405 (variable_tracking_main_1): Likewise.
406 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
407 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
408 (gimple_debug_nonbind_marker_p): New.
409 (gimple_build_debug_bind): Adjust.
410 (gimple_build_debug_begin_stmt): New.
411 * doc/invoke.texi (max-debug-marker-count): New param.
412 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
413 (debug_parameter_ref, debug_marker): New.
414 (NOTE_INSN_BEGIN_STMT): New.
415 (DEBUG_INSN): Describe begin stmt markers.
417 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
418 without debug insns, we wouldn't, but clean up debug insns
419 after a control flow insn nevertheless.
421 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
422 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
423 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
424 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
425 (rtl_tidy_fallthru_edge): Likewise.
426 (rtl_verify_fallthru): Likewise.
427 (rtl_verify_bb_layout): Likewise.
428 (skip_insns_after_block): Likewise.
429 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
430 * dwarf2out.c: Include print-rtl.h.
431 (dwarf2out_next_real_insn): New.
432 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
433 Dump debug binds in asm comments.
434 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
435 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
436 callers to use gsi_start_nondebug_bb instead.
437 (gsi_after_labels): Skip gimple debug stmts.
438 (gsi_start_nondebug): New.
439 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
440 (proper_loop_form_for_interchange): Adjust.
441 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
442 * gimple.h (gimple_seq_last_nondebug_stmt): New.
443 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
444 (collect_fallthrough_labels): Likewise.
445 (should_warn_for_implicit_fallthrough): Likewise.
446 (warn_implicit_fallthrough_r): Likewise.
447 (expand_FALLTHROUGH_r): Likewise.
448 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
449 (graphite_copy_stmts_from_block): Skip nonbind markers.
450 * haifa-sched.c (sched_extend_bb): Skip debug insns.
451 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
452 * jump.c (clean_barriers): Skip debug insns.
453 * omp-expand.c (expand_parallel_call): Skip debug insns.
454 (expand_task_call): Likewise.
455 (remove_exit_barrier): Likewise.
456 (expand_omp_taskreg): Likewise.
457 (expand_omp_for_init_counts): Likewise.
458 (expand_omp_for_generic): Likewise.
459 (expand_omp_for_static_nochunk): Likewise.
460 (expand_omp_for_static_chunk): Likewise.
461 (expand_omp_simd): Likewise.
462 (expand_omp_taskloop_for_outer): Likewise.
463 (expand_omp_taskloop_for_inner): Likewise.
464 (expand_oacc_for): Likewise.
465 (expand_omp_sections): Likewise.
466 (expand_omp_single): Likewise.
467 (expand_omp_synch): Likewise.
468 (expand_omp_atomic_load): Likewise.
469 (expand_omp_atomic_store): Likewise.
470 (expand_omp_atomic_fetch_op): Likewise.
471 (expand_omp_atomic_pipeline): Likewise.
472 (expand_omp_atomic_mutex): Likewise.
473 (expand_omp_target): Likewise.
474 (grid_expand_omp_for_loop): Likewise.
475 (grid_expand_target_grid_body): Likewise.
476 (build_omp_regions_1): Likewise.
477 * omp-low.c (check_combined_parallel): Skip debug stmts.
478 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
479 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
480 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
482 * tree-cfg.c (make_blobs_1): Skip debug stmts.
483 (make_edges): Likewise.
484 (cleanup_dead_labels): Likewise.
485 (gimple_can_merge_blocks_p): Likewise.
486 (stmt_starts_bb_p): Likewise.
487 (gimple_block_label): Likewise.
488 (gimple_redirect_edge_and_branch): Likewise.
489 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
491 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
493 * tree-pretty-print (print_declaration): Omit initializer in slim
495 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
497 (eliminate_unnecessary_stmts): Stabilize block removal order.
498 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
499 * var-tracking.c (get_first_insn): New.
500 (vt_emit_notes): Call it.
501 (vt_initialize): Walk any insns before the first BB.
502 (delete_debug_insns): Likewise.
504 * gimple.h (enum gimple_debug_subcode): Add
505 GIMPLE_DEBUG_BEGIN_STMT.
506 (gimple_debug_begin_stmt_p): New.
507 (gimple_debug_nonbind_marker_p): New.
508 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
509 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
510 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
511 * insn-notes.def (BEGIN_STMT): New.
512 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
513 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
514 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
515 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
516 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
517 (INSN_DEBUG_MARKER_KIND): New.
518 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
519 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
520 (INSN_VAR_LOCATION_PTR): New.
521 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
523 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
524 deps only if debug bind insns are enabled.
525 (pass_expand::execute): Avoid deep TER and expand
526 debug locations for debug bind insns only.
527 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
528 debug stmts special handling down to debug bind stmts.
529 * combine.c (try_combine): Narrow debug insns special handling
530 down to debug bind insns.
531 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
532 Narrow debug insns preexisting special handling down to debug
534 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
535 handling down to debug bind insns.
536 * function.c (instantiate_virtual_regs): Skip debug markers,
537 adjust handling of debug binds.
538 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
539 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
540 * haifa-sched.c (schedule_insn): Narrow special handling of debug
541 insns to debug bind insns.
542 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
543 special handling of debug stmts to debug bind stmts.
544 * ipa-split.c (split_function): Likewise.
545 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
546 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
548 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
549 * regrename.c (build_def_use): Likewise.
550 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
551 (pass_cprop_hardreg): Narrow special casing of debug insns to
553 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
554 * reload1.c (reload): Likewise.
555 * sese.c (sese_insert_phis_for_liveouts): Narrow special
556 casing of debug stmts to debug bind stmts.
557 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
558 * ssa-iterators.h (num_imm_uses): Likewise.
559 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
560 debug stmts to debug bind stmts.
561 * tree-inline.c (tree_function_versioning): Narrow special casing
562 of debug stmts to debug bind stmts.
563 * tree-loop-distribution.c (generate_loops_for_partition):
564 Narrow special casing of debug stmts to debug bind stmts.
565 * tree-sra.c (analyze_access_subtree): Narrow special casing
566 of debug stmts to debug bind stmts.
567 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
568 stmts to debug bind stmts.
569 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
570 casing of debug stmts to debug bind stmts.
571 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
572 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
573 casing of debug stmts to debug bind stmts.
574 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
576 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
577 debug stmts to debug bind stmts.
578 (gimple_replace_ssa_lhs): Likewise.
579 (insert_debug_temp_for_var_def): Likewise.
580 (insert_debug_temps_for_defs): Likewise.
581 (reset_debug_uses): Likewise.
582 * tree-ssanames.c (release_ssa_name_fn): Likewise.
583 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
584 (adjust_debug_stmts): Likewise.
585 (adjust_phi_and_debug_stmts): Likewise.
586 (vect_do_peeling): Likewise.
587 * tree-vect-loop.c (vect_transform_loop): Likewise.
588 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
589 * var-tracking.c (adjust_mems): Narrow special casing of debug
590 insns to debug bind insns.
591 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
592 (compute_bb_dataflow, vt_find_locations): Likewise.
593 (vt_expand_loc, emit_notes_for_changes): Likewise.
594 (vt_init_cfa_base): Likewise.
595 (vt_emit_notes): Likewise.
596 (vt_initialize): Likewise.
597 (vt_finalize): Likewise.
599 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
600 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
601 (next_nonnote_nondebug_insn_bb): New.
602 (prev_nonnote_nondebug_insn_bb): New.
603 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
604 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
605 (prev_nonnote_nondebug_insn_bb): Declare.
606 (next_nonnote_nondebug_insn_bb): Declare.
607 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
608 * cfgrtl.c (get_last_bb_insn): Likewise.
609 * lra.c (push_insns): Likewise.
611 2017-12-11 David Malcolm <dmalcolm@redhat.com>
614 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
616 * selftest.c (selftest::test_runner::test_runner): New ctor.
617 (selftest::test_runner::~test_runner): New dtor.
618 * selftest.h (class selftest::test_runner): New class.
620 2017-12-11 Carl Love <cel@us.ibm.com>
622 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
623 vec_extract_fp32_from_shortl]): Add #defines.
624 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
625 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
626 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
627 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
628 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
629 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
630 * doc/extend.texi: Add documentation for the added builtins.
632 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
634 PR rtl-optimization/80693
635 PR rtl-optimization/81019
636 PR rtl-optimization/81020
637 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
638 are not mentioned in i3. Place the REG_UNUSED note on i2,
639 possibly modified to REG_DEAD, if it did not originate in i3.
641 2017-12-11 Jakub Jelinek <jakub@redhat.com>
643 * recog.c (store_data_bypass_p_1): New function.
644 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
645 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
647 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
649 PR rtl-optimization/83361
650 * ifcvt.c (if_convert): Call fixup_partitions.
652 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
654 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
655 early folding of splat_u{8,16,32}.
657 2017-12-11 Jakub Jelinek <jakub@redhat.com>
659 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
660 output_operand_lossage first argument with capital letter.
661 (aarch64_override_options): Don't start error and sorry first argument
664 2017-12-11 Andi Kleen <ak@linux.intel.com>
666 PR gcov-profile/83355
667 * auto-profile.c (string_table::get_index_by_decl): Don't
668 recurse when abstract origin points to itself.
670 2017-12-11 Bin Cheng <bin.cheng@arm.com>
672 PR tree-optimization/83320
673 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
674 (prune_datarefs_not_in_loop): Ditto.
676 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
678 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
680 2017-12-10 Jakub Jelinek <jakub@redhat.com>
682 PR tree-optimization/83337
683 * gimple-loop-interchange.cc (compute_access_stride): Handle
684 bitfield DRs properly.
686 2017-12-09 Jakub Jelinek <jakub@redhat.com>
688 PR tree-optimization/83338
689 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
690 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
691 vector of unsigned integers to vector of signed integers.
693 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
695 PR rtl-optimization/83317
696 * lra-constraints.c (process_address_1): Add insn code check.
698 2017-12-08 Michael Matz <matz@suse.de>
700 Fix PR tree-optimization/83323
701 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
702 head-controlled loops and loop BBs.
703 * common.opt (funroll-and-jam): Remove, instead ...
704 (floop-unroll-and-jam): ... reuse this option.
705 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
706 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
707 (-floop-unroll-and-jam): ... this option.
709 2017-12-08 Jakub Jelinek <jakub@redhat.com>
711 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
712 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
713 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
715 2017-12-08 Richard Biener <rguenther@suse.de>
718 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
719 handle accesses outside of zero-sized vars.
721 2017-12-08 Martin Jambor <mjambor@suse.cz>
723 PR tree-optimization/83141
724 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
725 test for MEM_REFs implicitely changing types with padding. Remove
727 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
729 2017-12-08 Jakub Jelinek <jakub@redhat.com>
731 * config/arc/arc.c (arc_attribute_table): Add exclusions to
733 * config/avr/avr.c (avr_attribute_table): Likewise.
734 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
735 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
736 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
737 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
738 of struct attribute_spec.
739 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
741 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
744 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
746 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
749 * profile-count.c (profile_count::from_gcov_type): Move from
750 profile-count.h; handle overflow.
751 * profile-count.h (profile_count::from_gcov_type): Move offline.
753 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
755 PR rtl-optimization/83304
756 * combine.c (move_deaths): If we do not know where a register died,
759 2017-12-08 Richard Biener <rguenther@suse.de>
761 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
762 Provide -fopt-info-loop feedback when we interchange in a nest.
764 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
766 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
767 for armv6 ARM CPU IDs.
769 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
771 * common/config/arm/arm-common.c: Include <algorithm>.
772 (INCLUDE_VECTOR): Define.
773 (compare_opt_names): New function.
774 (arm_rewrite_selected_arch): Only strip out extensions that can be
775 expressed through -mfpu. Sort the remaining extensions
778 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
780 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
781 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
782 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
783 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
785 2017-06-08 Tristan Gingold <gindold@adacore.com>
788 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
789 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
791 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
793 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
794 of struct attribute_spec.
796 2017-12-08 Julia Koval <julia.koval@intel.com>
798 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
799 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
800 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
801 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
802 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
803 _mm_maskz_dpwssds_epi32): Ditto.
805 2017-12-08 Richard Biener <rguenther@suse.de>
807 PR tree-optimization/81303
808 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
809 conditions try to create a comparison vector type matching
810 the data vector type.
811 (vectorizable_condition): Adjust.
812 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
813 Leave invariant conditions alone in case we can vectorize those.
815 2017-12-08 Julia Koval <julia.koval@intel.com>
817 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
818 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
819 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
820 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
821 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
823 2017-12-08 Julia Koval <julia.koval@intel.com>
825 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
826 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
827 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
828 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
829 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
830 _mm_maskz_dpbusds_epi32): New intrinsics.
832 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
834 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
836 (ld<bh>io_signed): Likewise.
837 (st<bhw>io): Likewise.
838 * config/nios2/predicates.md (ldstio_memory_operand): Allow
839 SMALL_INT12 constant integer operand.
841 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
843 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
845 (nios2_large_constant_allowed): ...this. Adjust uses.
846 (nios2_plus_symbolic_constant_p): Rename to...
847 (nios2_plus_large_constant_p): ...this. Adjust uses.
848 (nios2_legitimate_address_p): Correct CONST_INT handling.
849 (nios2_symbolic_memory_operand_p): Rename to...
850 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
851 (nios2_large_constant_p): Check for large constant integers too.
852 (nios2_split_large_constant): Handle constant integers.
853 (nios2_split_symbolic_memory_operand): Rename to...
854 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
855 (nios2_legitimize_constant_address): Handle constant integers.
856 (r0rel_constant_p): Handle small constant integers.
857 (nios2_print_operand_address): Handle r0-relative integer addresses.
858 * config/nios2/nios2-protos.h: Adjust for renamed functions.
859 * config/nios2/nios2.md: Adjust for renamed functions.
861 2017-12-07 Andrew Waterman <andrew@sifive.com>
863 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
864 (TARGET_HAVE_SRODATA_SECTION): New define.
865 (riscv_select_section): New function.
867 2017-12-08 Joseph Myers <joseph@codesourcery.com>
868 Alexander Monakov <amonakov@ispras.ru>
869 Jakub Jelinek <jakub@redhat.com>
872 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
874 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
876 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
877 a bool success value. Don't call output_operand_lossage here.
878 (aarch64_print_ldpstp_address): Return a bool success value.
879 (aarch64_print_operand_address): Call output_addr_const if
880 aarch64_print_address_internal fails.
881 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
882 'y'; call output_operand_lossage instead. Call output_operand_lossage
883 if aarch64_print_ldpstp_address fails.
885 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
887 * tree-vector-builder.h
888 (tree_vector_builder::binary_encoded_nelts): Declare.
889 * tree-vector-builder.c
890 (tree_vector_builder::binary_encoded_nelts): New function.
891 * fold-const.c (negate_expr_p): Likewise.
892 (operand_equal_p, fold_checksum_tree): Likewise.
893 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
894 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
895 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
896 (uniform_vector_p): Likewise.
897 * varasm.c (const_hash_1, compare_constant): Likewise.
898 * tree-ssa-ccp.c: Include tree-vector-builder.h.
899 (valid_lattice_transition): Operate directly on the VECTOR_CST
901 * ipa-icf.c: Include tree-vector-builder.h.
902 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
903 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
905 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
907 * tree.c (build_vector): Delete.
908 * tree.h (build_vector): Make static and move into the self-testing
911 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
913 * vector-builder.h (vector_builder::derived): New const overload.
914 (vector_builder::elt): New function.
915 * tree-vector-builder.h (tree_vector_builder::type): New function.
916 (tree_vector_builder::apply_step): Declare.
917 * tree-vector-builder.c (tree_vector_builder::apply_step): New
919 * gimple-fold.h (tree_vector_builder): Declare.
920 (gimple_build_vector): Take a tree_vector_builder instead of a
921 type and vector of elements.
922 * gimple-fold.c (gimple_build_vector): Likewise.
923 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
925 (get_initial_defs_for_reduction): Likewise.
926 (vectorizable_induction): Likewise.
928 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
930 * tree-vector-builder.h
931 (tree_vector_builder::new_binary_operation): Declare.
932 * tree-vector-builder.c
933 (tree_vector_builder::new_binary_operation): New function.
934 * fold-const.c (fold_relational_const): Use it.
935 (const_binop): Likewise. Check that both input vectors have
936 the same number of elements, thus excluding things like WIDEN_SUM.
937 Check whether it is possible to operate directly on the encodings
940 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
942 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
943 new_unary_operation, operating only on the encoded elements.
944 (const_unop): Likewise.
945 (exact_inverse): Likewise.
946 (distributes_over_addition_p): New function.
947 (const_binop): Use tree_vector_builder and new_unary_operation
948 for combinations of VECTOR_CST and INTEGER_CST. Operate only
949 on the encoded elements unless the encoding is strided and the
950 operation does not distribute over addition.
951 (fold_convert_const): Use tree_vector_builder and
952 new_unary_operation. Operate only on the encoded elements
953 for truncating integer conversions, or for non-stepped encodings.
955 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
957 * config/sparc/sparc.c: Include tree-vector-builder.h.
958 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
959 * expmed.c: Include tree-vector-builder.h.
960 (make_tree): Use tree_vector_builder instead of build_vector.
961 * fold-const.c: Include tree-vector-builder.h.
962 (const_binop): Use tree_vector_builder instead of build_vector.
963 (const_unop): Likewise.
964 (native_interpret_vector): Likewise.
965 (fold_vec_perm): Likewise.
966 (fold_ternary_loc): Likewise.
967 * gimple-fold.c: Include tree-vector-builder.h.
968 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
970 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
971 (simplify_vector_constructor): Use tree_vector_builder instead
973 * tree-vect-generic.c: Include tree-vector-builder.h.
974 (add_rshift): Use tree_vector_builder instead of build_vector.
975 (expand_vector_divmod): Likewise.
976 (optimize_vector_constructor): Likewise.
977 * tree-vect-loop.c: Include tree-vector-builder.h.
978 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
979 of build_vector. Explicitly use a stepped encoding for
981 * tree-vect-slp.c: Include tree-vector-builder.h.
982 (vect_get_constant_vectors): Use tree_vector_builder instead
984 (vect_transform_slp_perm_load): Likewise.
985 (vect_schedule_slp_instance): Likewise.
986 * tree-vect-stmts.c: Include tree-vector-builder.h.
987 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
988 (vect_gen_perm_mask_any): Likewise.
989 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
990 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
992 (build_vector_from_val): Likewise. Explicitly use a duplicate
995 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
997 * doc/generic.texi (VECTOR_CST): Describe new representation of
999 * vector-builder.h: New file.
1000 * tree-vector-builder.h: Likewise.
1001 * tree-vector-builder.c: Likewise.
1002 * Makefile.in (OBJS): Add tree-vector-builder.o.
1003 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
1004 * tree-core.h (tree_base): Add a vector_cst field to the u union.
1005 (tree_vector): Change the number of elements to
1006 vector_cst_encoded_nelts.
1007 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
1008 (VECTOR_CST_ELTS): Delete.
1009 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
1010 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
1011 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
1012 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
1013 (VECTOR_CST_ENCODED_ELT): Likewise.
1014 (vector_cst_encoded_nelts): New function.
1015 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
1016 VECTOR_CST_NELTS_PER_PATTERN as arguments.
1017 (vector_cst_int_elt, vector_cst_elt): Declare.
1018 * tree.c: Include tree-vector-builder.h.
1019 (tree_code_size): Abort if passed VECTOR_CST.
1020 (tree_size): Update for new VECTOR_CST layout.
1021 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
1022 VECTOR_CST_NELTS_PER_PATTERN as arguments.
1023 (build_vector): Use tree_vector_builder.
1024 (vector_cst_int_elt, vector_cst_elt): New functions.
1025 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
1026 encoded elements and then create the vector in the canonical form.
1027 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
1028 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
1029 (tree_c_tests): Call test_vector_cst_patterns.
1030 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
1032 (hash_tree): Likewise.
1033 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
1034 (streamer_write_tree_header): Likewise.
1035 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
1036 (streamer_alloc_tree): Likewise. Update call to make_vector.
1037 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
1039 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1041 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
1042 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
1043 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
1045 2017-12-07 Bin Cheng <bin.cheng@arm.com>
1046 Richard Biener <rguenther@suse.de>
1048 PR tree-optimization/81303
1049 * Makefile.in (gimple-loop-interchange.o): New object file.
1050 * common.opt (floop-interchange): Reuse the option from graphite.
1051 * doc/invoke.texi (-floop-interchange): Ditto. New document for
1052 -floop-interchange and mention it for -O3.
1053 * opts.c (default_options_table): Enable -floop-interchange at -O3.
1054 * gimple-loop-interchange.cc: New file.
1055 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
1056 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
1057 * passes.def (pass_linterchange): New pass.
1058 * timevar.def (TV_LINTERCHANGE): New time var.
1059 * tree-pass.h (make_pass_linterchange): New declaration.
1060 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
1061 interchange. Record IV before/after increment in new parameters.
1062 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
1063 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
1065 (check_reduction_path): ...New function here.
1066 * tree-vectorizer.h (check_reduction_path): New declaration.
1068 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
1071 PR rtl-optimization/80818
1072 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
1073 always early clobbered.
1074 * lra-lives.c (process_bb_lives): Check input hard regs for early
1075 clobbered non-operand hard reg.
1077 2017-12-07 Jakub Jelinek <jakub@redhat.com>
1080 * tree-cfg.c (verify_gimple_assign_binary): Don't require
1081 types_compatible_p, just that TYPE_MODE is the same.
1083 2017-12-07 Martin Sebor <msebor@redhat.com>
1086 * attribs.c (empty_attribute_table): Initialize new member of
1087 struct attribute_spec.
1088 (decl_attributes): Add argument. Handle mutually exclusive
1089 combinations of attributes.
1090 (selftests::test_attribute_exclusions): New function.
1091 (selftests::attribute_c_tests): Ditto.
1092 * attribs.h (decl_attributes): Add default argument.
1093 * selftest.h (attribute_c_tests): Declare.
1094 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
1095 * tree-core.h (attribute_spec::exclusions, exclude): New type and
1097 * doc/extend.texi (Common Function Attributes): Update const and pure.
1098 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
1099 of struct attribute_spec.
1100 * config/arc/arc.c (arc_attribute_table): Same.
1101 * config/arm/arm.c (arm_attribute_table): Same.
1102 * config/avr/avr.c ( avr_attribute_table): Same.
1103 * config/bfin/bfin.c (bfin_attribute_table): Same.
1104 * config/cr16/cr16.c (cr16_attribute_table): Same.
1105 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
1106 * config/h8300/h8300.c (h8300_attribute_table): Same.
1107 * config/i386/i386.c (ix86_attribute_table): Same.
1108 * config/ia64/ia64.c (ia64_attribute_table): Same.
1109 * config/m32c/m32c.c (m32c_attribute_table): Same.
1110 * config/m32r/m32r.c (m32r_attribute_table): Same.
1111 * config/m68k/m68k.c (m68k_attribute_table): Same.
1112 * config/mcore/mcore.c (mcore_attribute_table): Same.
1113 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
1114 * config/mips/mips.c (mips_attribute_table): Same.
1115 * config/msp430/msp430.c (msp430_attribute_table): Same.
1116 * config/nds32/nds32.c (nds32_attribute_table): Same.
1117 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
1118 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
1119 * config/rl78/rl78.c (rl78__attribute_table): Same.
1120 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
1121 * onfig/rx/rx.c (rx_attribute_table): Same.
1122 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
1123 * config/sh/sh.c (sh_attribute_table): Same.
1124 * config/sparc/sparc.c (sparc_attribute_table): Same.
1125 * config/spu/spu.c (spu_attribute_table): Same.
1126 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
1127 * config/v850/v850.c (v850_attribute_table): Same.
1128 * config/visium/visium.c (visium_attribute_table): Same.
1130 2017-12-07 Tamar Christina <tamar.christina@arm.com>
1133 * config/arm/arm.c (INCLUDE_STRING): Define.
1134 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
1135 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
1138 2017-12-07 Michael Matz <matz@suse.de>
1140 Add unroll and jam pass
1142 * gimple-loop-jam.c: New file.
1143 * Makefile.in (OBJS): Add gimple-loop-jam.o.
1144 * common.opt (funroll-and-jam): New option.
1145 * opts.c (default_options_table): Add unroll-and-jam at -O3.
1146 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
1147 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
1148 * passes.def: Add pass_loop_jam.
1149 * timevar.def (TV_LOOP_JAM): Add.
1150 * tree-pass.h (make_pass_loop_jam): Declare.
1151 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
1152 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
1153 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
1154 to flow_loop_tree_node_add.
1155 (duplicate_subloops, copy_loops_to): Append to sibling list.
1156 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
1157 * doc/invoke.texi (-funroll-and-jam): Document new option.
1158 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
1160 2017-12-07 Richard Biener <rguenther@suse.de>
1162 PR tree-optimization/83296
1163 PR tree-optimization/67769
1164 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
1165 flow sensitive info in an unrelated BB.
1166 (value_replacement): Use reset_flow_sensitive_info.
1167 (minmax_replacement): Reset flow sensitive info on the def
1168 we move. Do not reset flow sensitive info in the whole BB
1169 we move the stmt to.
1170 (abs_replacement): Likewise.
1172 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1175 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1176 rs6000_cpu to the given -mcpu=, or to the default processor.
1178 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1180 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
1181 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
1182 instead of rs6000_cpu_attr.
1183 (rs6000_adjust_cost): Ditto.
1184 (is_microcoded_insn): Ditto.
1185 (rs6000_adjust_priority): Ditto.
1186 (rs6000_issue_rate): Ditto.
1187 (rs6000_use_sched_lookahead): Ditto.
1188 (rs6000_use_sched_lookahead_guard): Ditto.
1189 (rs6000_sched_reorder): Ditto.
1190 (force_new_group): Ditto.
1191 * config/rs6000/rs6000.md (cpu attribute): Ditto.
1192 (group_ending_nop): Ditto.
1194 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1196 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
1197 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
1198 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
1199 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
1201 (rs6000_reassociation_width): Ditto.
1202 (rs6000_emit_epilogue): Ditto.
1203 (rs6000_adjust_cost): Ditto.
1204 (is_microcoded_insn): Ditto.
1205 (is_cracked_insn): Ditto.
1206 (rs6000_adjust_priority): Ditto.
1207 (rs6000_sched_reorder): Ditto.
1208 (rs6000_sched_reorder2): Ditto.
1209 (insn_must_be_first_in_group): Ditto.
1210 (insn_must_be_last_in_group): Ditto.
1211 (rs6000_register_move_cost): Ditto.
1212 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
1215 2017-12-07 Julia Koval <julia.koval@intel.com>
1217 * config.gcc: Add vaesintrin.h.
1218 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
1220 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
1221 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
1223 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1224 * config/i386/immintrin.h: Include vaesintrin.h.
1225 * config/i386/sse.md (vaesdec_<mode>): New pattern.
1226 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
1227 _mm_aesdec_epi128): New intrinsics.
1229 2017-12-06 David Malcolm <dmalcolm@redhat.com>
1231 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
1232 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
1233 c-family/c-spellcheck.cc.
1234 (best_macro_match::best_macro_match): Likewise.
1236 (struct edit_distance_traits<cpp_hashnode *>): Move to
1237 c-family/c-spellcheck.h.
1238 (class best_macro_match): Likewise.
1240 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1242 PR tree-optimization/83293
1243 * gimple-ssa-strength-reduction.c (insert_initializers): Use
1244 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
1245 might insert into empty bb.
1248 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
1249 simplify for plus with :c added, and pointer_plus without that.
1250 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
1251 with undefined overflow and the conversion is not widening,
1252 perform negation in utype and only convert to type afterwards.
1253 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
1254 simplify for plus with :c added, and pointer_plus without that.
1255 If type is integral with undefined overflow and the conversion is
1256 not widening, perform minus in utype and only convert to type
1257 afterwards. Move the last pointer_diff_expr simplify into the
1260 2017-12-06 Martin Sebor <msebor@redhat.com>
1262 PR tree-optimization/82646
1263 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
1264 strncpy, not maxlen.
1266 2017-12-06 Martin Sebor <msebor@redhat.com>
1268 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
1271 PR tree-optimization/83075
1272 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
1273 strncat/strncpy don't change length of source string.
1275 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
1278 2017-11-29 Martin Aberg <maberg@gaisler.com>
1280 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1281 to prevent b2bst errata sequence.
1282 (sqrtdf2_fix): Likewise.
1284 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1286 PR tree-optimization/81945
1287 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
1288 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
1289 to dest_cfun has orig_loop_num set, either remap it to the new
1290 loop number if the loop got moved too, or clear it.
1292 2017-12-05 Steve Ellcey <sellcey@cavium.com>
1294 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
1296 (thunderx2t99_nothing): New insn reservation.
1297 (thunderx2t99_mrs): New insn reservation.
1298 (thunderx2t99_multiple): New insn reservation.
1299 (thunderx2t99_alu_basi): Add bfx to reservation.
1300 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
1302 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1305 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
1307 2017-12-05 Bin Cheng <bin.cheng@arm.com>
1309 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
1310 tree-ssa-pre.c::remove_dead_inserted_code.
1311 * tree-ssa-dce.h: New file.
1312 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
1313 (remove_dead_inserted_code): Move and rename to function
1314 tree-ssa-dce.c::simple_dce_from_worklist.
1315 (pass_pre::execute): Update use.
1317 2017-12-05 Richard Biener <rguenther@suse.de>
1319 PR tree-optimization/83277
1320 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
1321 to code-gen liveout vars.
1323 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
1325 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
1326 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
1327 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
1330 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
1332 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
1334 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
1335 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
1338 2017-12-05 Richard Biener <rguenther@suse.de>
1340 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
1341 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
1342 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
1343 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
1345 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1347 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
1349 2017-12-05 Martin Liska <mliska@suse.cz>
1350 Jakub Jelinek <jakub@redhat.com>
1352 * doc/invoke.texi: Document the options.
1353 * flag-types.h (enum sanitize_code): Add
1354 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1355 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
1356 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1357 * opts.c: Define new sanitizer options.
1358 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
1359 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
1361 2017-12-05 Julia Koval <julia.koval@intel.com>
1363 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
1364 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
1365 (ix86_handle_option): Handle -mavx512vnni.
1366 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
1367 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
1368 * config/i386/i386-c (__AVX512VNNI__): New.
1369 * config/i386/i386.c (ix86_target_string): Handle new option.
1370 (ix86_valid_target_attribute_inner_p): Handle new option.
1371 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
1372 * config/i386/i386.opt (mavx512vnni): New option.
1374 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1377 * athlon.md: Disable for generic.
1378 * haswell.md: Enable for generic.
1379 * i386.c (ix86_sched_init_global): Add core hooks for generic.
1380 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
1382 (ix86_adjust_cost): Move generic to haswell path.
1384 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
1386 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
1387 instead of MEM_P in a couple more places. Fix formatting issues.
1389 2017-12-04 Jim Wilson <jimw@sifive.com>
1391 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
1392 instead of GP_REG_LAST-1.
1393 (riscv_adjust_libcall_cfi_prologue): Likewise.
1394 (riscv_adjust_libcall_cri_epilogue): Likewise.
1395 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
1398 2017-12-04 Luis Machado <luis.machado@linaro.org>
1400 * ipa-pure-const.c (check_decl): Add missing newline.
1401 (state_from_flags): Likewise.
1403 2017-12-04 Jeff Law <law@redhat.com>
1405 PR tree-optimizatin/78496
1406 * gimple-ssa-evrp-analyze.h
1407 (evrp_range_analyzer::get_vr_values): Simplify.
1408 * gimple-ssa-evrp-analyze.c: Corresponding changes.
1409 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
1410 and gimple-ssa-evrp-analyze.h.
1411 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
1412 (simplify_stmt_for_jump_threading): Copy a blob of code from
1413 tree-vrp.c to use ranges to simplify statements.
1414 (dom_opt_dom_walker::before_dom_children): Call
1415 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
1416 (dom_opt_dom_walker::after_dom_children): Similarly for
1417 evrp_range_analyzer::leave.
1418 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
1421 * gimple-ssa-evrp-analyze.c
1422 (evrp_range_analyzer::extract_range_from_stmt): Always use
1423 vr_values::update_value_range so preexisting range info is
1424 medged with new range info, even if the new range is VR_VARYING.
1426 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1428 * combine.c: Adjust comment.
1429 (use_crosses_set_p): Delete.
1430 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
1431 (try_combine): Ditto.
1433 2017-12-04 Richard Biener <rguenther@suse.de>
1435 PR tree-optimization/83255
1436 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
1437 Re-add zero-iteration check.
1439 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1441 PR rtl-optimization/83245
1442 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
1443 hard registers as earlyclobber, also if not in an asm.
1445 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1449 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1452 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1453 rs6000_cpu based on cpu_index, not tune_index.
1455 2017-12-04 Richard Biener <rguenther@suse.de>
1457 PR tree-optimization/83238
1458 * graphite-scop-detection.c (scop_detection::merge_sese): Make
1459 code match comment, rejecting invalid SESE regions.
1461 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
1463 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
1464 require base operand is a REG_POINTER prior to reload on targets
1465 with non-equivalent space registers.
1467 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1469 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
1470 (update_specialized_profile): Fix updating of counts.
1471 (perhaps_add_new_callers): Likewise.
1473 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1476 * x86-tune.def: Remove obsolette FIXMEs.
1477 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
1478 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
1479 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
1481 (X86_TUNE_PAD_RETURNS): Disable for generic.
1483 2017-12-02 Jakub Jelinek <jakub@redhat.com>
1485 PR tree-optimization/83170
1486 PR tree-optimization/83241
1487 * gimple-ssa-store-merging.c
1488 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
1489 gimple_vuse (ins_stmt) in case it has changed.
1490 (imm_store_chain_info::output_merged_store): Likewise.
1492 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
1496 * tree-cfg.c (pass_warn_function_return::execute): Handle
1497 __builtin_ubsan_handle_missing_return like __builtin_unreachable
1498 with BUILTINS_LOCATION.
1502 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
1503 is BLKmode for vector field with vector raw mode, use TYPE_MODE
1504 instead of DECL_MODE.
1506 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
1507 last argument to rtx pointer.
1508 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
1509 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
1510 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
1511 depending on the chosen ISAs.
1512 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
1513 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
1514 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
1516 * config/i386/sse.md (mov<mode>_internal): Likewise.
1517 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
1519 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1521 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
1523 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1525 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
1526 parameters from prototype.
1527 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
1528 parameters. Don't print a comment.
1529 (emit_fusion_gpr_load): Adjust.
1530 (emit_fusion_load_store): Adjust.
1531 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
1532 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
1533 comment on the second line.
1535 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1538 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1539 rs6000_cpu based on cpu_index, not tune_index.
1541 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1543 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
1544 which_alternative instead of which_alternative + 1.
1545 (output_asm_insn): Print an extra tab if the template is short.
1547 2017-12-01 Jim Wilson <jimw@sifive.com>
1549 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
1551 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
1552 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
1554 * doc/tm.texi: Regenerate.
1556 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
1559 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
1560 whether we can allocate pseudos before trying to fix an address.
1561 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
1562 memory address is indexed or indirect.
1563 (floatuns_<mode>si2_hw2): Likewise.
1565 2017-12-01 Jason Merrill <jason@redhat.com>
1567 * Makefile.in (TAGS): Add c-family/*.cc.
1569 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
1571 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
1572 (call_value_insn): Likewise.
1573 (sibcall_insn): Likewise.
1574 (sibcall_value_insn): Likewise.
1575 (movsi_aarch64): Likewise.
1576 (movdi_aarch64): Likewise.
1577 (add_losym_): Likewise.
1578 (ldr_got_small_): Likewise.
1579 (ldr_got_small_sidi): Likewise.
1580 (ldr_got_small_28k_): Likewise.
1581 (ldr_got_small_28k_sidi): Likewise.
1582 * config/aarch64/aarch64.c (aarch64_print_address_internal):
1583 Move output_addr_const to symbolic case. Add error check.
1585 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1587 * config/s390/predicates.md (plus16_Q_operand): New predicate.
1588 * config/s390/s390.md: Disable MVC merging peephole if it would
1589 disable operand forwarding.
1590 (new peephole2): Split MVCs if it would turn them into up to 2
1593 2017-12-01 Richard Biener <rguenther@suse.de>
1595 PR tree-optimization/83232
1596 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
1597 detection of same access. Instead of breaking the group here
1598 do not consider the duplicate. Add comment explaining real fix.
1600 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
1602 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
1604 2017-12-01 Sudakshina Das <sudi.das@arm.com>
1606 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
1608 2017-12-01 Jakub Jelinek <jakub@redhat.com>
1610 * function.h (struct function): Remove cilk_frame_decl,
1611 is_cilk_function and calls_cilk_spawn fields.
1612 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
1614 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
1615 cilk_elemental field.
1616 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
1617 * target.def: Adjust comment.
1618 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
1619 Don't test cilk_elemental.
1621 PR tree-optimization/83233
1622 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
1623 bswap_stat name for the struct.
1626 * tree.h (SWITCH_BREAK_LABEL_P): Define.
1627 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
1628 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
1629 SWITCH_BREAK_LABEL_P set on the label.
1630 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
1631 added for default case if it was missing and not all cases covered.
1632 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
1633 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
1635 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
1636 to avoid -Wimplicit-fallthrough warning.
1637 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
1638 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
1640 PR tree-optimization/83221
1641 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
1643 (init_reassoc): Formatting fix.
1646 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
1647 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
1651 * tree-cfg.c: Include asan.h.
1652 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
1654 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1656 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
1658 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1660 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
1661 vpcmpeqd instruction.
1663 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1665 * config/i386/i386.c (standard_sse_constant_opcode): Fix
1666 registers type for 128bit mode.
1668 2017-11-30 Jakub Jelinek <jakub@redhat.com>
1670 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
1672 * gengtype-state.c (read_state_pair): Likewise.
1673 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
1674 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
1675 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
1676 * ifcvt.c (noce_try_store_flag_constants): Likewise.
1677 * tree-ssa-ccp.c (ccp_finalize): Likewise.
1678 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
1679 * builtins.c (fold_builtin_3): Likewise.
1680 * graphite-scop-detection.c
1681 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
1682 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
1685 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
1688 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
1689 parameter. Return true if ignore_decl_rtl_set_p is true and other
1690 conditions are satisfied.
1691 * asan.h (asan_protect_global): Add new parameter.
1692 * varasm.c (categorize_decl_for_section): Pass true as second parameter
1693 to asan_protect_global calls.
1695 2017-11-30 Jim Wilson <jimw@sifive.com>
1697 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
1698 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
1699 -mstrict-align, add info on default value. Delete redundant lines for
1700 -mabi. Add missing -mexplicit-relocs docs.
1702 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1704 * config/arc/arc.md (trap): New pattern.
1706 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1708 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
1709 instruction to end into a delay slot.
1710 * config/arc/arc.md (cond_delay_insn): Check if the instruction
1711 can be placed into a delay slot against reg_note.
1713 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1715 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
1716 labels number of usages.
1718 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1720 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
1722 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
1724 2017-11-30 Jakub Jelinek <jakub@redhat.com>
1727 * internal-fn.c (expand_mul_overflow): Optimize unsigned
1728 multiplication by power of 2 constant into two shifts + comparison.
1730 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
1733 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
1735 2017-11-30 Richard Biener <rguenther@suse.de>
1737 PR tree-optimization/83202
1738 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
1739 allow_peel argument and guard peeling.
1740 (canonicalize_loop_induction_variables): Likewise.
1741 (canonicalize_induction_variables): Pass false.
1742 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
1743 peeling from cunrolli.
1745 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1747 * combine.c (try_combine): Print a message to dump file whenever
1748 I0, I1, or I2 cannot be combined into I3.
1750 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1752 PR rtl-optimization/83156
1753 PR rtl-optimization/82621
1754 * combine.c (try_combine): Don't split an I2 if one of the dests is
1755 set again before I3. Allow unused dests.
1757 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1759 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
1761 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
1763 PR rtl-optimization/80818
1764 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
1765 recursively. Use insn code for clobber.
1766 (lra_set_insn_recog_data): Pass the new arg to
1767 collect_non_operand_hard_regs.
1768 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
1770 (lra_update_insn_regno_info): Pass insn to
1771 add_regs_to_insn_regno_info.
1773 2017-11-29 Jim Wilson <jimw@sifive.com>
1774 Andrew Waterman <andrew@sifive.com>
1776 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
1777 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
1778 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
1779 use SINGLE_SHIFT_COST.
1780 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
1781 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
1783 2017-11-29 Julia Koval <julia.koval@intel.com>
1785 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
1786 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
1787 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
1788 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
1789 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
1790 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
1791 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
1792 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
1793 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
1794 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
1795 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
1796 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
1797 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
1798 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
1799 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
1800 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
1801 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
1802 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
1803 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
1804 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
1805 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
1806 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
1807 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
1808 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
1809 __builtin_ia32_vpshldv_v2di_mask,
1810 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
1811 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
1812 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
1814 2017-11-29 Julia Koval <julia.koval@intel.com>
1816 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
1817 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
1818 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
1819 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
1820 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
1821 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
1822 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
1823 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
1824 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
1825 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
1826 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
1827 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
1828 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
1829 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
1830 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
1831 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
1832 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
1833 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
1834 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1835 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
1836 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
1838 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1840 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
1841 movsi_pic_gotdata_op instruction as a load for the UT699 errata
1844 2017-11-29 Martin Aberg <maberg@gaisler.com>
1846 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1847 to prevent b2bst errata sequence.
1848 (sqrtdf2_fix): Likewise.
1850 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1852 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
1853 (div_sqrt_insn_p): New function.
1854 (sparc_do_work_around_errata): Insert NOP instructions to
1855 prevent sequences that could trigger the TN-0013 errata for
1856 certain LEON3 processors.
1857 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
1858 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
1859 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
1860 (in_branch_delay): Prevent div and sqrt in delay slot if
1862 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
1864 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1866 * config/sparc/sparc.c (atomic_insn_p): New function.
1867 (sparc_do_work_around_errata): Insert NOP instructions to
1868 prevent sequences that could trigger the TN-0010 errata for
1870 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
1871 instruction referable in atomic_insns_p.
1873 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1875 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
1876 (atomic_compare_and_swap_leon3_1): Likewise.
1879 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1881 * config/sparc/sparc.c (fpop_insn_p): New function.
1882 (sparc_do_work_around_errata): Insert NOP instructions to
1883 prevent sequences that could trigger the TN-0012 errata for
1885 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
1886 * config/sparc/sparc.md (fix_gr712rc): New attribute.
1887 (in_branch_annul_delay): Prevent floating-point instructions
1888 in delay slot of annulled integer branch.
1890 2017-11-29 Richard Biener <rguenther@suse.de>
1892 PR tree-optimization/83202
1893 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
1895 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
1896 nodes vectorized to the same stmts multiple times.
1897 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
1898 (vect_analyze_slp_instance): Adjust.
1899 (scalar_stmts_to_slp_tree_map_t): New typedef.
1900 (vect_schedule_slp_instance): Add a map recording the SLP node
1901 representing the vectorized stmts for a set of scalar stmts.
1902 Avoid code-generating redundancies.
1903 (vect_schedule_slp): Allocate map and pass it down.
1905 2017-11-29 Nathan Sidwell <nathan@acm.org>
1908 * tree.c (build_complex_type): Fix canonicalization. Only fill in
1911 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
1913 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
1914 cases for printing LDP/STP memory addresses.
1915 (aarch64_print_address_internal): Renamed from
1916 aarch64_print_operand_address, added parameter, add Pmode check.
1917 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
1918 (aarch64_print_operand_address): Indirect to
1919 aarch64_print_address_internal.
1920 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
1921 'y' operand output specifier.
1923 2017-11-29 Jakub Jelinek <jakub@redhat.com>
1926 * tree.c (build_simple_mem_ref_loc): Handle
1927 get_addr_base_and_unit_offset returning a MEM_REF.
1930 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
1933 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
1936 2017-11-28 David Malcolm <dmalcolm@redhat.com>
1938 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
1940 (layout::move_to_column): Likewise.
1942 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1945 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
1946 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
1948 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1949 Martin Jambor <mjambor@suse.cz>
1952 * tree.h (expr_type_first_operand_type_p): Declare
1953 * tree.c (expr_type_first_operand_type_p): New function.
1954 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
1955 (ipa_value_from_jfunc): Adjust declaration.
1956 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
1957 Use it as result type for arithmetics, unless it is NULL in which case
1958 be more conservative.
1959 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
1960 ipa_get_jf_pass_through_result.
1961 (propagate_vals_across_pass_through): Likewise.
1962 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
1963 is to propagate_vals_across_pass_through.
1964 (propagate_constants_across_call): Pass PARM_TYPE to
1965 propagate_scalar_across_jump_function.
1966 (find_more_scalar_values_for_callers_subset): Pass parameter type to
1967 ipa_value_from_jfunc.
1968 (cgraph_edge_brings_all_scalars_for_node): Likewise.
1969 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
1970 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
1971 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
1972 target_type, pass it to ipa_value_from_jfunc.
1973 (update_indirect_edges_after_inlining): Pass parameter type to
1974 try_make_edge_direct_simple_call.
1976 2017-11-28 Jeff Law <law@redhat.com>
1978 * gimple-ssa-evrp-analyze.c
1979 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
1980 refine ranges if scev_initialized_p returns true.
1981 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
1983 2017-11-28 Julia Koval <julia.koval@intel.com>
1985 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
1986 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
1987 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
1988 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
1989 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
1990 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
1991 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
1992 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
1993 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
1994 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
1995 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
1996 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
1997 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
1998 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
1999 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
2000 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
2001 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
2002 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
2003 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
2004 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
2005 __builtin_ia32_vpshrd_v2di_mask): New builtins.
2006 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
2008 2017-11-28 Julia Koval <julia.koval@intel.com>
2010 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
2011 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
2012 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
2013 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
2014 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
2015 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
2016 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
2017 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
2018 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
2019 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
2020 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
2021 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
2022 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
2023 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
2024 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
2025 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
2026 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
2027 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
2028 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
2029 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
2030 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
2031 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
2032 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
2033 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
2034 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
2035 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
2036 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
2037 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
2038 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
2039 __builtin_ia32_vpshld_v2di_mask): New builtins.
2040 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2041 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
2043 2017-11-28 Richard Biener <rguenther@suse.de>
2045 PR tree-optimization/80776
2046 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
2048 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
2050 (evrp_range_analyzer::record_ranges_from_incoming_edges):
2051 If the incoming edge is an effective fallthru because the other
2052 edge only reaches a __builtin_unreachable () then record ranges
2053 derived from the controlling condition in SSA info.
2054 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
2055 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
2057 2017-11-28 Olivier Hainque <hainque@adacore.com>
2059 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
2060 on mingw build hosts.
2062 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2065 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
2067 2017-11-28 Richard Biener <rguenther@suse.de>
2070 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
2071 copies generated from memcpy use a character array as reference
2074 2017-11-28 Julia Koval <julia.koval@intel.com>
2075 Sebastian Peryt <sebastian.peryt@intel.com>
2077 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
2078 c-family/c-cilkplus.o, c-family/array-notation-common.o,
2079 cilk-common.o, cilk.h, cilk-common.c): Remove.
2081 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
2082 * builtins.c (is_builtin_name): Remove cilkplus condition.
2083 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
2084 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
2085 cilk-builtins.def, cilkplus.def): Remove.
2086 * cif-code.def (CILK_SPAWN): Remove.
2087 * cilk-builtins.def: Delete.
2088 * cilk-common.c: Ditto.
2090 * cilkplus.def: Ditto.
2091 * config/darwin.h (fcilkplus): Delete.
2092 * cppbuiltin.c: Ditto.
2093 * doc/extend.texi: Remove cilkplus doc.
2094 * doc/generic.texi: Ditto.
2095 * doc/invoke.texi: Ditto.
2096 * doc/passes.texi: Ditto.
2097 * gcc.c (fcilkplus): Remove.
2098 * gengtype.c (cilk.h): Remove.
2099 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
2101 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
2103 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
2105 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
2106 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
2107 cilkplus conditions.
2108 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
2109 inline_read_section): Ditto.
2110 * ipa-inline-analysis.c (cilk.h): Remove.
2111 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
2112 * lto-wrapper.c (merge_and_complain, append_compiler_options,
2113 append_linker_options): Remove condition for fcilkplus.
2114 * lto/lto-lang.c (cilk.h): Remove.
2115 (lto_init): Remove condition for fcilkplus.
2116 * omp-expand.c (expand_cilk_for_call): Delete.
2117 (expand_omp_taskreg, expand_omp_for_static_chunk,
2118 expand_omp_for): Remove cilkplus
2120 (expand_cilk_for): Delete.
2121 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
2122 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
2123 execute_lower_omp, diagnose_sb_0): Ditto.
2124 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
2125 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2126 * tree-nested.c: Ditto.
2127 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
2128 (dump_generic_node): Ditto.
2129 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2130 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
2132 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
2134 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2136 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
2137 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
2138 for vector mode and !TARGET_SIMD.
2140 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2142 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
2144 * tree.h (SWITCH_LABELS): Remove.
2145 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
2146 assert SWITCH_BODY is non-NULL.
2147 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
2149 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
2151 PR tree-optimization/80788
2152 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
2153 has TREE_OVERFLOW set, call drop_tree_overflow.
2155 2017-11-28 Richard Biener <rguenther@suse.de>
2157 PR tree-optimization/83158
2158 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
2160 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
2163 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
2164 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
2166 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
2168 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
2169 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
2171 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2174 * varasm.c (compare_constant): Take the mode of the constants into
2175 account when comparing floating point constants.
2177 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
2179 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
2180 from explicit instantiation of debug_helper.
2181 * vec.h (DEFINE_DEBUG_VEC): Ditto.
2183 2017-11-27 Richard Biener <rguenther@suse.de>
2185 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
2188 2017-11-27 Richard Biener <rguenther@suse.de>
2190 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
2191 (type_hash_canon): Also clear int_cst_hash_table entry for
2193 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
2195 2017-11-27 Tamar Christina <tamar.christina@arm.com>
2197 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2198 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2199 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2201 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2203 * hash-map.h (gt_cleare_cache): Avoid UB.
2205 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
2207 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
2208 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
2209 (decide_unroll_runtime_iterations): Likewise.
2210 (decide_unroll_stupid): Likewise.
2212 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2215 * config/i386/i386.md: Add a loop with incssp.
2217 2017-11-27 Martin Jambor <mjambor@suse.cz>
2219 PR tree-optimization/81248
2220 * tree-sra.c (splice_param_accesses): Remove size check.
2221 (decide_one_param_reduction): Fix size check.
2222 * gimple-pretty-print.c (dump_profile): Silence warning.
2223 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
2225 2017-11-27 Jakub Jelinek <jakub@redhat.com>
2228 * dbxout.c (lastlineno): New variable.
2229 (dbx_debug_hooks): Use dbxout_switch_text_section as
2230 switch_text_section debug hook.
2231 (dbxout_function_end): Switch to current_function_section
2232 rather than function_section. If crtl->has_bb_partition,
2233 output just one N_FUN, depending on in_cold_section_p.
2234 (dbxout_source_line): Remember last lineno in lastlineno.
2235 (dbxout_switch_text_section): New function.
2236 (dbxout_function_decl): Adjust dbxout_block caller.
2237 (dbx_block_with_cold_children): New function.
2238 (dbxout_block): Return true if any LBRAC/RBRAC have been
2239 emitted. Use dbx_block_with_cold_children at depth == 0
2240 in second partition. Add PARENT_BLOCKNUM argument, pass
2241 it optionally adjusted to children. Output LBRAC/RBRAC
2242 around recursive call only if the block is in the current
2243 partition, if not and anything was output, emit empty
2245 * final.c (final_scan_insn): Compute cold_function_name
2246 before calling switch_text_section debug hook. Call
2247 that hook even if dwarf2out_do_frame if not emitting
2251 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
2252 TREE_READONLY decls.
2254 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2256 PR rtl-optimization/82488
2257 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
2259 2017-11-26 Julia Koval <julia.koval@intel.com>
2261 * config/i386/i386.c (processor_target_table): Add skylake_cost for
2263 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
2266 2017-11-26 Julia Koval <julia.koval@intel.com>
2268 * config/i386/driver-i386.c (host_detect_local_cpu):
2269 Detect skylake-avx512.
2271 2017-11-26 Julia Koval <julia.koval@intel.com>
2273 * config.gcc: Add -march=cannonlake.
2274 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
2275 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
2276 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
2277 (PTA_CANNONLAKE): New.
2278 (processor_target_table): Add cannonlake.
2279 (ix86_option_override_internal): Ditto.
2280 (fold_builtin_cpu): Ditto.
2281 (get_builtin_code_for_version): Handle cannonlake.
2282 (M_INTEL_COREI7_CANNONLAKE): New.
2283 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
2284 * doc/invoke.texi: Add -march=cannonlake.
2286 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
2288 * plugin.c (add_new_plugin): Use platform-specific library extensions.
2289 (try_init_one_plugin): Alternative implementation for MinGW.
2290 * Makefile.in (plugin_implib): New.
2291 (gengtype-lex.c): Fix broken AIX workaround.
2292 * configure: Regenerate.
2293 * doc/plugins.texi: Document support for MinGW.
2295 2017-11-25 Jakub Jelinek <jakub@redhat.com>
2297 PR rtl-optimization/81553
2298 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
2299 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
2300 is a shift where C1 has different mode than the whole shift, use C1's
2301 mode for MULT rather than the shift's mode.
2304 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
2305 builtins not enabled in the currently selected ISA.
2307 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
2309 PR tree-optimization/71026
2310 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
2311 (insert_reciprocals): Change to insert reciprocals before a division
2312 by a square and to insert the square of a reciprocal.
2313 (execute_cse_reciprocals_1): Change to consider division by a square.
2314 (register_division_in): Add importance parameter.
2316 2017-11-24 Richard Biener <rguenther@suse.de>
2318 PR tree-optimization/82402
2319 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
2320 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
2322 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
2324 * match.pd (0-ptr): New transformation.
2326 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
2329 * ipa-inline.c (inline_small_functions): Set current badnes correctly
2330 when skipping checking.
2332 2017-11-24 Richard Biener <rguenther@suse.de>
2334 PR tree-optimization/83128
2335 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
2336 (vn_reference_lookup_3): Likewise.
2338 2017-11-24 Jakub Jelinek <jakub@redhat.com>
2341 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
2342 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
2345 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
2346 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
2347 check for the latter with gimple_call_builtin_p. Do not handle
2348 BUILT_IN_STPNCPY_CHK which is not a pass through call.
2350 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
2352 * config/arm/arm_neon.h: Fix pragma GCC push_options before
2355 2017-11-23 Julia Koval <julia.koval@intel.com>
2357 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
2358 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
2359 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
2360 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
2361 _mm512_maskz_expandloadu_epi16): New intrinsics.
2362 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
2363 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
2364 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
2365 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
2366 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
2367 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
2368 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
2369 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
2370 _mm256_maskz_expandloadu_epi8): New intrinsics.
2371 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
2372 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
2373 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
2374 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
2375 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
2376 * config/i386/sse.md (VI248_VLBW): New iterator.
2377 (expand<mode>_mask, expand<mode>_maskz): New patterns.
2379 2017-11-23 Julia Koval <julia.koval@intel.com>
2381 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
2382 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
2383 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
2384 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
2385 _mm512_mask_compressstoreu_epi16): New.
2386 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
2387 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
2388 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
2389 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
2390 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
2391 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
2392 _mm256_mask_compressstoreu_epi8): New.
2393 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
2394 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
2395 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
2396 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
2397 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
2398 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
2399 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
2400 __builtin_ia32_compresshi128_mask,
2401 __builtin_ia32_compressstoreuqi512_mask,
2402 __builtin_ia32_compressstoreuhi512_mask,
2403 __builtin_ia32_compressstoreuqi256_mask,
2404 __builtin_ia32_compressstoreuqi128_mask,
2405 __builtin_ia32_compressstoreuhi256_mask,
2406 __builtin_ia32_compressstoreuhi128_mask): New builtins.
2407 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
2409 (ix86_expand_special_args_builtin): Handle new types.
2410 (s4fma_expand): Handle new builtin array.
2411 * config/i386/immintrin.h: Include new headers.
2412 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
2413 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
2415 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2417 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
2418 fixes. Declare temp and g variables at the top in order to avoid
2419 {} in most of the cases.
2421 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
2423 * match.pd (ptr-0): New transformation.
2425 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2427 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
2428 (struct addr_mode_cost_table): New.
2429 (struct tune_params): Add field addr_mode_costs.
2430 * config/arm/arm.c (generic_addr_mode_costs): New.
2431 (arm_slowmul_tune): Initialise addr_mode_costs field.
2432 (arm_fastmul_tune): Likewise.
2433 (arm_strongarm_tune): Likewise.
2434 (arm_xscale_tune): Likewise.
2435 (arm_9e_tune): Likewise.
2436 (arm_marvell_pj4_tune): Likewise.
2437 (arm_v6t2_tune): Likewise.
2438 (arm_cortex_tune): Likewise.
2439 (arm_cortex_a8_tune): Likewise.
2440 (arm_cortex_a7_tune): Likewise.
2441 (arm_cortex_a15_tune): Likewise.
2442 (arm_cortex_a35_tune): Likewise.
2443 (arm_cortex_a53_tune): Likewise.
2444 (arm_cortex_a57_tune): Likewise.
2445 (arm_exynosm1_tune): Likewise.
2446 (arm_xgene1_tune): Likewise.
2447 (arm_cortex_a5_tune): Likewise.
2448 (arm_cortex_a9_tune): Likewise.
2449 (arm_cortex_a12_tune): Likewise.
2450 (arm_cortex_a73_tune): Likewise.
2451 (arm_v7m_tune): Likewise.
2452 (arm_cortex_m7_tune): Likewise.
2453 (arm_v6m_tune): Likewise.
2454 (arm_fa726te_tune): Likewise.
2455 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
2457 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2459 * config/arm/arm.c (arm_mem_costs): New function.
2460 (arm_rtx_costs_internal): Use arm_mem_costs.
2462 2017-11-23 Mark Wielaard <mark@klomp.org>
2464 * dwarf2out.c (init_sections_and_labels): Use generation to create
2465 unique ranges_section_label and ranges_base_label. Return generation.
2466 (output_rnglists): Add generation argument. Use generation to create
2467 unique ranges labels.
2468 (dwarf2out_finish): Get generation from init_sections_and_labels
2469 and pass generation to output_rnglists.
2471 2017-11-23 Mike Stump <mikestump@comcast.net>
2472 Eric Botcazou <ebotcazou@adacore.com>
2474 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
2475 * cfgloop.h (struct loop): Add unroll field.
2476 * function.h (struct function): Add has_unroll bitfield.
2477 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
2478 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
2479 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
2480 (pass_rtl_unroll_loops::gate): Likewise.
2481 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
2482 for which loop->unroll==1.
2483 (decide_unroll_constant_iterations): Use note for consistency and
2484 take loop->unroll into account. Return early if loop->unroll is set.
2485 Fix thinko in existing test.
2486 (decide_unroll_runtime_iterations): Use note for consistency and
2487 take loop->unroll into account.
2488 (decide_unroll_stupid): Likewise.
2489 * lto-streamer-in.c (input_cfg): Read loop->unroll.
2490 * lto-streamer-out.c (output_cfg): Write loop->unroll.
2491 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
2493 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
2494 (print_loop): Print loop->unroll if set.
2495 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
2496 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
2497 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
2499 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
2500 loop->unroll is set and smaller than the trip count. Otherwise bypass
2501 entirely the heuristics if loop->unroll is set. Remove dead note.
2502 Fix off-by-one bug in other note.
2503 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
2504 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
2506 (tree_unroll_loops_completely): Make static.
2507 (pass_complete_unroll::execute): Use correct type for variable.
2508 (pass_complete_unrolli::execute): Fix formatting.
2509 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
2511 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
2513 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
2514 enable when TARGET_PREFER_AVX128 is set.
2516 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2518 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
2520 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
2521 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
2523 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2525 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
2526 frequencies. Fix estimation of aggregate parameters.
2528 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2530 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
2533 2017-11-23 Tom de Vries <tom@codesourcery.com>
2535 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
2537 2017-11-23 Tom de Vries <tom@codesourcery.com>
2539 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
2542 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
2545 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
2546 sibcall_value_pcrel_fdpic): Use local variable instead of
2548 (calli_tbr_rel): Add missing operand 2.
2549 (call_valuei_tbr_rel): Add missing operand 3.
2551 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2554 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
2555 bitpos/bitsize covering the whole destination, use store_expr only if
2556 the complex mode is the same. Otherwise, use expand_normal and if
2557 it returns CONCAT, subreg each part separately instead of trying to
2558 subreg the whole result.
2560 2017-11-23 Richard Biener <rguenther@suse.de>
2562 PR tree-optimization/23094
2563 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
2564 come here from walking over backedges in the first iteration.
2565 (vn_reference_lookup_3): Skip clobbers that store the same value.
2567 2017-11-23 Richard Biener <rguenther@suse.de>
2569 PR tree-optimization/81403
2570 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
2571 a block we need a leader relative to.
2572 (phi_translate_1): For nary processing require a leader from
2573 get_representative_for given we run expression simplification
2574 using match-and-simplify. Remove previous fix.
2576 2017-11-22 Jeff Law <law@redhat.com>
2578 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
2579 Use new method allocate_value_range rather than accessing the
2580 vrp_value_range_pool data member directly.
2581 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
2582 to use extract_range_from_stmt method to avoid need for
2583 extract_range_from_assignment method.
2584 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
2585 method rather than setting values_propgated data member directly.
2586 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
2587 and values propagated data members and extract_range_from_assignment
2588 method. Reorder private data members to conform to standards.
2589 Add new methods set_lattice_propagation_complete and
2590 allocate_value_range.
2592 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
2594 PR rtl-optimization/83030
2595 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
2596 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
2597 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
2598 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
2599 CROSSING_JUMP_P flag is already set before setting it.
2600 * cfgrtl.c (fixup_partition_crossing): Likewise.
2601 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
2604 2017-11-22 Jakub Jelinek <jakub@redhat.com>
2606 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
2607 Handle the case where both arguments are using gen_const_vec_series.
2609 2017-11-22 David Malcolm <dmalcolm@redhat.com>
2612 * pretty-print.c (pp_format): Move quoting implementation to
2613 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
2614 to pass address of "quote" local.
2615 (pp_begin_quote): New function.
2616 (pp_end_quote): New function.
2617 * pretty-print.h (printer_fn): Convert penultimate param from bool
2619 (pp_begin_quote): New decl.
2620 (pp_end_quote): New decl.
2621 * tree-diagnostic.c (default_tree_printer): Convert penultimate
2622 param from bool to bool *.
2623 * tree-diagnostic.h (default_tree_printer): Likewise.
2625 2017-11-22 Jeff Law <law@redhat.com>
2627 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
2628 Initialize vr_values.
2629 (evrp_range_analyzer::try_find_new_range): Call methods attached to
2630 vr_values via vr_values class instance rather than delegators.
2631 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
2632 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
2633 (evrp_range_analyzer::push_value_range): Likewise.
2634 (evrp_range_analyzer::pop_value_range): Likewise.
2635 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
2636 most delegators. Those remaining are exposed as public interfaces.
2637 Make vr_values a pointer and private.
2638 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
2640 (evrp_range_analyzer::get_vr_value): New method.
2641 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
2642 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
2643 (class evrp_dom_walker): Attach evrp_folder class, initialize
2644 it in the ctor. Remove temporary delegators.
2645 (evrp_dom_walker::before_dom_children): Call methods in attached
2646 evrp_range_analyzer class via class instance pointer. Use
2647 free value_range_constant_singleton to remove need for
2648 op_with_constant_singleton_value delegator method. Do not
2649 create a vrp_prop class instance for every call! Narrow
2650 scope of a couple variables.
2651 (evrp_dom_walker::cleanup): Call methods in attached
2652 evrp_range_analyzer class via class instance pointer.
2653 * vr-values.h (class vr_values): Privatize many methods and
2656 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2658 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
2659 * cfgexpand.c (expand_debug_expr): Remove handling for them.
2660 * expr.c (expand_expr_real_2): Likewise.
2661 * fold-const.c (const_unop): Likewise.
2662 * optabs-tree.c (optab_for_tree_code): Likewise.
2663 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2664 * tree-inline.c (estimate_operator_cost): Likewise.
2665 * tree-pretty-print.c (dump_generic_node): Likewise.
2666 (op_code_prio): Likewise.
2667 (op_symbol_code): Likewise.
2668 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
2669 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
2670 * internal-fn.c (direct_internal_fn_optab): New function.
2671 (direct_internal_fn_array, direct_internal_fn_supported_p
2672 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
2673 * fold-const-call.c (fold_const_reduction): New function.
2674 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
2676 * tree-vect-loop.c: Include internal-fn.h.
2677 (reduction_code_for_scalar_code): Rename to...
2678 (reduction_fn_for_scalar_code): ...this and return an internal
2680 (vect_model_reduction_cost): Take an internal_fn rather than
2682 (vect_create_epilog_for_reduction): Likewise. Build calls rather
2684 (vectorizable_reduction): Use internal functions rather than tree
2685 codes for the reduction operation. Update calls to the functions
2687 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
2688 Use calls to internal functions rather than REDUC tree codes.
2689 * config/aarch64/aarch64-simd.md: Update comment accordingly.
2691 2017-11-22 Olivier Hainque <hainque@adacore.com>
2693 * config/vxworks.c (vxworks_override_options): Pick default
2694 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
2695 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
2696 DWARF_GNAT_ENCODINGS_DEFAULT.
2697 * config/vxworksae.h: Likewise.
2699 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
2701 PR tree-optimization/83104
2702 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
2705 2017-11-22 Marek Polacek <polacek@redhat.com>
2706 H.J. Lu <hongjiu.lu@intel.com>
2707 Jason Merrill <jason@redhat.com>
2712 * calls.c (initialize_argument_information): Call
2713 warn_parameter_passing_abi target hook.
2714 (store_one_arg): Use 0 for empty record size. Don't push 0 size
2715 argument onto stack.
2716 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
2717 * common.opt: Update -fabi-version description.
2718 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
2719 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
2721 (ix86_is_empty_record): New function.
2722 (ix86_warn_parameter_passing_abi): New function.
2723 (TARGET_EMPTY_RECORD_P): Redefine.
2724 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
2725 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
2726 * doc/tm.texi: Regenerated.
2727 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
2728 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
2729 * dwarf2out.c (get_ultimate_context): Move to tree.c.
2730 * explow.c (hard_function_value): Call arg_int_size_in_bytes
2731 instead of int_size_in_bytes.
2732 * expr.c (copy_blkmode_to_reg): Likewise.
2733 * function.c (aggregate_value_p): Return 0 for empty types.
2734 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
2736 (locate_and_pad_parm): Call arg size_in_bytes instead
2738 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
2739 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
2740 * target.def (empty_record_p, warn_parameter_passing_abi): New target
2742 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
2743 (std_gimplify_va_arg_expr): Skip empty records. Call
2744 arg_size_in_bytes instead size_in_bytes.
2745 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
2746 * tree-core.h (tree_type_common): Add empty_flag.
2747 (tree_decl_common): Update comments.
2748 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
2750 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
2751 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
2753 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
2754 * tree.c (default_is_empty_type): New function.
2755 (default_is_empty_record): New function.
2756 (arg_int_size_in_bytes): New function.
2757 (arg_size_in_bytes): New function.
2758 (get_ultimate_context): New function.
2759 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
2760 TRANSLATION_UNIT_WARN_EMPTY_P.
2761 (default_is_empty_record, arg_int_size_in_bytes,
2762 arg_size_in_bytes, get_ultimate_context): Declare.
2764 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2766 * config/arm/arm.c (cmse_clear_registers): New function.
2767 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
2768 code by call to cmse_clear_registers.
2769 (cmse_nonsecure_entry_clear_before_return): Likewise.
2771 2017-11-22 Tamar Christina <tamar.christina@arm.com>
2773 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
2774 (vdot_s32, vdotq_s32): New.
2775 (vdot_lane_u32, vdotq_lane_u32): New.
2776 (vdot_lane_s32, vdotq_lane_s32): New.
2779 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2782 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
2783 for unsigned values with fewer HWIs than the precision.
2784 (test_overflow): New function.
2785 (wide_int_cc_tests): Call it.
2787 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2789 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
2790 has the same mode class as Pmode.
2792 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2794 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
2795 padding_bits_to_clear_ptr.
2796 (cmse_nonsecure_entry_clear_before_return): Likewise.
2798 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2800 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
2801 auto_sbitap instead of integer bitfield to control register needing
2804 2017-11-22 Jakub Jelinek <jakub@redhat.com>
2806 PR tree-optimization/83044
2807 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
2808 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
2809 that need the upper bound. Subtract offset from
2810 get_addr_base_and_unit_offset only if positive and subtract it
2811 before division by eltsize rather than after it.
2814 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
2815 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
2816 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
2820 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
2821 expand_binop with *mul_widen_optab, make sure at least one of the
2822 operands doesn't have VOIDmode.
2825 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
2827 PR rtl-optimization/82044
2828 PR tree-optimization/82042
2829 * dse.c (record_store): Check for overflow.
2830 (check_mem_read_rtx): Properly check for overflow if width == -1, call
2831 add_wild_read instead of clear_rhs_from_active_local_stores on
2832 overflow and log it into dump_file.
2834 2017-11-22 Richard Biener <rguenther@suse.de>
2836 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
2837 fake edges to exit when looking for a place to insert.
2838 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
2840 (insert_into_preds_of_block): Commit edge insertion immediately,
2841 assert that doesn't require new BBs.
2842 (fini_pre): Release expressions.
2843 (pass_pre::execute): Shuffle things around a bit, if the fn
2844 is too large do not compute AVAIL either as this is really the
2847 2017-11-22 Richard Biener <rguenther@suse.de>
2849 PR tree-optimization/83089
2850 * tree-if-conv.c (pass_if_conversion::execute): If anything
2851 changed reset SCEV and free the number of iteration estimates.
2853 2017-11-21 Martin Sebor <msebor@redhat.com>
2855 PR tree-optimization/82945
2856 * calls.h (warn_nonstring_bound): Remove unused function.
2858 2017-11-21 Martin Sebor <msebor@redhat.com>
2860 PR tree-optimization/82945
2861 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
2862 * calls.h (maybe_warn_nonstring_arg): Declare new function.
2863 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
2865 (initialize_argument_information): Call maybe_warn_nonstring_arg.
2866 * calls.h (get_attr_nonstring_decl): Declare new function.
2867 * doc/extend.texi (attribute nonstring): Update.
2868 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
2869 get_attr_nonstring_decl and handle it.
2870 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
2871 detection of nul-termination.
2872 (strlen_to_stridx): Change to a pointer.
2873 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
2874 (pass_strlen::execute): Same.
2876 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
2878 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
2879 for the new option -mprefer-vector-width=[none|128|256|512].
2880 * config/i386/i386.c (ix86_target_string): remove old style options
2881 -mprefer-avx256 and make -mprefer-avx128 as alias.
2882 (ix86_option_override_internal): Apply defaults for the
2883 -mprefer-vector-width=[128|256] option.
2884 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
2885 Implement macros to work with -mprefer-vector-width=.
2886 * config/i386/i386.opt: Implemented option
2887 -mprefer-vector-width=[none|128|256|512].
2888 * doc/invoke.texi: Documentation for
2889 -mprefer-vector-width=[none|128|256|512].
2891 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
2893 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
2895 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
2897 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
2899 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
2901 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
2903 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
2904 * expr.c (expand_expr_real_2): Likewise.
2905 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
2906 fold_binary_loc): Likewise.
2907 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
2908 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
2909 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
2910 MINUS_EXPR transformations.
2911 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
2912 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
2913 * tree-inline.c (estimate_operator_cost): Likewise.
2914 * tree-pretty-print.c (dump_generic_node, op_code_prio,
2915 op_symbol_code): Likewise.
2916 * tree-vect-stmts.c (vectorizable_operation): Likewise.
2917 * vr-values.c (extract_range_from_binary_expr): Likewise.
2918 * varasm.c (initializer_constant_valid_p_1): Likewise.
2919 * tree.def: New tree code POINTER_DIFF_EXPR.
2921 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
2923 * config/i386/i386.md (*bswap<mode>2_movbe): Add
2924 integer suffix to movbe mnemonic.
2925 (*bswaphi2_movbe): Ditto.
2926 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
2928 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2931 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
2932 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
2933 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
2935 2017-11-21 Martin Liska <mliska@suse.cz>
2937 * tree-inline.c (expand_call_inline): Remove not needed
2940 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
2941 Jakub Jelinek <jakub@redhat.com>
2944 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
2945 Remove static keyword from f variable.
2947 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2949 PR tree-optimization/83086
2950 * gimple-ssa-store-merging.c
2951 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
2952 rather than n.base_addr.
2954 2017-11-21 Martin Liska <mliska@suse.cz>
2956 PR rtl-optimization/82044
2957 PR tree-optimization/82042
2958 * dse.c (check_mem_read_rtx): Check for overflow.
2960 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2962 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
2965 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2967 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
2968 and its corresponding call arg location note.
2970 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2971 Andrew Burgess <andrew.burgess@embecosm.com>
2973 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
2975 (arc_return_slot_offset): Likewise.
2976 (arc_eh_return_address_location): New declaration.
2977 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
2978 (MUST_SAVE_REGISTER): Add exception handler case.
2979 (MUST_SAVE_RETURN_ADDR): Likewise.
2980 (arc_frame_pointer_required): Likewise.
2981 (arc_frame_pointer_needed): New function.
2982 (arc_compute_frame_size): Changed.
2983 (arc_expand_prologue): Likewise.
2984 (arc_expand_epilogue): Likewise.
2985 (arc_initial_elimination_offset): Likewise.
2986 (arc_return_slot_offset): Delete.
2987 (arc_eh_return_address_location): New function.
2988 (arc_builtin_setjmp_frame_value): Likewise.
2989 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
2990 (EH_RETURN_STACKADJ_RTX): Define.
2991 (EH_RETURN_HANDLER_RTX): Likewise.
2992 * config/arc/arc.md (eh_return): Delete.
2994 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2996 * print-tree.h (debug_vec_tree): Remove prototype.
2997 * gdbinit.in (pvt): Remove macro.
2999 2017-11-21 Jakub Jelinek <jakub@redhat.com>
3001 PR tree-optimization/83047
3002 * gimple-ssa-store-merging.c
3003 (imm_store_chain_info::output_merged_store): If the loads with the
3004 same vuse are in different basic blocks, for load_gsi pick a load
3005 location that is dominated by the other loads.
3008 * config/i386/i386.c (ix86_memmodel_check): Start
3009 -Winvalid-memory-model diagnostics with lowercase letter.
3012 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
3013 set in_cold_section_p to the partition containing loc_list->first.
3014 When seeing loc_list->last_before_switch node, update secname and
3015 perform range_across_switch second partition handling only after that.
3018 * run-rtl-passes.c: Include debug.h.
3019 (run_rtl_passes): Call debug_hooks->assembly_start.
3020 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
3024 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
3025 OPTAB_DIRECT in calls to expand_simple_binop.
3027 2017-11-20 David Malcolm <dmalcolm@redhat.com>
3030 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
3032 2017-11-20 David Malcolm <dmalcolm@redhat.com>
3035 * spellcheck.h (best_match::blithely_get_best_candidate): New
3038 2017-11-20 Jakub Jelinek <jakub@redhat.com>
3040 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
3041 with lower case letter.
3043 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
3045 * config/i386/i386.md (bswaphi2): New expander.
3046 (*bswaphi2_movbe): New insn pattern.
3047 (bswaphi -> rorhi pepehole2): New peephole pattern.
3049 2017-11-20 Jeff Law <law@redhat.com>
3051 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
3052 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
3053 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
3054 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
3055 gimple-ssa-evrp-analyze.h.
3057 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
3058 set BB_VISITED here.
3059 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
3061 2017-11-20 Bin Cheng <bin.cheng@arm.com>
3063 * tree-predcom.c: Add general comment on Store-Store chains.
3064 (split_data_refs_to_components): Postpone clearing eliminate_store_p
3066 (get_chain_last_ref_at): Rename into...
3067 (get_chain_last_write_at): ...this.
3068 (get_chain_last_write_before_load): New function.
3069 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
3070 CT_STORE_STORE when write reference is added.
3071 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
3072 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
3073 (initialize_root_vars_store_elim_1): Ditto.
3074 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
3075 definition is created.
3076 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
3077 chain by replacing it with dominant stored value.
3079 2017-11-20 Bin Cheng <bin.cheng@arm.com>
3081 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
3083 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
3085 * vr-values.c (extract_range_from_binary_expr): Use a full range
3088 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
3090 * config/arm/arm.md (R4_REGNUM): Define constant.
3091 (nonsecure_call_internal): Remove r4 clobber.
3092 (nonsecure_call_value_internal): Likewise.
3093 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
3094 clobber and resequence match_operands.
3095 (nonsecure_call_value_reg_thumb1_v5): Likewise.
3096 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
3097 (nonsecure_call_value_reg_thumb2): Likewise.
3099 2017-11-20 Jakub Jelinek <jakub@redhat.com>
3101 PR tree-optimization/78821
3102 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
3103 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
3104 to the address of the base rather than the base itself.
3105 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
3106 (find_bswap_or_nop_finalize): New function.
3107 (find_bswap_or_nop): Use it.
3108 (bswap_replace): Return a tree rather than bool, change first
3109 argument from gimple * to gimple_stmt_iterator, allow inserting
3110 into an empty sequence, allow ins_stmt to be NULL - then emit
3111 all stmts into gsi. Fix up MEM_REF address gimplification.
3112 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
3113 (struct store_immediate_info): Add N and INS_STMT non-static
3115 (store_immediate_info::store_immediate_info): Initialize them
3116 from newly added ctor args.
3117 (merged_store_group::apply_stores): Formatting fixes. Sort by
3119 (stmts_may_clobber_ref_p): For stores call also
3120 refs_anti_dependent_p.
3121 (gather_bswap_load_refs): New function.
3122 (imm_store_chain_info::try_coalesce_bswap): New method.
3123 (imm_store_chain_info::coalesce_immediate_stores): Use it.
3124 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
3125 (imm_store_chain_info::output_merged_store): Fail if number of
3126 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
3127 and NOP_EXPR rhs_code.
3128 (pass_store_merging::process_store): Compute n and ins_stmt, if
3129 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
3130 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
3132 (pass_store_merging::execute): Calculate dominators.
3134 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
3135 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
3136 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
3137 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3138 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
3139 class pass_optimize_bswap, bswap_replace,
3140 pass_optimize_bswap::execute): Moved to ...
3141 * gimple-ssa-store-merging.c: ... this file.
3142 Include optabs-tree.h.
3143 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
3144 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3145 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
3146 anonymous namespace, remove static keywords.
3147 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
3148 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
3150 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
3153 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
3155 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
3157 * vec.h (debug_helper): New function.
3158 (DEFINE_DEBUG_VEC): New macro.
3159 * hash-set.h (debug_helper): New function.
3160 (DEFINE_DEBUG_HASH_SET): New macro.
3161 * cfg.c (debug_slim (edge)): New function.
3162 Call DEFINE_DEBUG_VEC for edges.
3163 Call DEFINE_DEBUG_HASH_SET for edges.
3164 * cfghooks.c (debug_slim (basic_block)): New function.
3165 Call DEFINE_DEBUG_VEC for basic blocks.
3166 Call DEFINE_DEBUG_HASH_SET for basic blocks.
3167 * print-tree.c (debug_slim): New function to handle trees.
3168 Call DEFINE_DEBUG_VEC for trees.
3169 Call DEFINE_DEBUG_HASH_SET for trees.
3170 (debug (vec<tree, va_gc>) &): Remove.
3171 (debug (<vec<tree, va_gc>) *): Remove.
3172 * print-rtl.c (debug_slim): New function to handle const_rtx.
3173 Call DEFINE_DEBUG_VEC for rtx_def.
3174 Call DEFINE_DEBUG_VEC for rtx_insn.
3175 Call DEFINE_DEBUG_HASH_SET for rtx_def.
3176 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
3177 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
3178 (debug (vec<rtx_insn *> *ptr): Remove.
3179 (debug_insn_vector): Remove.
3180 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
3182 2017-11-20 Tom de Vries <tom@codesourcery.com>
3184 PR rtl-optimization/82020
3185 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
3186 IF_THEN_ELSE condition.
3188 2017-11-19 Jeff Law <law@redhat.com>
3190 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
3191 of degenerates resulting from ignoring an edge.
3193 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3196 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
3198 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3201 * profile-count.c (profile_count::to_sreal_scale): Fix return value
3202 for uninitialied counts.
3204 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3207 * tree-inline.c (estimate_num_insns): Set to 1 at least.
3209 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3212 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
3214 2017-11-19 Tom de Vries <tom@codesourcery.com>
3216 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
3219 2017-11-19 Tom de Vries <tom@codesourcery.com>
3221 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
3224 2017-11-19 Tom de Vries <tom@codesourcery.com>
3226 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
3229 2017-11-19 Tom de Vries <tom@codesourcery.com>
3231 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
3233 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
3235 2017-11-19 Tom de Vries <tom@codesourcery.com>
3237 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
3239 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
3240 semicolon after MCORE_EXPORT_NAME call.
3242 2017-11-19 Tom de Vries <tom@codesourcery.com>
3245 * vmsdbgout.c (vmsdbgout_early_finish): New function.
3246 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
3248 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
3250 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
3251 (cgraph_edge::clone): Cleanup updating of profile.
3252 * ipa-cp.c (update_profiling_info): Likewise.
3253 * ipa-inline-transform.c (inline_transform): Likewise.
3254 * ipa-inline.c (inline_small_functions): Add missing space to dump.
3255 * ipa-split.c (execute_split_functions): Do not split when function
3257 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
3258 * profile-count.c (profile_count::dump): Add global0.
3259 (profile_count::to_cgraph_frequency): Do not ICE when entry is
3261 (profile_count::to_sreal_scale): Likewise.
3262 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
3263 (profile_count::combine_with_ipa_count): New function.
3264 * profile-count.h (profile_guessed_global0adjusted): New.
3265 (profile_count::adjusted_zero): New.
3266 (profile_count::global0adjusted): New.
3267 (profile_count::combine_with_ipa_count): New.
3268 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
3269 correct profile of return block of split functions.
3270 (copy_cfg_body): Remove unused profile_count.
3271 (copy_body): Likewise.
3272 (expand_call_inline): Update.
3273 (tree_function_versioning): Update.
3275 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
3277 * hash-set.h (hash_set::empty): New.
3278 * tree-ssa-threadbackward.h: Delete.
3279 * tree-ssa-threadbackward.c (class thread_jumps): New.
3280 Move max_threaded_paths into class.
3281 (fsm_find_thread_path): Remove arguments that are now in class.
3282 (profitable_jump_thread_path): Rename to...
3283 (thread_jumps::profitable_jump_thread_path): ...this.
3284 (convert_and_register_jump_thread_path): Rename to...
3285 (thread_jumps::convert_and_register_current_path): ...this.
3286 (check_subpath_and_update_thread_path): Rename to...
3287 (thread_jumps::check_subpath_and_update_thread_path): ...this.
3288 (register_jump_thread_path_if_profitable): Rename to...
3289 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
3290 (handle_phi): Rename to...
3291 (thread_jumps::handle_phi): ...this.
3292 (handle_assignment): Rename to...
3293 (thread_jumps::handle_assignment): ...this.
3294 (fsm_find_control_statement_thread_paths): Rename to...
3295 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
3296 (find_jump_threads_backwards): Rename to...
3297 (thread_jumps::find_jump_threads_backwards): ...this.
3298 Initialize path local data.
3299 (pass_thread_jumps::execute): Call find_jump_threads_backwards
3300 from within thread_jumps class.
3301 (pass_early_thread_jumps::execute): Same.
3303 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3305 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
3307 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3309 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
3310 to canonical location.
3312 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3315 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
3317 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
3319 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
3321 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
3322 rcrt1.o%s/grcrt1.o%s for -static-pie.
3324 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3326 * i386.c (ix86_multiplication_cost, ix86_division_cost,
3327 ix86_shift_rotate_cost): Break out from ...
3328 (ix86_rtx_costs): ... here.
3329 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
3332 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3334 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
3335 when entry block was promoted unlikely.
3336 (estimate_bb_frequencies): Increase frequency scale.
3337 * profile-count.h (profile_count): Export precision info.
3339 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3341 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
3342 disturbin profile of entry block.
3344 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3346 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
3349 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3351 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
3352 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
3354 2017-11-17 Jeff Law <law@redhat.com>
3356 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
3357 from evrp_dom_walker class. Various methods moved into new class.
3358 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
3359 (evrp_range_analyzer::enter): New method.
3360 (evrp_range_analyzer::leave): New method.
3361 (evrp_dom_walker): Remove delegators no longer needed by this class.
3362 Replace vr_values data member with evrp_range_analyzer
3364 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
3365 method extracted from evrp_dom_walker::before_dom_children.
3366 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
3367 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
3369 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
3370 Add private copy constructor and move assignment operators.
3371 Privatize methods and class data where trivially possible.
3372 (evrp_dom_walker::cleanup): New function, extracted from
3373 execute_early_vrp. Simplify access to class data.
3375 * vr-values.h (get_output_for_vrp): Prototype.
3376 * vr-values.c (get_output_for_vrp): New function extracted from
3377 vrp_visit_assignment_or_call and extract_range_from_stmt.
3378 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
3380 2017-11-17 Luis Machado <luis.machado@linaro.org>
3382 * config/aarch64/aarch64.c
3383 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
3384 (qdf24xx_tunings) <autoprefetcher_model>: Set to
3385 tune_params::AUTOPREFETCHER_WEAK.
3387 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3390 * config/arm/arm.c (arm_valid_target_attribute_rec):
3391 Parse "arch=" and "+<ext>".
3392 (arm_valid_target_attribute_tree): Re-init global options.
3393 (arm_option_override): Make non-static.
3394 (arm_options_perform_arch_sanity_checks): Make errors fatal.
3395 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
3396 (__ARM_FEATURE_CRC32): Support undef.
3397 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
3398 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
3400 2017-11-17 David Malcolm <dmalcolm@redhat.com>
3402 * gdbinit.in (break-on-diagnostic): New command.
3404 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3406 * config/i386/i386.c (ix86_expand_epilogue): Change simple
3407 return to indirect jump for EH return if control-flow
3408 protection is enabled. Change explicit 'false' argument in
3409 pro_epilogue_adjust_stack with a value of
3411 * config/i386/i386.md (simple_return_indirect_internal):
3412 Remove SImode restriction to support 64-bit.
3414 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3416 * combine.c (added_notes_insn): New.
3417 (try_combine): Handle added_notes_insn like added_links_insn.
3418 Rewrite return value code.
3419 (distribute_notes): Set added_notes_insn to the earliest insn we added
3422 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3424 PR rtl-optimization/82621
3425 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
3426 dest of one of those SETs is unused.
3428 2017-11-17 Richard Biener <rguenther@suse.de>
3431 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
3432 * tree-pretty-print.c (dump_generic_node): Handle
3433 annot_expr_parallel_kind.
3434 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
3435 * gimplify.c (gimple_boolify): Likewise.
3437 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3439 * config.gcc (extra_headers): Add cet.h for x86 targets.
3440 * config/i386/cet.h: New file.
3441 * doc/install.texi: Add --enable-cet/--disable-cet.
3443 2017-11-17 Richard Biener <rguenther@suse.de>
3445 PR tree-optimization/83017
3446 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
3447 (gen_parallel_loop): Properly count iterations.
3448 (parallelize_loops): Handle loop->can_be_parallel independent
3449 of flag_loop_parallelize_all. Make static profitability test match
3451 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
3452 * invoke.texi (parloops-min-per-thread): Document.
3454 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
3456 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
3457 upstreaming review comments.
3459 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3461 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
3462 with fast unaligned access.
3463 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
3465 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
3467 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
3468 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
3470 2017-11-17 Richard Biener <rguenther@suse.de>
3472 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
3473 folding of references.
3475 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
3478 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
3479 of replacing call to strncmp with corresponding call to strcmp when
3482 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
3484 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
3485 option prefer-avx256 for skylake-avx512 configuration.
3486 * config/i386/i386.c (ix86_option_override_internal): Ditto.
3487 (get_builtin_code_for_version): Ditto.
3489 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3490 Monk Chiang <sh.chiang04@gmail.com>
3492 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
3493 (FIXED_REGISTERS): Reserve more register numbers.
3494 (CALL_USED_REGISTERS): Likewise.
3495 (REG_ALLOC_ORDER): Likewise.
3496 (REG_CLASS_CONTENTS): Likewise.
3497 (REGISTER_NAMES): Likewise.
3499 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3500 Kito Cheng <kito.cheng@gmail.com>
3502 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
3504 * config/nds32/iterators.md: Add vector mode iterators and attributes.
3506 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
3508 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
3509 parameter list for vec_splats.
3511 2017-11-16 Joseph Myers <joseph@codesourcery.com>
3513 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
3515 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
3517 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
3520 * doc/install.texi: Document incompatibility of Perl >=5.6.26
3521 with the required version of automake 1.11.6.
3523 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
3525 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
3527 (DU_C2_3_power9): Correct reservation combinations.
3528 (FP_DIV_power9, VEC_DIV_power9): New.
3529 (power9-alu): Split out rotate/shift...
3530 (power9-rot): ...to here, correct dispatch resource.
3531 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
3533 (power9-fp): Correct latency.
3534 (power9-sdiv): Add div/sqrt resource.
3535 (power9-ddiv): Correct latency, add div/sqrt resource.
3536 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
3537 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
3539 (power9-qpdiv, power9-qpmul): Adjust resource usage.
3541 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
3543 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
3544 switch statement mapping KF built-ins to TF built-ins if we don't
3545 have the proper ISA 3.0 assembler support.
3547 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3549 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
3550 (lower_emutls_function_body): Do not compute it.
3552 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3554 * ipa-split.c (split_bb_info): Turn time to sreal.
3555 (split_point): Likewise.
3556 (dump_split_point): Likewise.
3557 (fine_split_points): Likewise.
3558 (execute_split_functions): Only zero split_bbs; turn time to sreals.
3560 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3562 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
3565 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3567 * predict.c (combine_predictions_for_bb): Preserve zero predicted
3569 (expensive_function_p): Remove useless assert.
3570 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
3572 2017-11-16 Martin Sebor <msebor@redhat.com>
3574 PR tree-optimization/82588
3575 PR tree-optimization/82583
3576 * tree-vrp.c (check_array_ref): Handle flexible array members,
3577 string literals, and inner indices.
3578 (search_for_addr_array): Add detail to diagnostics.
3580 2017-11-16 Nathan Sidwell <nathan@acm.org>
3584 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
3585 (SET_DECL_ASSEMBLER_NAME): Forward to
3586 overwrite_decl_assembler_name.
3587 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
3588 (overwrite_decl_assembler_name): Declare.
3589 * tree.c (overwrite_decl_assembler_name): New.
3590 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
3591 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
3592 (LANG_HOOKS_INITIALIZER): Add it.
3593 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
3594 * langhooks.c (lhd_set_decl_assembler_name): Use
3595 SET_DECL_ASSEMBLER_NAME.
3596 (lhd_overwrite_decl_assembler_name): Default implementation.
3598 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
3599 Jackson Woodruff <jackson.woodruff@arm.com>
3601 PR tree-optimization/71026
3602 * match.pd: Canonicalize constant multiplies in division.
3604 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
3606 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
3607 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
3608 Remove OPT_fomit_frame_pointer entry.
3609 * common/config/arc/arc-common.c: Likewise.
3610 * common/config/arm/arm-common.c: Likewise.
3611 * common/config/avr/avr-common.c: Likewise.
3612 * common/config/c6x/c6x-common.c: Likewise.
3613 * common/config/cr16/cr16-common.c: Likewise.
3614 * common/config/cris/cris-common.c: Likewise.
3615 * common/config/epiphany/epiphany-common.c: Likewise.
3616 * common/config/fr30/fr30-common.c: Likewise.
3617 * common/config/frv/frv-common.c: Likewise.
3618 * common/config/ia64/ia64-common.c: Likewise.
3619 * common/config/iq2000/iq2000-common.c: Likewise.
3620 * common/config/lm32/lm32-common.c: Likewise.
3621 * common/config/m32r/m32r-common.c: Likewise.
3622 * common/config/mcore/mcore-common.c: Likewise.
3623 * common/config/microblaze/microblaze-common.c: Likewise.
3624 * common/config/mips/mips-common.c: Likewise.
3625 * common/config/mmix/mmix-common.c: Likewise.
3626 * common/config/mn10300/mn10300-common.c: Likewise.
3627 * common/config/nios2/nios2-common.c: Likewise.
3628 * common/config/pa/pa-common.c: Likewise.
3629 * common/config/pdp11/pdp11-common.c: Likewise.
3630 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
3631 * common/config/riscv/riscv-common.c: Likewise.
3632 * common/config/rs6000/rs6000-common.c: Likewise.
3633 * common/config/rx/rx-common.c: Likewise.
3634 * common/config/s390/s390-common.c: Likewise.
3635 * common/config/sh/sh-common.c: Likewise.
3636 * common/config/sparc/sparc-common.c: Likewise.
3637 * common/config/tilegx/tilegx-common.c: Likewise.
3638 * common/config/tilepro/tilepro-common.c: Likewise.
3639 * common/config/v850/v850-common.c: Likewise.
3640 * common/config/visium/visium-common.c: Likewise.
3641 * common/config/xstormy16/xstormy16-common.c: Likewise.
3642 * common/config/xtensa/xtensa-common.c: Likewise.
3643 * invoke.texi (-fomit-frame-pointer): Update documentation.
3645 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3647 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
3649 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3651 * cfg.c (scale_bbs_frequencies_int,
3652 cale_bbs_frequencies_gcov_type): Remove.
3653 * cfg.h (scale_bbs_frequencies_int,
3654 cale_bbs_frequencies_gcov_type): Remove.
3656 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3658 * tree-ssa-loop-manip.c
3659 (scale_dominated_blocks_in_loop): Update to profile counts.
3660 (tree_transform_and_unroll_loop): Likewise.
3662 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3664 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
3665 scale_bbs_frequencies_int.
3667 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3669 * final.c (compute_alignments): Use counts rather than frequencies.
3671 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3673 * cfgloopanal.c: Include sreal.h
3674 (average_num_loop_insns): Use counts and sreal for accounting.
3676 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3678 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
3681 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3682 Kito Cheng <kito.cheng@gmail.com>
3684 * config/nds32/constraints.md: Provide more constraints.
3685 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
3686 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
3687 support constraints usage.
3689 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3691 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
3693 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3694 Kito Cheng <kito.cheng@gmail.com>
3696 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
3697 * config/nds32/nds32.opt: Refine the layout.
3698 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
3699 TARGET_EXT_STRING): Support new options.
3700 * config/nds32/nds32.h: Likewise.
3701 * config/nds32/nds32.md: Likewise.
3702 * config/nds32/nds32-predicates.c: Likewise.
3703 * config/nds32/constraints.md: Likewise.
3704 * common/config/nds32/nds32-common.c: Likewise.
3706 2017-11-16 Julia Koval <julia.koval@intel.com>
3709 * config/i386/gfniintrin.h: Add sse check.
3710 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
3712 2017-11-16 Julia Koval <julia.koval@intel.com>
3714 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
3715 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
3716 (ix86_handle_option): Handle -mavx512vbmi2.
3717 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
3718 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3719 * config/i386/i386-c.c (__AVX512VBMI2__): New.
3720 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
3721 (ix86_valid_target_attribute_inner_p): Ditto.
3722 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
3723 * config/i386/i386.opt (mavx512vbmi2): New option.
3724 * doc/invoke.texi: Add new option.
3726 2017-11-16 Julia Koval <julia.koval@intel.com>
3728 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
3729 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
3730 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
3731 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
3732 _mm512_gf2p8mul_epi8): New intrinsics.
3733 * config/i386/i386-builtin-types.def
3734 (V64QI_FTYPE_V64QI_V64QI): New type.
3735 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
3736 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
3737 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
3738 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
3739 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
3740 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3742 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
3744 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
3745 explicitly as a stream of bytes.
3747 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3749 * config/rs6000/altivec.h (vec_xst_be): New #define.
3750 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
3751 and externalize from *altivec_vperm_<mode>_internal.
3752 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
3754 (XL_BE_V8HI): Likewise.
3755 (XL_BE_V4SI): Likewise.
3756 (XL_BE_V4SI): Likewise.
3757 (XL_BE_V2DI): Likewise.
3758 (XL_BE_V4SF): Likewise.
3759 (XL_BE_V2DF): Likewise.
3760 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
3761 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
3762 all array entries with these keys: VSX_BUILTIN_VEC_XL,
3763 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
3764 VSX_BUILTIN_VEC_XST_BE.
3765 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
3766 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
3768 (altivec_init_builtins): Replace conditional calls to def_builtin
3769 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
3770 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
3771 with unconditional calls. Remove calls to def_builtin for
3772 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
3773 __builtin_vec_xst_be.
3774 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
3775 to define_expand, and add alternate RTL generation for P8.
3776 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
3777 vsx_ld_elemrev_v8hi.
3778 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
3779 add alternate RTL generation for P8.
3780 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
3781 vsx_ld_elemrev_v16qi.
3782 (vsx_st_elemrev_v8hi): Convert define_insn
3783 to define_expand, and add alternate RTL generation for P8.
3784 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
3785 vsx_st_elemrev_v8hi.
3786 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
3787 add alternate RTL generation for P8.
3788 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
3789 vsx_st_elemrev_v16qi.
3791 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
3794 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
3795 TARGET_AVX512ER check.
3796 (ix86_option_override_internal): Set MASK_VZEROUPPER if
3797 neither -mzeroupper nor -mno-zeroupper is used and
3798 TARGET_EMIT_VZEROUPPER is set.
3799 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
3800 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
3802 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
3804 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3805 folding of vector compares.
3806 (fold_build_vec_cmp): New helper function.
3807 (fold_compare_helper): New helper function.
3808 (builtin_function_type): Add compare builtins to the list of functions
3809 having unsigned arguments. Cosmetic updates to comment indentation.
3810 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
3811 the not+eq combination.
3813 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3815 PR tree-optimization/82726
3816 PR tree-optimization/70754
3817 * tree-predcom.c (order_drefs_by_pos): New function.
3818 (combine_chains): Move code setting has_max_use_after to...
3819 (try_combine_chains): ...here. New parameter. Sort combined chains
3820 according to position information.
3821 (tree_predictive_commoning_loop): Update call to above function.
3822 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
3824 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3826 PR tree-optimization/82726
3828 2017-01-23 Bin Cheng <bin.cheng@arm.com>
3830 PR tree-optimization/70754
3831 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
3832 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
3833 combined stmt before it if not NULL.
3834 (combine_chains): Process refs reversely and compute dominance point
3838 2017-02-23 Bin Cheng <bin.cheng@arm.com>
3840 PR tree-optimization/79663
3841 * tree-predcom.c (combine_chains): Process refs in reverse order
3842 only for ZERO length chains, and add explaining comment.
3844 2017-11-15 Tamar Christina <tamar.christina@arm.com>
3846 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
3847 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
3848 * doc/invoke.texi (ARM Options): Add armv8.3-a.
3850 2017-11-15 Tamar Christina <tamar.christina@arm.com>
3852 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
3854 2017-11-15 Martin Liska <mliska@suse.cz>
3856 * tree-cfg.c (pass_warn_function_return::execute):
3857 Compare warn_return_type for greater than zero.
3859 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
3863 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
3864 to return true on Xeon and not on Xeon Phi.
3865 (ix86_check_avx256_register): Changed to ...
3866 (ix86_check_avx_upper_register): ... this. Add extra check for
3867 VALID_AVX512F_REG_OR_XI_MODE.
3868 (ix86_avx_u128_mode_needed): Changed
3869 ix86_check_avx256_register to ix86_check_avx_upper_register.
3870 (ix86_check_avx256_stores): Changed to ...
3871 (ix86_check_avx_upper_stores): ... this. Changed
3872 ix86_check_avx256_register to ix86_check_avx_upper_register.
3873 (ix86_avx_u128_mode_after): Changed
3874 avx_reg256_found to avx_upper_reg_found. Changed
3875 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
3876 (ix86_avx_u128_mode_entry): Changed
3877 ix86_check_avx256_register to ix86_check_avx_upper_register.
3878 (ix86_avx_u128_mode_exit): Ditto.
3879 * config/i386/i386.h: (host_detect_local_cpu): New define.
3881 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3883 * config/arm/xgene1.md (xgene1): Split into automatons
3884 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
3885 (xgene1_f_load): Adjust reservations and/or types.
3886 (xgene1_f_store): Likewise.
3887 (xgene1_load_pair): Likewise.
3888 (xgene1_store_pair): Likewise.
3889 (xgene1_fp_load1): Likewise.
3890 (xgene1_load1): Likewise.
3891 (xgene1_store1): Likewise.
3892 (xgene1_move): Likewise.
3893 (xgene1_alu): Likewise.
3894 (xgene1_simd): Likewise.
3895 (xgene1_bfm): Likewise.
3896 (xgene1_neon_load1): Likewise.
3897 (xgene1_neon_store1): Likewise.
3898 (xgene1_neon_logic): Likewise.
3899 (xgene1_neon_st1): Likewise.
3900 (xgene1_neon_ld1r): Likewise.
3901 (xgene1_alu_cond): Added.
3902 (xgene1_shift_reg): Likwise.
3903 (xgene1_bfx): Likewise.
3904 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
3906 2017-11-15 Jakub Jelinek <jakub@redhat.com>
3909 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
3911 (can_widen_mult_without_libcall): New function.
3912 (expand_mul_overflow): If only checking unsigned mul overflow,
3913 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
3914 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
3915 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
3916 (expand_DIVMOD): Formatting fix.
3917 * expmed.h (expand_mult): Add NO_LIBCALL argument.
3918 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
3919 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
3921 PR tree-optimization/82977
3922 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
3923 constructed temporary to strlen_to_stridx.put.
3925 2017-11-15 Martin Liska <mliska@suse.cz>
3927 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
3928 * configure: Regenerate.
3930 2017-11-15 Martin Liska <mliska@suse.cz>
3933 * config/sh/sh-mem.cc: Use proper probability for
3936 2017-11-14 Jeff Law <law@redhat.com>
3938 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
3939 the red zone for stack_clash_protection_final_dynamic_probe targets
3940 when the total dynamic stack size is zero bytes.
3942 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
3943 blocks is post order.
3945 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
3947 * dumpfile.h (TDF_COMPARE_DEBUG): New.
3948 * final.c (rest_of_clean_state): Set it for the
3949 -fcompare-debug dump.
3950 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
3951 class when TDF_COMPARE_DEBUG is set.
3953 * dwarf2out.c (gen_producer_string): Discard
3956 2017-11-15 Joseph Myers <joseph@codesourcery.com>
3959 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
3960 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
3961 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
3962 (__TGMATH_REAL_2_3): Remove macros.
3963 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
3964 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
3966 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
3968 (remquo): Define using __TGMATH_REAL_3.
3970 2017-11-14 Jeff Law <law@redhat.com>
3972 * vr-values.c: New file with contents extracted from tree-vrp.c.
3973 * Makefile.in (OBJS): Add vr-values.o
3974 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
3975 (set_value_range, set_and_canonicalize_value_range): Likewise.
3976 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
3977 (value_range_constant_singleton, symbolic_range_p): Likewise.
3978 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
3979 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
3980 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
3981 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
3982 (find_case_label_range, find_case_label_index): Likewise.
3983 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
3984 (range_int_cst_singleton_p, value_inside_range): Likewise.
3985 (get_single_symbol): Likewise.
3986 (switch_update): Move structure definition here.
3987 (to_remove_edges, to_update_switch_stmts): Provide externs.
3988 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
3989 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
3990 (vrp_val_is_min, set_value_range): Likewise.
3991 (set_and_canonicalize_value_range, copy_value_range): Likewise.
3992 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
3993 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
3994 (range_is_nonnull, range_int_cst_p): Likewwise.
3995 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
3996 (get_single_symbol, operand_less_p): Likewise
3997 (compare_values_warnv, compare_values): Likewise.
3998 (value_inside_range, value_range_constant_singleton): Likewise.
3999 (zero_nonzero_bitgs_from_vr): Likewise.
4000 (extract_range_from_binary_expr_1): Likewise.
4001 (overflow_comparison_p): Likewise.
4002 (to_remove_edges, to_update_switch_stmts): Likewise.
4003 (find_case_label-index, find_case_label_range): Likewise.
4004 (switch_update, set_value_range_to_nonnegative): Remove.
4005 (set_value_range_to_truthvalue): Likewise.
4006 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
4007 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
4008 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
4009 (find_case_label_ranges, test_for_singularity): Likewise.
4010 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
4011 (x_vr_values): Move to its remaining use site.
4013 2017-11-10 Jeff Law <law@redhat.com>
4015 * vr-values.h (VR_INITIALIZER): Move #define here.
4016 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
4017 * Makefile.in (OBJS): Add tree-evrp.o
4018 * tree-vrp.h (assert_info): Move structure definition here.
4019 (set_value_range_to_varying): Prototype.
4020 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
4021 (infer_value_range, register_edge_assert_for): Likewise.
4022 (stmt_interesting_for_vrp): Likewise.
4023 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
4024 (set_value_range_to_varying): No longer static.
4025 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
4026 (infer_value_range, register_edge_assert_for): Likewise.
4028 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
4030 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
4031 generate the XXBRD instruction.
4033 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
4034 (rs6000_builtin_type_compatible): Treat _Float128 and long double
4035 as being compatible if -mabi=ieeelongdouble.
4036 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
4037 to setup float128 built-ins with hardware support.
4038 (BU_FLOAT128_HW_2): Likewise.
4039 (BU_FLOAT128_HW_3): Likewise.
4040 (BU_FLOAT128_HW_VSX_1): Likewise.
4041 (BU_FLOAT128_HW_VSX_2): Likewise.
4042 (scalar_extract_expq): Change float128 built-in functions to
4043 accommodate having both KFmode and TFmode functions. Use the
4044 KFmode variant as the default.
4045 (scalar_extract_sigq): Likewise.
4046 (scalar_test_neg_qp): Likewise.
4047 (scalar_insert_exp_q): Likewise.
4048 (scalar_insert_exp_qp): Likewise.
4049 (scalar_test_data_class_qp): Likewise.
4050 (sqrtf128_round_to_odd): Delete processing the round to odd
4051 built-in functions as special built-in functions, and define them
4052 as float128 built-ins. Use the KFmode variant as the default.
4053 (truncf128_round_to_odd): Likewise.
4054 (addf128_round_to_odd): Likewise.
4055 (subf128_round_to_odd): Likewise.
4056 (mulf128_round_to_odd): Likewise.
4057 (divf128_round_to_odd): Likewise.
4058 (fmaf128_round_to_odd): Likewise.
4059 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
4060 support for KFmode and TFmode xststdcqp calls.
4061 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
4062 point, switch the built-in handlers for the get/set float128
4063 exponent, get float128 mantissa, float128 test built-ins, and the
4064 float128 round to odd built-in functions. Eliminate creating the
4065 float128 round to odd built-in functions as special built-ins.
4066 (rs6000_init_builtins): Eliminate special creation of the float128
4067 round to odd built-in functions.
4068 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
4069 function insns to support both TFmode and KFmode variants.
4070 (xsxsigqp_<mode>): Likewise.
4071 (xsiexpqpf_<mode>): Likewise.
4072 (xsiexpqp_<mode>): Likewise.
4073 (xststdcqp_<mode>): Likewise.
4074 (xststdcnegqp_<mode>): Likewise.
4075 (xststdcqp_<mode>): Likewise.
4077 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4079 * tree-ssa-threadupdate.c (compute_path_counts): Remove
4080 unused path_in_freq_ptr parameter.
4081 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
4083 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4085 * ipa-inline.c (edge_badness): Dump sreal frequency.
4086 (compute_inlined_call_time): Match natural implementaiton ...
4087 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
4088 forgotten division by CGRAPH_FREQ_BASE.
4090 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4092 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
4093 Solaris 11. Update comment.
4094 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
4096 * config/sol2.h (STARTFILE_SPEC): Likewise.
4097 * configure: Regenerate.
4099 2017-11-14 Carl Love <cel@us.ibm.com>
4101 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
4102 le_ and be_ prefixes to swap* variables. Remove
4103 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
4106 2017-11-14 Jason Merrill <jason@redhat.com>
4108 Support GTY((cache)) on hash_map.
4109 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
4110 (ggc_cache_remove): Override it instead of ggc_mx.
4111 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
4112 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
4113 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
4114 (simple_cache_map_traits): Override maybe_mx.
4115 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
4116 (hash_map): Friend gt_cleare_cache.
4117 (gt_cleare_cache): New.
4118 * tree.h (tree_cache_traits): New hash_map traits class.
4119 (tree_cache_map): New typedef.
4121 2017-11-14 Richard Biener <rguenther@suse.de>
4123 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
4124 paramter and handling.
4125 (cleanup_control_flow_bb): Likewise.
4126 (cleanup_control_flow_pre): New helper performing a DFS walk
4127 to call cleanup_control_flow_bb in PRE order.
4128 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
4129 via cleanup_control_flow_pre.
4131 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
4133 * config/aarch64/aarch64-simd.md
4134 (aarch64_simd_bsl<mode>_internal): Remove DImode.
4135 (*aarch64_simd_bsl<mode>_alt): Likewise.
4136 (aarch64_simd_bsldi_internal): New.
4137 (aarch64_simd_bsldi_alt): Likewise.
4139 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4141 * tracer.c (better_p): Do not compare frequencies.
4142 * reg-stack.c (better_edge): Likewise.
4143 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
4146 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4148 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
4149 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
4150 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
4151 * lto-streamer-in.c (input_function): Use update_max_bb_count.
4152 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
4153 * predict.c (maybe_hot_frequency_p): Inline to ...
4154 (maybe_hot_count_p): ... here; rewrite to counts.
4155 (counts_to_freqs): Rename to ...
4156 (update_max_bb_count): ... this one.
4157 (expensive_function_p): Use counts.
4158 (estimate_bb_frequencies): Update.
4159 (rebuild_frequencies): Update.
4160 * predict.h (counts_to_freqs): Rename to ...
4161 (update_max_bb_count): ... this one.
4162 * profile.c (compute_branch_probabilities): Add debug info
4163 * tree-inline.c (expand_call_inline): Update debug info.
4164 (optimize_inline_calls): Use update_max_bb_count..
4165 (tree_function_versioning): Use update_max_bb_count..
4166 * value-prof.c (gimple_value_profile_transformations):
4167 Do not use update_max_bb_count.
4169 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4171 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
4172 always use frequencies.
4174 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4176 * bb-reorder.c: Remove frequencies from comments.
4177 (better_edge_p): Use profile counts.
4178 (find_traces): Dump profile counts.
4179 (rotate_loop): Use profile counts.
4180 (find_traces_1_round): Likewise.
4181 (connect_better_edge_p): Use counts instead of probabilities for
4183 (copy_bb_p): Drop early check for non-0 frequency.
4184 (sanitize_hot_paths): Update comments.
4186 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4188 * ipa-split.c (struct split_point): Add count.
4189 (consider_split): Do not compute incoming frequency; compute incoming
4190 count and store it to split_point.
4191 (split_function): Set count of the call to split part correctly.
4193 2017-11-13 Carl Love <cel@us.ibm.com>
4195 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
4197 2017-11-13 Tom Tromey <tom@tromey.com>
4199 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
4201 2017-11-13 Carl Love <cel@us.ibm.com>
4203 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
4204 Add support for builtins:
4205 unsigned int vec_first_{,miss}_match_{,or_eos}index,
4206 vector {un,}signed {char,int,short},
4207 vector {un,}signed {char,int,short}) arguments.
4208 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
4209 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
4210 Add BU_P9V_AV_2 expansions for the builtins.
4211 * config/rs6000/altivec.h (vec_first_match_index,
4212 vec_first_mismatch_index, vec_first_match_or_eos_index,
4213 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
4214 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
4215 new extern declaration.
4216 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
4217 * config/rs6000/vsx.md (first_match_index_<mode>,
4218 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
4219 first_mismatch_or_eos_index_<mode>): Add define expand.
4220 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
4221 * doc/extend.texi: Update the built-in documenation file for the new
4224 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
4226 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
4227 into the min/max operations for _Float<N> and _Float<N>X types.
4229 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4232 * dwarf2out.c (do_eh_frame): New static variable.
4233 (dwarf2out_begin_prologue): Set it.
4234 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
4236 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4238 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
4240 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
4242 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4245 * debug.h (dwarf2out_do_eh_frame): Declare.
4246 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
4247 (dwarf2out_do_frame): Use it.
4248 (dwarf2out_do_cfi_asm): Likewise.
4249 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
4250 (dwarf2out_assembly_start): Likewise.
4251 (dwarf2out_begin_prologue): Fix comment.
4252 * toplev.c (compile_file): Always call dwarf2out_frame_finish
4253 if the target needs either debug or unwind DWARF2 info.
4254 * lto-opts.c (lto_write_options): Do not save -fexceptions,
4255 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
4256 -ftrapping-math, -ftrapv and -fwrapv.
4258 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4260 * cgraph.c (cgraph_edge::sreal_frequency): New function.
4261 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
4262 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
4263 (estimate_edge_size_and_time): Likewise.
4264 (ipa_merge_fn_summary_after_inlining): Likewise.
4265 * ipa-inline.c (cgraph_freq_base_rec): Remove.
4266 (compute_uninlined_call_time): Use sreal_frequency.
4267 (compute_inlined_call_time): Likewise.
4268 (ipa_inline): Do not initialize cgraph_freq_base_rec.
4269 * profile-count.c: Include sreal.h.
4270 (profile_count::to_sreal_scale): New.
4271 * profile-count.h: Forward declare sreal.
4272 (profile_count::to_sreal_scale): Declare.
4274 2017-11-13 Nathan Sidwell <nathan@acm.org>
4276 * diagnostic.c (maybe_line_and_column): New.
4277 (diagnostic_get_location_text): Use it.
4278 (diagnostic_report_current_module): Likewise.
4279 (test_diagnostic_get_location_text): Add tests.
4281 2017-11-13 Luis Machado <luis.machado@linaro.org>
4283 * doc/md.texi (Specifying processor pipeline description): Fix
4284 incorrect latency for the div instruction example.
4286 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4288 PR tree-optimization/78821
4289 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
4290 that bit_not_p is the same.
4291 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
4292 (split_group): Count precisely bit_not_p bits in each statement.
4293 (invert_op): New function.
4294 (imm_store_chain_info::output_merged_store): Use invert_op to
4295 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
4296 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
4298 2017-11-13 Martin Liska <mliska@suse.cz>
4300 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
4301 (struct source_info): Likewise.
4302 (add_branch_counts): Likewise.
4303 (add_line_counts): Likewise.
4304 (function_summary): Likewise.
4305 (output_intermediate_line): Likewise.
4306 (generate_results): Likewise.
4308 2017-11-13 Martin Liska <mliska@suse.cz>
4310 * gcov.c (struct block_info): Remove typedef for block_t.
4311 (struct line_info): Likewise.
4312 (line_info::has_block): Likewise.
4313 (EXIT_BLOCK): Likewise.
4314 (unblock): Likewise.
4315 (circuit): Likewise.
4316 (get_cycles_count): Likewise.
4317 (process_file): Likewise.
4318 (read_graph_file): Likewise.
4319 (solve_flow_graph): Likewise.
4320 (find_exception_blocks): Likewise.
4321 (add_line_counts): Likewise.
4322 (accumulate_line_info): Likewise.
4323 (output_line_details): Likewise.
4325 2017-11-13 Martin Liska <mliska@suse.cz>
4327 * gcov.c (struct arc_info): Remove typedef for arc_t.
4328 (struct line_info): Likewise.
4329 (add_branch_counts): Likewise.
4330 (output_branch_count): Likewise.
4331 (function_info::~function_info): Likewise.
4332 (circuit): Likewise.
4333 (output_intermediate_line): Likewise.
4334 (read_graph_file): Likewise.
4335 (solve_flow_graph): Likewise.
4336 (find_exception_blocks): Likewise.
4337 (add_line_counts): Likewise.
4338 (accumulate_line_info): Likewise.
4339 (output_line_details): Likewise.
4340 (output_function_details): Likewise.
4342 2017-11-13 Martin Liska <mliska@suse.cz>
4344 * gcov.c (struct function_info): Remove typedef for function_t.
4345 (struct source_info): Likewise.
4346 (source_info::get_functions_at_location): Likewise.
4347 (solve_flow_graph): Likewise.
4348 (find_exception_blocks): Likewise.
4349 (add_line_counts): Likewise.
4350 (output_intermediate_file): Likewise.
4351 (process_file): Likewise.
4352 (generate_results): Likewise.
4353 (release_structures): Likewise.
4354 (read_graph_file): Likewise.
4355 (read_count_file): Likewise.
4356 (accumulate_line_counts): Likewise.
4357 (output_lines): Likewise.
4359 2017-11-13 Martin Liska <mliska@suse.cz>
4361 * gcov.c (function_info::function_info): Remove num_counts
4362 and add vector<gcov_type>.
4363 (function_info::~function_info): Use the vector.
4364 (process_file): Likewise.
4365 (read_graph_file): Likewise.
4366 (read_count_file): Likewise.
4367 (solve_flow_graph): Likewise.
4369 2017-11-13 Martin Liska <mliska@suse.cz>
4371 * gcov.c (function_info::is_artificial): New function.
4372 (process_file): Erase all artificial early.
4373 (generate_results): Skip as all artificial are already
4376 2017-11-13 Martin Liska <mliska@suse.cz>
4378 * gcov.c (read_graph_file): Store to global vector of functions.
4379 (read_count_file): Iterate the vector.
4380 (process_file): Likewise.
4381 (generate_results): Likewise.
4382 (release_structures): Likewise.
4384 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4386 PR tree-optimization/82954
4387 * gimple-ssa-store-merging.c
4388 (imm_store_chain_info::coalesce_immediate_stores): If
4389 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
4391 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
4393 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
4394 Upddate call to ENDIAN_LANE_N.
4395 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
4396 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
4397 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
4398 and use aarch64_endian_lane_rtx.
4399 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
4401 2017-11-12 Tom de Vries <tom@codesourcery.com>
4403 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
4405 2017-11-12 Tom de Vries <tom@codesourcery.com>
4407 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
4409 2017-11-12 Tom de Vries <tom@codesourcery.com>
4411 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
4413 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4414 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4415 * defaults.h (ASM_OUTPUT_LABELREF): Same.
4417 2017-11-11 Martin Sebor <msebor@redhat.com>
4420 * doc/extend.texi (attribute nonstring): Remove spurious argument.
4423 * prefic.c (translate_name): Replace strncpy with memcpy to
4424 avoid -Wstringop-truncation.
4426 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4428 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
4430 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4432 * predict.c (maybe_hot_frequency_p): Do not use cfun.
4434 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4436 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
4439 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
4441 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
4442 enable generating XXBRH if the value is in a vector register.
4443 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
4444 value is in a vector register.
4445 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
4446 register to register bswap64's instead of doing the GPR sequence
4447 used on previous machines.
4448 (bswapdi2_xxbrd): New insn.
4449 (bswapdi2_reg): Disallow on ISA 3.0.
4450 (register to register bswap64 splitter): Do not split the insn on
4451 ISA 3.0 systems that use XXBRD.
4453 2017-11-10 Martin Sebor <msebor@redhat.com>
4456 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
4458 (find_subframework_file): Same.
4460 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4462 * auto-profile.c (afdo_indirect_call): Drop frequency.
4463 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
4464 (cgraph_node::create_edge): Drop frequency argument.
4465 (cgraph_node::create_indirect_edge): Drop frequency argument.
4466 (cgraph_edge::make_speculative): Drop frequency arguments.
4467 (cgraph_edge::resolve_speculation): Do not update frequencies
4468 (cgraph_edge::dump_edge_flags): Do not dump frequency.
4469 (cgraph_node::dump): Check consistency in IPA mode.
4470 (cgraph_edge::maybe_hot_p): Use IPA counter.
4471 (cgraph_edge::verify_count_and_frequency): Rename to ...
4472 (cgraph_edge::verify_count): ... this one; drop frequency checking.
4473 (cgraph_node::verify_node): Update.
4474 * cgraph.h (struct cgraph_edge): Drop frequency.
4475 (cgraph_edge::frequency): New function.
4476 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
4478 (cgraph_edge::rebuild_edges): Likewise.
4479 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
4480 (duplicate_thunk_for_node): Do not pass frequency.
4481 (cgraph_node::create_clone): Scale only counts.
4482 (cgraph_node::create_virtual_clone): Do not pass frequency.
4483 (cgraph_node::create_edge_including_clones): Do not pass frequency.
4484 (cgraph_node::create_version_clone): Do not pass frequency.
4485 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
4486 (cgraph_node::expand_thunk): Do not pass frequency.
4487 (cgraph_node::create_wrapper): Do not pass frequency.
4488 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
4490 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
4491 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
4492 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
4493 (gather_caller_stats): Use frequency function.
4494 (ipcp_cloning_candidate_p): Use frequency function.
4495 (ipcp_propagate_stage): Use frequency function.
4496 (get_info_about_necessary_edges): Use frequency function.
4497 (update_profiling_info): Update only IPA profile.
4498 (update_specialized_profile): Use frequency functoin.
4499 (perhaps_add_new_callers): Update only IPA profile.
4500 * ipa-devirt.c (ipa_devirt): Use IPA profile.
4501 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
4502 (dump_ipa_call_summary): Use frequency function.
4503 (estimate_edge_size_and_time): Use frequency function.
4504 (ipa_merge_fn_summary_after_inlining): Use frequency function.
4505 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
4506 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
4507 (update_noncloned_counts): ... ths one; scale counts only.
4508 (clone_inlined_nodes): Do not scale frequency.
4509 (inline_call): Do not pass frequency.
4510 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
4511 (compute_inlined_call_time): Use IPA profile.
4512 (want_inline_small_function_p): Use IPA profile.
4513 (want_inline_self_recursive_call_p): Use IPA profile.
4514 (edge_badness): Use IPA profile.
4515 (lookup_recursive_calls): Use IPA profile.
4516 (recursive_inlining): Do not pass frequency.
4517 (resolve_noninline_speculation): Do not update frequency.
4518 (inline_small_functions): Collect max of IPA profile.
4519 (dump_overall_stats): Dump IPA porfile.
4520 (dump_inline_stats): Dump IPA porfile.
4521 (ipa_inline): Collect IPA stats.
4522 * ipa-inline.h (clone_inlined_nodes): Update prototype.
4523 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
4524 (ipa_propagate_frequency): Use frequency function.
4525 (ipa_profile): Cleanup.
4526 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
4527 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
4528 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
4529 (input_node): Do not stream frequency.
4530 (input_edge): Do not stream frequency.
4531 (merge_profile_summaries): Scale only IPA profiles.
4532 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
4533 * predict.c (drop_profile): Do not recompute frequency.
4534 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
4535 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
4536 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
4537 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
4538 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
4539 * tree-inline.c (copy_bb): Do not scale frequency.
4540 (expand_call_inline): Do not scale frequency.
4541 (tree_function_versioning): Do not scale frequency.
4542 * ubsan.c (ubsan_create_edge): Do not pass frequency.
4544 2017-11-10 Julia Koval <julia.koval@intel.com>
4546 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
4547 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
4548 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
4549 (_mm256_maskz_gf2p8affine_epi64_epi8)
4550 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
4551 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
4552 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
4553 (__builtin_ia32_vgf2p8affineqb_v32qi)
4554 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
4555 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
4557 2017-11-10 Tamar Christina <tamar.christina@arm.com>
4561 (arm_option_override): Refactor.
4562 (arm_option_reconfigure_globals): New.
4563 (arm_options_perform_arch_sanity_checks): New.
4564 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
4566 (arm_options_perform_arch_sanity_checks): Likewise
4568 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
4570 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
4571 (power9-qpmul): New.
4572 * rs6000/rs6000.md ("type" attr): Add qmul.
4573 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
4574 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
4575 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
4577 2017-11-10 Martin Sebor <msebor@redhat.com>
4580 * builtins.c (compute_objsize): Handle arrays that
4581 compute_builtin_object_size likes to fail for. Make extern.
4582 * builtins.h (compute_objsize): Declare.
4583 (check_strncpy_sizes): New function.
4584 (expand_builtin_strncpy): Call check_strncpy_sizes.
4585 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
4586 -Wstringop-truncation.
4587 (gimple_fold_builtin_strncat): Same.
4588 * gimple.c (gimple_build_call_from_tree): Set call location.
4589 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
4590 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
4591 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
4592 (handle_builtin_strlen): Use strlen_to_stridx.
4593 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
4595 Use strlen_to_stridx.
4596 (pass_strlen::execute): Release strlen_to_stridx.
4597 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
4598 (-Wstringop-truncation): Document new option.
4600 2017-11-10 Martin Liska <mliska@suse.cz>
4602 PR gcov-profile/82702
4603 * gcov.c (main): Handle intermediate files in a different
4605 (get_gcov_intermediate_filename): New function.
4606 (output_gcov_file): Remove support of intermediate files.
4607 (generate_results): Allocate intermediate file.
4608 (release_structures): Clean-up properly fn_end.
4609 (output_intermediate_file): Start iterating with line 1.
4611 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4613 PR tree-optimization/82929
4614 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
4615 ops_swapped_p non-static data member.
4616 (store_immediate_info::store_immediate_info): Clear it.
4617 (imm_store_chain_info::coalesce_immediate_stores): If swapping
4618 ops set ops_swapped_p.
4619 (count_multiple_uses): Handle ops_swapped_p.
4621 2017-11-10 Martin Liska <mliska@suse.cz>
4623 * coverage.c (coverage_init): Stream information about
4624 support of has_unexecuted_blocks.
4625 * doc/gcov.texi: Document that.
4626 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
4627 * gcov.c (read_graph_file): Likewise.
4628 (output_line_beginning): Fix a small issue with
4631 2017-11-10 Bin Cheng <bin.cheng@arm.com>
4633 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
4634 reference of trivial component.
4636 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4639 * gimple-ssa-store-merging.c
4640 (pass_store_merging::terminate_all_aliasing_chains): For
4641 gimple_store_p stmts also call refs_output_dependent_p.
4643 PR rtl-optimization/82913
4644 * compare-elim.c (try_merge_compare): Punt if def_insn is not
4647 2017-11-09 Jeff Law <law@redhat.com>
4649 * vr-values.h: New file with vr_values class.
4650 * tree-vrp.c: Include vr-values.h
4651 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
4652 data objects into the vr_values class.
4653 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
4654 (get_value_range): Make it a member function within vr_values class.
4655 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
4656 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
4657 (op_with_constant_singleton_value_range): Likewise.
4658 (extract_range_for_var_from_comparison_expr): Likewise.
4659 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
4660 (extract_range_from_binary_expr): Likewise.
4661 (extract_range_from_unary_expr): Likewise.
4662 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
4663 (check_for_binary_op_overflow, extract_range_basic): Likewise.
4664 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
4665 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
4666 (compare_name_with_value, compare_names): Likewise.
4667 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
4668 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
4669 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
4670 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
4671 (extract_range_from_phi_node): Likewise.
4672 (simplify_truth_ops_using_ranges): Likewise.
4673 (simplify_div_or_mod_using_ranges): Likewise.
4674 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
4675 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
4676 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
4677 (simplify_float_conversion_using_ranges): Likewise.
4678 (simplify_internal_call_using_ranges): Likewise.
4679 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
4680 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
4681 poitner via x_vr_values for calls into gimple folder.
4682 (vrp_initialize_lattice): Make this the vr_values ctor.
4683 (vrp_free_lattice): Make this the vr_values dtor.
4684 (set_vr_value): New function.
4685 (class vrp_prop): Add vr_values data member. Add various member
4686 functions as well as member functions that delegate to vr_values.
4687 (check_array_ref): Make a member function within vrp_prop class.
4688 (search_for_addr_array, vrp_initialize): Likewise.
4689 (vrp_finalize): Likewise. Revamp to avoid direct access to
4690 vr_value, values_propagated, etc.
4691 (check_array_bounds): Extract vrp_prop class instance pointer from
4692 walk info structure. Use it to call member functions.
4693 (check_all_array_refs): Make a member function within vrp_prop class.
4694 Smuggle class instance pointer via walk info structure.
4695 (x_vr_values): New local static.
4696 (vrp_valueize): Use x_vr_values to get class instance.
4697 (vr_valueize_1): Likewise.
4698 (class vrp_folder): Add vr_values data member. Add various member
4699 functions as well as member functions that delegate to vr_values.
4700 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
4701 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
4702 class instance from vr_values. Use it to call member functions.
4703 (vrp_dom_walker): Add vr_values data member.
4704 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
4705 instance via x_vr_values.
4706 (identify_jump_threads): Accept vr_values as argument. Store
4707 it into the walker structure.
4708 (evrp_dom_walker): Add vr_values class data member. Add various
4710 (evrp_dom_walker::try_find_new_range): Use vr_values data
4711 member to access the memory allocator.
4712 (evrp_dom_walker::before_dom_children): Store vr_values class
4713 instance into the vrp_folder class.
4714 (evrp_dom_walker::push_value_range): Rework to avoid direct
4715 access to num_vr_values and vr_value.
4716 (evrp_dom_walker::pop_value_range): Likewise.
4717 (execute_early_vrp): Remove call to vrp_initialize_lattice.
4718 Use vr_values to get to dump_all_value_ranges member function.
4719 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
4720 and simplify_cond_using_ranges_2 via vrp_prop class instance.
4721 Pass vr_values class instance down to identify_jump_threads.
4722 Remove call to vrp_free_lattice.
4723 (debug_all_value_ranges): Remove.
4725 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
4726 (vrp_folder): Likewise.
4728 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
4729 Get it from the existing bitmap instead.
4730 (vrp_intersect_ranges_1): Likewise.
4732 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4734 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
4736 (store_immediate_info::store_immediate_info): Add bitnotp argument,
4737 set bit_not_p to it.
4738 (imm_store_chain_info::coalesce_immediate_stores): Break group
4739 if bit_not_p is different.
4740 (count_multiple_uses, split_group,
4741 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
4742 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
4743 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
4744 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
4745 to store_immediate_info ctor.
4747 2017-11-09 Jim Wilson <jimw@sifive.com>
4749 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
4750 (scan_prog_file): Likewise.
4752 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4754 * bb-reorder.c (max_entry_frequency): Remove.
4755 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
4756 connect_traces, push_to_next_round_p): Remove prototypes.
4757 (find_traces_1_round): Use counts only.
4758 (push_to_next_round_p): Likewise.
4759 (find_traces): Likewise.
4760 (rotate_loop): Likewise.
4761 (find_traces_1_round): Likewise.
4762 (connect_traces): Likewise.
4763 (edge_order): Likewise.
4765 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
4767 * config/arm/arm.c (output_return_instruction): Add comments to
4768 indicate requirement for cmse_nonsecure_entry return to account
4769 for the size of clearing instruction output here.
4770 (thumb_exit): Likewise.
4771 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
4772 return in hardfloat mode.
4774 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
4776 * config/rs6000/rs6000.c (machine_function): Add a bool,
4777 "toc_is_wrapped_separately".
4778 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
4779 if it wasn't explicitly set or unset, we are optimizing for speed, and
4780 doing separate shrink-wrapping.
4781 (rs6000_get_separate_components): Enable the TOC component if
4782 saving the TOC register in the prologue.
4783 (rs6000_components_for_bb): Handle the TOC component.
4784 (rs6000_emit_prologue_components): Store the TOC register where needed.
4785 (rs6000_set_handled_components): Mark TOC as handled, if handled.
4786 (rs6000_emit_prologue): Don't save the TOC if that is already done.
4788 2017-11-09 Martin Jambor <mjambor@suse.cz>
4790 * ipa-param-manipulation.c: New file.
4791 * ipa-param-manipulation.h: Likewise.
4792 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
4793 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
4794 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
4795 (ipa_parm_adjustment): Likewise.
4796 (ipa_parm_adjustment_vec): Likewise.
4797 (ipa_get_vector_of_formal_parms): Moved declaration to
4798 ipa-param-manipulation.h.
4799 (ipa_get_vector_of_formal_parm_types): Likewise.
4800 (ipa_modify_formal_parameters): Likewise.
4801 (ipa_modify_call_arguments): Likewise.
4802 (ipa_combine_adjustments): Likewise.
4803 (ipa_dump_param_adjustments): Likewise.
4804 (ipa_modify_expr): Likewise.
4805 (ipa_get_adjustment_candidate): Likewise.
4806 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
4807 ipa-param-manipulation.c.
4808 (ipa_get_vector_of_formal_parm_types): Likewise.
4809 (ipa_modify_formal_parameters): Likewise.
4810 (ipa_modify_call_arguments): Likewise.
4811 (ipa_modify_expr): Likewise.
4812 (get_ssa_base_param): Likewise.
4813 (ipa_get_adjustment_candidate): Likewise.
4814 (index_in_adjustments_multiple_times_p): Likewise.
4815 (ipa_combine_adjustments): Likewise.
4816 (ipa_dump_param_adjustments): Likewise.
4817 * tree-sra.c: Also include ipa-param-manipulation.h
4818 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
4821 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4822 Alan Hayward <alan.hayward@arm.com>
4823 David Sherwood <david.sherwood@arm.com>
4825 * doc/sourcebuild.texi (vect_masked_store): Document.
4827 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4828 Alan Hayward <alan.hayward@arm.com>
4829 David Sherwood <david.sherwood@arm.com>
4831 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
4833 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4834 Alan Hayward <alan.hayward@arm.com>
4835 David Sherwood <david.sherwood@arm.com>
4837 * doc/sourcebuild.texi (vect_variable_length): Document.
4839 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4840 Alan Hayward <alan.hayward@arm.com>
4841 David Sherwood <david.sherwood@arm.com>
4843 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
4845 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4846 Alan Hayward <alan.hayward@arm.com>
4847 David Sherwood <david.sherwood@arm.com>
4849 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
4851 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4852 Alan Hayward <alan.hayward@arm.com>
4853 David Sherwood <david.sherwood@arm.com>
4855 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
4856 previously undocumented selectors.
4857 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
4859 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4861 * doc/rtl.texi (const_vector): Say that elements can be
4862 const_wide_ints too.
4863 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
4864 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
4865 (gen_vec_duplicate): Use it instead of CONSTANT_P.
4866 * optabs.c (expand_vector_broadcast): Likewise.
4868 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4869 Alan Hayward <alan.hayward@arm.com>
4870 David Sherwood <david.sherwood@arm.com>
4872 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
4873 scaled index even if the unscaled address was invalid.
4874 Don't increase the complexity of using a scale in that case.
4876 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4877 Alan Hayward <alan.hayward@arm.com>
4878 David Sherwood <david.sherwood@arm.com>
4880 * doc/rtl.texi: Rewrite the subreg rules so that they partition
4881 the inner register into REGMODE_NATURAL_SIZE bytes rather than
4882 UNITS_PER_WORD bytes.
4883 * emit-rtl.c (validate_subreg): Divide subregs into blocks
4884 based on REGMODE_NATURAL_SIZE of the inner mode.
4885 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
4886 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
4887 * expmed.c (lowpart_bit_field_p): Divide the value up into
4888 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
4889 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
4890 whether something is likely to occupy more than one register.
4892 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4895 * ipa-inline-transform.c (update_noncloned_frequencies): Use
4896 profile_count::adjust_for_ipa_scaling.
4897 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
4898 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
4900 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
4902 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4904 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
4905 (split_group): Add total_orig and total_new arguments, estimate the
4906 number of statements related to the store group without store merging
4907 and with store merging.
4908 (imm_store_chain_info::output_merged_store): Adjust split_group
4909 callers, punt if estimated number of statements with store merging
4910 is not smaller than estimated number of statements without it.
4912 (handled_load): Remove has_single_use checks.
4913 (pass_store_merging::process_store): Likewise.
4915 2017-11-09 Richard Biener <rguenther@suse.de>
4917 PR tree-optimization/82902
4918 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
4920 2017-11-09 Martin Liska <mliska@suse.cz>
4923 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
4926 2017-11-09 Martin Liska <mliska@suse.cz>
4928 PR tree-optimization/82669
4929 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
4931 2017-11-09 Martin Liska <mliska@suse.cz>
4933 PR gcov-profile/48463
4934 * coverage.c (coverage_begin_function): Output also end locus
4935 of a function and information whether the function is
4937 * gcov-dump.c (tag_function): Parse and print the information.
4938 * gcov.c (INCLUDE_MAP): Add include.
4939 (INCLUDE_SET): Likewise.
4940 (struct line_info): Move earlier in the source file because
4941 of vector<line_info> in function_info structure.
4942 (line_info::line_info): Likewise.
4943 (line_info::has_block): Likewise.
4944 (struct source_info): Add new member index.
4945 (source_info::get_functions_at_location): New function.
4946 (function_info::group_line_p): New function.
4947 (output_intermediate_line): New function.
4948 (output_intermediate_file): Use the mentioned function.
4949 (struct function_start): New.
4950 (struct function_start_pair_hash): Likewise.
4951 (process_file): Add code that identifies group functions.
4952 Assign lines either to global or function scope.
4953 (generate_results): Skip artificial functions.
4954 (find_source): Assign index for each source file.
4955 (read_graph_file): Read new flag artificial and end_line.
4956 (add_line_counts): Assign it either to global of function scope.
4957 (accumulate_line_counts): Isolate core of the function to
4958 accumulate_line_info and call it for both function and global
4960 (accumulate_line_info): New function.
4961 (output_line_beginning): Fix GNU coding style.
4962 (print_source_line): New function.
4963 (output_line_details): Likewise.
4964 (output_function_details): Likewise.
4965 (output_lines): Iterate both source (global) scope and function
4967 (struct function_line_start_cmp): New class.
4968 * doc/gcov.texi: Reflect changes in documentation.
4970 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4973 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
4974 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
4975 and similarly for not instead of neg.
4977 2017-11-08 Andi Kleen <ak@linux.intel.com>
4979 * config/i386/i386.opt: Add -mforce-indirect-call.
4980 * config/i386/predicates.md: Check for flag_force_indirect_call.
4981 * doc/invoke.texi: Document -mforce-indirect-call
4983 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
4985 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
4987 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
4988 config/riscv/riscv.c (predict.h): New include.
4989 (riscv_slow_unaligned_access_p): No longer static.
4990 (riscv_block_move_straight): Add require.
4991 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
4993 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4996 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
4997 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
4998 nonimmediate_operand predicate for operand 1 instead of
5001 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5003 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
5005 * config/aarch64/constraints.md (Uml): New constraint.
5006 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
5009 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5011 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
5012 of two vec_duplicates into a vec_concat.
5014 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5016 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
5017 Simplify vec_merge of vec_duplicate and vec_concat.
5018 * config/aarch64/constraints.md (Utq): New constraint.
5019 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
5022 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5024 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
5025 Simplify vec_merge of vec_duplicate and const_vector.
5026 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
5028 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
5029 mode iterator. Update predicate on operand 1 to
5030 handle non-const_vec constants. Delete constraints.
5031 (*aarch64_combinez_be<mode>): Likewise for operand 2.
5033 2017-11-08 Jakub Jelinek <jakub@redhat.com>
5035 PR tree-optimization/78821
5036 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
5038 (store_operand_info::store_operand_info): Initialize it to false.
5039 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
5040 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
5041 store in the group, and if chain_info is non-NULL, to ignore altogether
5043 (compatible_load_p): Fail if bit_not_p does not match.
5044 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
5045 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
5046 (pass_store_merging::process_store): Adjust
5047 terminate_all_aliasing_chains calls to pass NULL in all current spots,
5048 call terminate_all_aliasing_chains newly when adding a store into
5049 a chain with non-NULL chain_info.
5051 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
5053 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
5055 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
5057 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
5059 (aarch64_layout_frame): Initialise emit_frame_chain.
5060 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
5061 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
5063 2017-11-08 Martin Liska <mliska@suse.cz>
5065 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
5066 of gimple_call_internal_p.
5068 2017-11-07 Tom de Vries <tom@codesourcery.com>
5070 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
5072 2017-11-07 Tom de Vries <tom@codesourcery.com>
5074 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5077 2017-11-08 Martin Liska <mliska@suse.cz>
5080 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
5082 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
5084 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
5086 2017-11-07 Jakub Jelinek <jakub@redhat.com>
5089 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
5090 (*cmp<mode>_ccz_1): New insn with $k alternative.
5093 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
5094 RTX_COMM_COMPARE as commutative as well.
5095 (ix86_binary_operator_ok): Formatting fix.
5096 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
5097 *<code><mode>3<mask_name><round_saeonly_name>,
5098 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
5099 *<s>mul<mode>3_highpart<mask_name>,
5100 *vec_widen_umult_even_v16si<mask_name>,
5101 *vec_widen_umult_even_v8si<mask_name>,
5102 *vec_widen_umult_even_v4si<mask_name>,
5103 *vec_widen_smult_even_v16si<mask_name>,
5104 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
5105 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
5106 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
5107 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
5108 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
5109 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
5110 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
5111 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
5112 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
5113 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
5114 ix86_binary_operator_ok. Formatting fixes.
5115 (*<plusminus_insn><mode>3<mask_name><round_name>,
5116 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
5119 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
5121 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
5122 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
5123 eq and ne if TARGET_ISEL.
5124 (cmp): New code_iterator.
5125 (UNS, UNSU_, UNSIK): New code_attrs.
5126 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
5127 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
5129 ("eq<mode>3"): ... this.
5130 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
5132 ("ne<mode>3"): ... this.
5134 2017-11-07 Julia Koval <julia.koval@intel.com>
5137 * common/config/i386/i386-common.c
5138 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
5139 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
5140 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
5141 * config/i386/i386.opt: Ditto.
5142 * config/i386/i386.c (ix86_target_string): Ditto.
5143 (ix86_option_override_internal): Ditto.
5144 (ix86_init_mpx_builtins): Move MPX to args2.
5145 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
5146 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
5147 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
5148 __builtin_ia32_vgf2p8affineinvqb_v32qi,
5149 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
5150 __builtin_ia32_vgf2p8affineinvqb_v16qi,
5151 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
5153 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
5156 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
5157 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
5158 (zero-extendsidi peephole2): Remove peephole.
5160 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
5163 * stor-layout.c: Include attribs.h.
5164 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
5165 explicit lookup of "aligned" attribute.
5167 2017-11-07 Andrew Waterman <andrew@sifive.com>
5169 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
5170 (riscv_expand_block_move): Likewise.
5171 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
5173 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
5174 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
5175 * config/riscv/riscv.c (riscv_block_move_straight): New function.
5176 (riscv_adjust_block_mem): Likewise.
5177 (riscv_block_move_loop): Likewise.
5178 (riscv_expand_block_move): Likewise.
5179 * config/riscv/riscv.md (movmemsi): New pattern.
5181 2017-11-07 Michael Clark <michaeljclark@mac.com>
5183 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
5184 (MUSL_DYNAMIC_LINKER): Likewise.
5186 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
5188 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
5189 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
5192 2017-11-07 Richard Biener <rguenther@suse.de>
5194 * match.pd: Fix build.
5196 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
5197 Jackson Woodruff <jackson.woodruff@arm.com>
5199 PR tree-optimization/71026
5200 * match.pd: Canonicalize negate in division.
5202 2017-11-07 Sudakshina Das <sudi.das@arm.com>
5205 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
5207 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5209 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
5210 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
5212 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5214 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
5215 non-scalar integral types.
5216 * match.pd (negate_expr_p): Handle MINUS_EXPR.
5217 (-(A-B), -(~A)): New transformations.
5219 2017-11-07 Tom de Vries <tom@codesourcery.com>
5221 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
5222 semicolon after "do {} while (0)".
5223 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5224 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5225 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5226 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5227 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5229 2017-11-07 Tom de Vries <tom@codesourcery.com>
5231 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
5232 after "do {} while (0)".
5233 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5234 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5235 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5236 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5237 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5239 2017-11-07 Tom de Vries <tom@codesourcery.com>
5242 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
5244 (arm_rtx_costs_internal): Add missing semicolon after
5245 HANDLE_NARROW_SHIFT_ARITH call.
5247 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5249 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
5250 disable isel if it was not set explicitly.
5252 2017-11-06 James Bowman <james.bowman@ftdichip.com>
5254 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
5255 (add_type_attribute) likewise.
5257 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
5259 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
5260 of struct ix86_frame.
5261 (ix86_initial_elimination_offset): Likewise.
5262 (ix86_expand_split_stack_prologue): Likewise.
5264 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
5266 * tree-vrp.h (enum value_range_type): Update stale comment.
5268 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5269 Alan Hayward <alan.hayward@arm.com>
5270 David Sherwood <david.sherwood@arm.com>
5272 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
5273 (aarch64_expand_vec_perm_const): Take the number of units too.
5274 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
5275 (aarch64_expand_vec_perm_const): Likewise.
5276 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
5277 (vec_perm<mode>): Update accordingly.
5279 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5280 Alan Hayward <alan.hayward@arm.com>
5281 David Sherwood <david.sherwood@arm.com>
5283 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
5284 Take the number of units too.
5285 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
5286 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
5287 but check for a vector mode before rather than after the call.
5288 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
5289 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
5290 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
5291 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
5292 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
5293 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
5294 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
5295 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
5296 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
5297 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
5298 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
5299 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
5300 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
5301 (aarch64_sqdmull2_n<mode>): Update accordingly.
5303 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5304 Alan Hayward <alan.hayward@arm.com>
5305 David Sherwood <david.sherwood@arm.com>
5307 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
5308 the number of units too.
5309 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
5310 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
5311 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
5312 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
5313 (vec_store_lanesxi<mode>): Update accordingly.
5315 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5316 Alan Hayward <alan.hayward@arm.com>
5317 David Sherwood <david.sherwood@arm.com>
5319 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
5320 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
5321 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
5322 of units rather than the mode.
5323 * config/aarch64/iterators.md (nunits): New mode attribute.
5324 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
5325 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
5326 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
5327 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
5328 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
5329 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
5330 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
5331 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
5332 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
5333 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
5334 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5335 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
5336 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
5337 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
5338 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
5339 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
5340 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
5341 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
5342 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
5343 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
5344 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
5345 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
5346 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
5347 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
5348 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
5349 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
5350 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
5351 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
5352 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
5353 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
5354 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
5355 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
5356 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
5357 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
5358 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
5359 (aarch64_simd_vec_setv2di): Likewise.
5361 2017-11-06 Carl Love <cel@us.ibm.com>
5363 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
5365 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
5366 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
5368 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
5370 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
5371 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
5372 Add power 8 macro expansions.
5373 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
5374 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
5375 power 8 instructions. (VSX_XXBR): Add iterator.
5377 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
5379 * config/arm/arm.md (predicable_short_it): Change default to "no",
5380 improve documentation, remove uses that are identical to the default.
5381 (enabled_for_depr_it): Rename to enabled_for_short_it.
5382 * gcc/config/arm/arm-fixed.md (predicable_short_it):
5383 Remove default uses.
5384 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
5385 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
5386 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
5387 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
5389 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
5392 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
5393 float128 helper macros, which are no longer used after deleting
5394 the old 'q' built-in functions, and moving the round to odd
5395 built-in functions to being special built-in functions.
5396 (BU_FLOAT128_2): Likewise.
5397 (BU_FLOAT128_1_HW): Likewise.
5398 (BU_FLOAT128_2_HW): Likewise.
5399 (BU_FLOAT128_3_HW): Likewise.
5400 (FABSQ): Delete old 'q' built-in functions.
5401 (COPYSIGNQ): Likewise.
5402 (SQRTF128_ODD): Move round to odd built-in functions to be
5403 special built-in functions, so that we can handle
5404 -mabi=ieeelongdouble.
5405 (TRUNCF128_ODD): Likewise.
5406 (ADDF128_ODD): Likewise.
5407 (SUBF128_ODD): Likewise.
5408 (MULF128_ODD): Likewise.
5409 (DIVF128_ODD): Likewise.
5410 (FMAF128_ODD): Likewise.
5411 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
5412 built-in names to 'f128'.
5413 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
5414 old 'q' built-in functions, as the machine independent code for
5415 'f128' built-in functions handles this.
5416 (rs6000_expand_builtin): Add expansion for float128 round to odd
5417 functions, keying off on -mabi=ieeelongdouble of whether to use
5418 the KFmode or TFmode variant.
5419 (rs6000_init_builtins): Initialize the _Float128 round to odd
5421 * doc/extend.texi (PowerPC Built-in Functions): Document the old
5422 _Float128 'q' built-in functions are now mapped into the new
5423 'f128' built-in functions.
5425 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
5427 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
5428 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
5430 2017-11-06 David Malcolm <dmalcolm@redhat.com>
5433 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
5434 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
5435 * toplev.c: Include "ipa-fnsummary.h".
5436 (toplev::finalize): Call ipa_fnsummary_c_finalize.
5438 2017-11-06 Jakub Jelinek <jakub@redhat.com>
5440 PR tree-optimization/82838
5441 * gimple-ssa-store-merging.c
5442 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
5443 on a separate gimple_seq which is then appended to seq.
5445 2017-11-06 Jeff Law <law@redhat.com>
5448 * config/i386/i386.c (PROBE_INTERVAL): Remove.
5449 (get_probe_interval): New functions.
5450 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
5451 (ix86_adjust_stack_and_probe): Likewise.
5452 (output_adjust_stack_and_probe): Likewise.
5453 (ix86_emit_probe_stack_range): Likewise.
5454 (ix86_expand_prologue): Likewise.
5456 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5458 PR tree-optimization/82816
5459 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
5460 if the modes of the two types are the same.
5461 (convert_plusminus_to_widen): Likewise.
5463 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5465 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
5467 (usadv16qi): New define_expand.
5468 (usadv8hi): New define_expand.
5470 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
5473 * ipa-inline-transform.c (update_noncloned_frequencies): Always
5475 (inline_transform): Likewise.
5476 * predict.c (counts_to_freqs): Remove useless conditional.
5477 * profile-count.h (profile_count::apply_scale): Move sanity check.
5478 * tree-inline.c (copy_bb): Always scale.
5479 (copy_cfg_body): Likewise.
5481 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
5484 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
5486 (*cmp_ite0): Add enabled_for_depr_it attribute.
5487 (*cmp_ite1): Likewise.
5489 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5491 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
5494 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5496 * tree-vrp.c (vrp_int_const_binop): Return true on success and
5497 return the value by pointer.
5498 (extract_range_from_multiplicative_op_1): Update accordingly.
5499 Return as soon as an operation fails.
5501 2017-11-05 Tom de Vries <tom@codesourcery.com>
5504 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
5505 (DEF_SANITIZER_BUILTIN): ... here.
5506 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
5507 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
5509 2017-11-05 Tom de Vries <tom@codesourcery.com>
5512 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
5514 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
5515 ASM_OUTPUT_BEFORE_CASE_LABEL call.
5516 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
5518 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5519 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5520 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5522 2017-11-05 Tom de Vries <tom@codesourcery.com>
5525 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
5528 2017-11-04 Michael Clark <michaeljclark@mac.com>
5530 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
5531 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
5533 (*addsi3_extended): Likewise.
5534 (*addsi3_extended2): Likewise.
5535 (<optab>si3): Likewise.
5536 (<optab>di3): Likewise.
5537 (<optab><mode>3): Likewise.
5538 (<*optabe>si3_internal): Likewise.
5539 (zero_extendqi<SUPERQI:mode>2): Likewise.
5540 (*add<mode>hi3): Likewise.
5541 (*xor<mode>hi3): Likewise.
5542 (<optab>di3): Likewise.
5543 (*<optab>si3_extend): Likewise.
5544 (*sge<u>_<X:mode><GPR:mode>): Likewise.
5545 (*slt<u>_<X:mode><GPR:mode>): Likewise.
5546 (*sle<u>_<X:mode><GPR:mode>): Likewise.
5548 2017-11-04 Andrew Waterman <andrew@sifive.com>
5550 * config/riscv/riscv.c (riscv_option_override): Conditionally set
5551 TARGET_STRICT_ALIGN based upon -mtune argument.
5553 2017-11-04 Andrew Waterman <andrew@sifive.com>
5555 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
5557 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
5559 * config/i386/i386.c (choose_basereg): Use optional scratch
5560 register and add assertion.
5561 (x86_emit_outlined_ms2sysv_save): Use scratch register when
5562 needed, and don't allocate stack.
5563 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
5564 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
5565 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
5567 2017-11-03 Jeff Law <law@redhat.com>
5569 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
5570 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
5571 to probe at the start of a noreturn function.
5573 2017-11-03 Jakub Jelinek <jakub@redhat.com>
5575 PR tree-optimization/78821
5576 * gimple-ssa-store-merging.c: Update the file comment.
5577 (MAX_STORE_ALIAS_CHECKS): Define.
5578 (struct store_operand_info): New type.
5579 (store_operand_info::store_operand_info): New constructor.
5580 (struct store_immediate_info): Add rhs_code and ops data members.
5581 (store_immediate_info::store_immediate_info): Add rhscode, op0r
5582 and op1r arguments to the ctor, initialize corresponding data members.
5583 (struct merged_store_group): Add load_align_base and load_align
5585 (merged_store_group::merged_store_group): Initialize them.
5586 (merged_store_group::do_merge): Update them.
5587 (merged_store_group::apply_stores): Pick the constant for
5588 encode_tree_to_bitpos from one of the two operands, or skip
5589 encode_tree_to_bitpos if neither operand is a constant.
5590 (class pass_store_merging): Add process_store method decl. Remove
5591 bool argument from terminate_all_aliasing_chains method decl.
5592 (pass_store_merging::terminate_all_aliasing_chains): Remove
5593 var_offset_p argument and corresponding handling.
5594 (stmts_may_clobber_ref_p): New function.
5595 (compatible_load_p): New function.
5596 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
5597 if there is overlap and rhs_code is not INTEGER_CST. For
5598 non-overlapping stores terminate group if rhs is not mergeable.
5599 (get_alias_type_for_stmts): Change first argument from
5600 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
5601 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
5602 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
5604 (get_location_for_stmts): Change first argument from
5605 auto_vec<gimple *> & to vec<gimple *> &.
5606 (struct split_store): Remove orig_stmts data member, add orig_stores.
5607 (split_store::split_store): Create orig_stores rather than orig_stmts.
5608 (find_constituent_stmts): Renamed to ...
5609 (find_constituent_stores): ... this. Change second argument from
5610 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
5611 to info structures rather than the statements.
5612 (split_group): Rename ALLOW_UNALIGNED argument to
5613 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
5614 it. Adjust find_constituent_stores caller.
5615 (imm_store_chain_info::output_merged_store): Handle rhs_code other
5616 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
5617 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
5618 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
5619 (mem_valid_for_store_merging): New function.
5620 (handled_load): New function.
5621 (pass_store_merging::process_store): New method.
5622 (pass_store_merging::execute): Use process_store method. Adjust
5623 terminate_all_aliasing_chains caller.
5625 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5627 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
5628 Return true for more constants, symbols and label references.
5629 (aarch64_valid_floating_const): Remove unused function.
5631 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5634 * config/aarch64/aarch64.c (aarch64_layout_frame):
5635 Undo forcing of LR at bottom of frame.
5637 2017-11-03 Jeff Law <law@redhat.com>
5640 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
5641 for int_registers_saved.
5643 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
5644 extracted from tree-ssa-dom.c.
5645 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
5646 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
5647 (record_equivalences_from_incoming_edge): Add additional argument
5648 to single_pred_edge_ignoring_loop_edges call.
5649 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
5650 (uncprop_dom_walker::before_dom_children): Add additional argument
5651 to single_pred_edge_ignoring_loop_edges call.
5652 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
5653 single_pred_edge_ignoring_loop_edges rather than open coding.
5654 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
5656 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
5658 * match.pd (-(-A)): Rewrite.
5660 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5662 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
5663 (rs6000_emit_int_cmove): New declaration.
5664 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
5665 (rs6000_emit_sISEL): Delete.
5666 (rs6000_emit_int_cmove): Make non-static.
5667 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
5668 instead of rs6000_emit_sISEL.
5670 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
5672 * asan.c (create_cond_insert_point): Maintain profile.
5673 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
5675 * basic-block.h (struct basic_block_def): Remove frequency.
5676 (EDGE_FREQUENCY): Use to_frequency
5677 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
5679 (find_traces): Update to use to_frequency.
5680 (find_traces_1_round): Likewise; use only IPA counts.
5681 (bb_to_key): Likewise.
5682 (connect_traces): Use IPA counts only.
5683 (copy_bb_p): Update to use to_frequency.
5684 (fix_up_crossing_landing_pad): Likewise.
5685 (sanitize_hot_paths): Likewise.
5686 * bt-load.c (basic_block_freq): Likewise.
5687 * cfg.c (init_flow): Set count_max to uninitialized.
5688 (check_bb_profile): Remove frequencies; check counts.
5689 (dump_bb_info): Do not dump frequencies.
5690 (update_bb_profile_for_threading): Update counts only.
5691 (scale_bbs_frequencies_int): Likewise.
5692 (MAX_SAFE_MULTIPLIER): Remove.
5693 (scale_bbs_frequencies_gcov_type): Update counts only.
5694 (scale_bbs_frequencies_profile_count): Update counts only.
5695 (scale_bbs_frequencies): Update counts only.
5696 * cfg.h (struct control_flow_graph): Add count-max.
5697 (update_bb_profile_for_threading): Update prototype.
5698 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
5699 (find_many_sub_basic_blocks): Likewise.
5700 * cfgcleanup.c (try_forward_edges): Likewise.
5701 (try_crossjump_to_edge): Likewise.
5702 * cfgexpand.c (expand_gimple_cond): Likewise.
5703 (expand_gimple_tailcall): Likewise.
5704 (construct_init_block): Likewise.
5705 (construct_exit_block): Likewise.
5706 * cfghooks.c (verify_flow_info): Check consistency of counts.
5707 (dump_bb_for_graph): Do not dump frequencies.
5708 (split_block_1): Do not update frequencies.
5709 (split_edge): Do not update frequencies.
5710 (make_forwarder_block): Do not update frequencies.
5711 (duplicate_block): Do not update frequencies.
5712 (account_profile_record): Do not update frequencies.
5713 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
5714 for global heuristics.
5715 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
5716 (expected_loop_iterations_unbounded): Use counts only.
5717 * cfgloopmanip.c (scale_loop_profile): Simplify.
5718 (create_empty_loop_on_edge): Simplify
5720 (duplicate_loop_to_header_edge): Simplify
5721 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
5722 (update_br_prob_note): Take care of removing note when profile
5724 (relink_block_chain): Do not dump frequency.
5725 (rtl_account_profile_record): Use to_frequency.
5726 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
5727 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
5728 (cgraph_update_edges_for_call_stmt_node): Likewise.
5729 (cgraph_edge::verify_count_and_frequency): Update.
5730 (cgraph_node::verify_node): Temporarily disable frequency verification.
5731 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
5732 to_cgraph_frequency.
5733 (cgraph_edge::rebuild_edges): Convert to ipa counts.
5734 * cgraphunit.c (init_lowered_empty_function): Do not initialize
5736 (cgraph_node::expand_thunk): Update profile.
5737 * except.c (dw2_build_landing_pads): Do not update frequency.
5738 * final.c (compute_alignments): Use to_frequency.
5739 (dump_basic_block_info): Do not dump frequency.
5740 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
5741 (dump_gimple_bb_header): Do not dump frequency.
5742 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
5744 * gimple-streamer-in.c (input_bb): Do not stream frequency.
5745 * gimple-streamer-out.c (output_bb): Do not stream frequency.
5746 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
5747 (init_before_recovery): Do not update frequency.
5748 (sched_create_recovery_edges): Do not update frequency.
5749 * hsa-gen.c (convert_switch_statements): Do not update frequency.
5750 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
5751 (ipa_cp_c_finalize): Set max_count to uninitialized.
5752 * ipa-fnsummary.c (get_minimal_bb): Use counts.
5753 (param_change_prob): Use counts.
5754 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
5756 * ipa-split.c (consider_split): Use to_frequency.
5757 (split_function): Use to_frequency.
5758 * ira-build.c (loop_compare_func): Likewise.
5759 (mark_loops_for_removal): Likewise.
5760 (mark_all_loops_for_removal): Likewise.
5761 * loop-doloop.c (doloop_modify): Do not update frequency.
5762 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
5764 * lto-streamer-in.c (input_function): Update count_max.
5765 * omp-expand.c (expand_omp_taskreg): Update count_max.
5766 * omp-simd-clone.c (simd_clone_adjust): Update profile.
5767 * predict.c (maybe_hot_frequency_p): Use to_frequency.
5768 (maybe_hot_count_p): Use ipa counts only.
5769 (maybe_hot_bb_p): Simplify.
5770 (maybe_hot_edge_p): Simplify.
5771 (probably_never_executed): Do not take frequency argument.
5772 (probably_never_executed_bb_p): Do not pass frequency.
5773 (probably_never_executed_edge_p): Likewise.
5774 (combine_predictions_for_bb): Check that profile is nonzero.
5775 (propagate_freq): Do not set frequency.
5776 (drop_profile): Simplify.
5777 (counts_to_freqs): Simplify.
5778 (expensive_function_p): Use to_frequency.
5779 (propagate_unlikely_bbs_forward): Simplify.
5780 (determine_unlikely_bbs): Simplify.
5781 (estimate_bb_frequencies): Add hack to silence graphite issues.
5782 (compute_function_frequency): Use ipa counts.
5783 (pass_profile::execute): Update.
5784 (rebuild_frequencies): Use counts only.
5785 (force_edge_cold): Use counts only.
5786 * profile-count.c (profile_count::dump): Dump new count types.
5787 (profile_count::differs_from_p): Check compatiblity.
5788 (profile_count::to_frequency): New function.
5789 (profile_count::to_cgraph_frequency): New function.
5790 * profile-count.h (struct function): Declare.
5791 (enum profile_quality): Add profile_guessed_local and
5792 profile_guessed_global0.
5793 (class profile_proability): Decrease number of bits to 29;
5794 update from_reg_br_prob_note and to_reg_br_prob_note.
5795 (class profile_count: Update comment; decrease number of bits
5796 to 61. Check compatibility.
5797 (profile_count::compatible_p): New private member function.
5798 (profile_count::ipa_p): New member function.
5799 (profile_count::operator<): Handle global zero correctly.
5800 (profile_count::operator>): Handle global zero correctly.
5801 (profile_count::operator<=): Handle global zero correctly.
5802 (profile_count::operator>=): Handle global zero correctly.
5803 (profile_count::nonzero_p): New member function.
5804 (profile_count::force_nonzero): New member function.
5805 (profile_count::max): New member function.
5806 (profile_count::apply_scale): Handle IPA scalling.
5807 (profile_count::guessed_local): New member function.
5808 (profile_count::global0): New member function.
5809 (profile_count::ipa): New member function.
5810 (profile_count::to_frequency): Declare.
5811 (profile_count::to_cgraph_frequency): Declare.
5812 * profile.c (OVERLAP_BASE): Delete.
5813 (compute_frequency_overlap): Delete.
5814 (compute_branch_probabilities): Do not use compute_frequency_overlap.
5815 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
5816 * sched-ebb.c (rank): Use counts only.
5817 * shrink-wrap.c (handle_simple_exit): Use counts only.
5818 (try_shrink_wrapping): Use counts only.
5819 (place_prologue_for_one_component): Use counts only.
5820 * tracer.c (find_best_predecessor): Use to_frequency.
5821 (find_trace): Use to_frequency.
5822 (tail_duplicate): Use to_frequency.
5823 * trans-mem.c (expand_transaction): Do not update frequency.
5824 * tree-call-cdce.c: Do not update frequency.
5825 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
5826 (gimple_merge_blocks): Likewise.
5827 (gimple_split_edge): Likewise.
5828 (gimple_duplicate_sese_region): Likewise.
5829 (gimple_duplicate_sese_tail): Likewise.
5830 (move_sese_region_to_fn): Likewise.
5831 (gimple_account_profile_record): Likewise.
5832 (insert_cond_bb): Likewise.
5833 * tree-complex.c (expand_complex_div_wide): Likewise.
5834 * tree-eh.c (lower_resx): Update profile.
5835 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
5837 (initialize_cfun): Do not initialize frequencies
5838 (freqs_to_counts): Delete.
5839 (copy_cfg_body): Ignore count parameter.
5840 (copy_body): Update.
5841 (expand_call_inline): Update count_max.
5842 (optimize_inline_calls): Update count_max.
5843 (tree_function_versioning): Update count_max.
5844 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
5845 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
5847 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
5848 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
5849 (try_peel_loop): Likewise.
5850 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
5852 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
5853 (tree_transform_and_unroll_loop): Do not use frequencies
5854 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
5855 Use reliable prediction only.
5856 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
5857 * tree-ssa-sink.c (select_best_block): Use to_frequency.
5858 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
5859 probability scaling.
5860 * tree-ssa-threadupdate.c (create_block_for_threading): Do
5861 not update frequency
5862 (any_remaining_duplicated_blocks): Likewise.
5863 (update_profile): Likewise.
5864 (estimated_freqs_path): Delete.
5865 (freqs_to_counts_path): Delete.
5866 (clear_counts_path): Delete.
5867 (ssa_fix_duplicate_block_edges): Likewise.
5868 (duplicate_thread_path): Likewise.
5869 * tree-switch-conversion.c (gen_inbound_check): Use counts.
5870 * tree-tailcall.c (decrease_profile): Do not update frequency.
5871 (eliminate_tail_call): Likewise.
5872 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
5873 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
5874 (optimize_mask_stores): Likewise.
5875 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
5876 * ubsan.c (ubsan_expand_null_ifn): Update profile.
5877 (ubsan_expand_ptr_ifn): Update profile.
5878 * value-prof.c (gimple_ic): Simplify.
5879 * value-prof.h (gimple_ic): Update prototype.
5880 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
5881 * ipa-inline.c (compute_uninlined_call_time): Be sure that
5883 (want_inline_self_recursive_call_p): Likewise.
5884 (resolve_noninline_speculation): Only cummulate defined counts.
5885 (inline_small_functions): Use nonzero_p.
5886 (ipa_inline): Do not access freed node.
5888 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5890 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5891 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
5893 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
5895 * config/riscv/riscv.c (riscv_legitimize_move): Handle
5896 non-legitimate address.
5898 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5900 * config/rs6000/rs6000.md (*lt0_disi): Delete.
5901 (*lt0_<mode>di, *lt0_<mode>si): New.
5903 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5905 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
5906 TARGET_PAIRED_FLOAT.
5908 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
5909 Jim Wilson <jim.wilson@linaro.org>
5911 * config/aarch64/aarch64-cores.def (saphira): New CPU.
5912 * config/aarch64/aarch64-tune.md: Regenerated.
5913 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
5914 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
5916 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
5918 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
5919 (arc_expand_prologue): Restore blink for millicode.
5920 * config/arc/linux.h (LINK_EH_SPEC): Defined.
5922 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
5925 * config/i386/i386.c (ix86_vector_duplicate_value): Use
5926 gen_vec_duplicate after forcing the scalar into a register.
5928 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
5930 * combine (try_combine): Print the insns input to try_combine to the
5933 2017-11-02 Steve Ellcey <sellcey@cavium.com>
5936 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
5937 Remove second argument from aarch64_process_target_attr call.
5938 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
5940 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
5942 (aarch64_handle_attr_arch): Remove second argument.
5943 (aarch64_handle_attr_cpu): Ditto.
5944 (aarch64_handle_attr_tune): Ditto.
5945 (aarch64_handle_attr_isa_flags): Ditto.
5946 (aarch64_process_one_target_attr): Ditto.
5947 (aarch64_process_target_attr): Ditto.
5948 (aarch64_option_valid_attribute_p): Remove second argument.
5949 on aarch64_process_target_attr call.
5951 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5953 * diagnostic.c: Include "selftest-diagnostic.h".
5954 (selftest::assert_location_text): New function.
5955 (selftest::test_diagnostic_get_location_text): New function.
5956 (selftest::diagnostic_c_tests): Call it.
5958 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5960 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
5961 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
5962 (class selftest::test_diagnostic_context): Move to...
5963 * selftest-diagnostic.c: New file.
5964 * selftest-diagnostic.h: New file.
5966 2017-11-02 James Bowman <james.bowman@ftdichip.com>
5968 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
5969 offset range for FT32B.
5970 * config/ft32/ft32.h: option "mcompress" enables relaxation.
5971 * config/ft32/ft32.md: Add TARGET_NOPM.
5972 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
5973 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
5975 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
5977 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
5979 2017-11-02 Jeff Law <law@redhat.com>
5981 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
5982 virtual keyword on FINAL OVERRIDE members.
5984 * tree-ssa-propagate.h (ssa_propagation_engine): Group
5985 virtuals together. Add virtual destructor.
5986 (substitute_and_fold_engine): Similarly.
5988 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
5990 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
5992 2017-11-02 Richard Biener <rguenther@suse.de>
5994 PR tree-optimization/82795
5995 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
5997 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5999 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
6001 (gcc_GAS_CHECK_FEATURE): Remove.
6002 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
6003 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
6004 * configure: Regenerate.
6006 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
6008 * config/arc/arc.c (hwloop_optimize): Account for empty
6011 2017-11-02 Richard Biener <rguenther@suse.de>
6014 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
6015 Truncate ARRAY_REF index and element size.
6017 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
6019 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
6021 2017-11-01 Jeff Law <law@redhat.com>
6023 * tree-ssa-ccp.c (ccp_folder): New class derived from
6024 substitute_and_fold_engine.
6025 (ccp_folder::get_value): New member function.
6026 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
6027 (ccp_fold_stmt): Remove prototype.
6028 (ccp_finalize): Call substitute_and_fold from the ccp_class.
6029 * tree-ssa-copy.c (copy_folder): New class derived from
6030 substitute_and_fold_engine.
6031 (copy_folder::get_value): Renamed from get_value.
6032 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
6033 * tree-vrp.c (vrp_folder): New class derived from
6034 substitute_and_fold_engine.
6035 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
6036 (vrp_folder::get_value): New member function.
6037 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
6038 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
6039 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
6040 provide a class interface to folder/substitute routines.
6041 (ssa_prop_fold_stmt_fn): Remove typedef.
6042 (ssa_prop_get_value_fn): Likewise.
6043 (subsitute_and_fold): Remove prototype.
6044 (replace_uses_in): Likewise.
6045 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
6046 Renamed from replace_uses_in. Call the virtual member function
6047 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
6048 (substitute_and_fold_dom_walker): Remove initialization of
6049 data member entries for calbacks. Add substitute_and_fold_engine
6050 member and initialize it.
6051 (substitute_and_fold_dom_walker::before_dom_children0: Use the
6052 member functions for get_value, replace_phi_args_in c
6053 replace_uses_in, and fold_stmt calls.
6054 (substitute_and_fold_engine::substitute_and_fold): Renamed from
6055 substitute_and_fold. Remove assert. Update ctor call.
6057 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
6058 (ssa_prop_visit_phi_fn): Likewise.
6059 (class ssa_propagation_engine): New class to provide an interface
6061 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
6063 (ssa_prop_visit_phi): Likewise.
6064 (ssa_propagation_engine::simulate_stmt): Moved into class.
6065 Call visit_phi/visit_stmt from the class rather than via
6066 file scoped static variables.
6067 (ssa_propagation_engine::simulate_block): Moved into class.
6068 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
6069 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
6070 set file scoped statics for the visit_stmt/visit_phi callbacks.
6071 * tree-complex.c (complex_propagate): New class derived from
6072 ssa_propagation_engine.
6073 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
6074 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
6075 (tree_lower_complex): Call ssa_propagate via the complex_propagate
6077 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
6078 ssa_propagation_engine.
6079 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
6080 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
6081 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
6082 * tree-ssa-copy.c (copy_prop): New class derived from
6083 ssa_propagation_engine.
6084 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
6085 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
6086 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
6087 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
6088 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
6089 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
6090 (execute_vrp): Call ssa_propagate from the vrp_prop class.
6092 2017-11-01 Jakub Jelinek <jakub@redhat.com>
6094 PR rtl-optimization/82778
6095 PR rtl-optimization/82597
6096 * compare-elim.c (struct comparison): Add in_a_setter field.
6097 (find_comparison_dom_walker::before_dom_children): Remove killed
6098 bitmap and df_simulate_find_defs call, instead walk the defs.
6099 Compute last_setter and initialize in_a_setter. Merge definitions
6100 with first initialization for a few variables.
6101 (try_validate_parallel): Use insn_invalid_p instead of
6102 recog_memoized. Return insn rather than just the pattern.
6103 (try_merge_compare): Fix up comment. Don't uselessly test if
6104 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
6106 (execute_compare_elim_after_reload): Remove df_chain_add_problem
6109 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6110 Alan Hayward <alan.hayward@arm.com>
6111 David Sherwood <david.sherwood@arm.com>
6113 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
6114 aarch64_hard_regno_nregs to get the number of registers
6117 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6118 Alan Hayward <alan.hayward@arm.com>
6119 David Sherwood <david.sherwood@arm.com>
6121 * config/aarch64/constraints.md (Upl): Rename to...
6123 * config/aarch64/aarch64.md
6124 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
6127 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6128 Alan Hayward <alan.hayward@arm.com>
6129 David Sherwood <david.sherwood@arm.com>
6131 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
6132 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
6135 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6136 Alan Hayward <alan.hayward@arm.com>
6137 David Sherwood <david.sherwood@arm.com>
6139 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
6140 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
6141 (aarch64_evpc_dup): Generate rtl direcly, rather than using
6143 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
6144 of a single element.
6145 * config/aarch64/iterators.md: Add a comment above the permute
6146 unspecs to say that they are generated directly by
6147 aarch64_expand_vec_perm_const.
6148 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
6150 2017-11-01 Nathan Sidwell <nathan@acm.org>
6152 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
6154 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
6156 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
6157 and medany code models, and describe what they do.
6159 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6161 Revert accidental duplicate:
6163 * combine.c (can_change_dest_mode): Reject changes in
6164 REGMODE_NATURAL_SIZE.
6166 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
6168 PR rtl-optimization/64682
6169 PR rtl-optimization/69567
6170 PR rtl-optimization/69737
6171 PR rtl-optimization/82683
6172 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
6173 register mentioned in the note, drop the note, unless it came from I3,
6174 in which case it should go to I3 again.
6176 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6178 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
6179 and return false if not.
6180 (clear_bytes_written_by, live_bytes_read): Update accordingly.
6182 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6184 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
6185 range is known to be empty.
6187 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6188 Alan Hayward <alan.hayward@arm.com>
6189 David Sherwood <david.sherwood@arm.com>
6191 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
6192 and CONST_VECTOR_NUNITS instead of computing the number of units from
6193 the byte sizes of the vector and element.
6194 (simplify_binary_operation_1): Likewise.
6195 (simplify_const_binary_operation): Likewise.
6196 (simplify_ternary_operation): Likewise.
6198 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6199 Alan Hayward <alan.hayward@arm.com>
6200 David Sherwood <david.sherwood@arm.com>
6202 * var-tracking.c (INT_MEM_OFFSET): Replace with...
6203 (int_mem_offset): ...this new function.
6204 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
6205 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
6206 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
6209 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6210 Alan Hayward <alan.hayward@arm.com>
6211 David Sherwood <david.sherwood@arm.com>
6213 * lower-subreg.c (interesting_mode_p): New function.
6214 (compute_costs, find_decomposable_subregs, decompose_register)
6215 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
6216 (resolve_clobber, dump_choices): Use it.
6218 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6219 Alan Hayward <alan.hayward@arm.com>
6220 David Sherwood <david.sherwood@arm.com>
6222 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
6224 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6225 Alan Hayward <alan.hayward@arm.com>
6226 David Sherwood <david.sherwood@arm.com>
6228 * alias.c (find_base_value, find_base_term): Only process integer
6229 truncations. Check the precision rather than the size.
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 * machmode.h (is_narrower_int_mode): New function
6236 * optabs.c (expand_float, expand_fix): Use it.
6237 * dwarf2out.c (rotate_loc_descriptor): Likewise.
6239 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6240 Alan Hayward <alan.hayward@arm.com>
6241 David Sherwood <david.sherwood@arm.com>
6243 * rtl.h (narrower_subreg_mode): New function.
6244 * ira-color.c (update_costs_from_allocno): Use it.
6246 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6247 Alan Hayward <alan.hayward@arm.com>
6248 David Sherwood <david.sherwood@arm.com>
6250 * optabs-query.h (convert_optab_p): New function, split out from...
6251 (convert_optab_handler): ...here.
6252 (widening_optab_handler): Delete.
6253 (find_widening_optab_handler): Remove permit_non_widening parameter.
6254 (find_widening_optab_handler_and_mode): Likewise. Provide an
6255 override that operates on mode class wrappers.
6256 * optabs-query.c (widening_optab_handler): Delete.
6257 (find_widening_optab_handler_and_mode): Remove permit_non_widening
6258 parameter. Assert that the two modes are the same class and that
6259 the "from" mode is narrower than the "to" mode. Use
6260 convert_optab_handler instead of widening_optab_handler.
6261 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
6262 instead of widening_optab_handler.
6263 * expr.c (expand_expr_real_2): Update calls to
6264 find_widening_optab_handler.
6265 * optabs.c (expand_widen_pattern_expr): Likewise.
6266 (expand_binop_directly): Take the insn_code as a parameter.
6267 (expand_binop): Only call find_widening_optab_handler for
6268 conversion optabs; use optab_handler otherwise. Update calls
6269 to find_widening_optab_handler and expand_binop_directly.
6270 Use convert_optab_handler instead of widening_optab_handler.
6271 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
6272 find_widening_optab_handler and use scalar_mode rather than
6274 (convert_plusminus_to_widen): Likewise.
6276 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6277 Alan Hayward <alan.hayward@arm.com>
6278 David Sherwood <david.sherwood@arm.com>
6280 * machmode.h (fixed_size_mode): New class.
6281 * rtl.h (get_pool_mode): Return fixed_size_mode.
6282 * gengtype.c (main): Add fixed_size_mode.
6283 * target.def (get_raw_result_mode): Return a fixed_size_mode.
6284 (get_raw_arg_mode): Likewise.
6285 * doc/tm.texi: Regenerate.
6286 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
6287 * targhooks.c (default_get_reg_raw_mode): Likewise.
6288 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
6289 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
6290 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
6291 (msp430_get_raw_result_mode): Likewise.
6292 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
6293 * dbxout.c (dbxout_parms): Require fixed-size modes.
6294 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
6295 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
6296 * omp-low.c (lower_oacc_reductions): Likewise.
6297 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
6298 (simplify_subreg): Update accordingly.
6299 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
6300 (force_const_mem): Update accordingly. Return NULL_RTX for modes
6301 that aren't fixed-size.
6302 (get_pool_mode): Return a fixed_size_mode.
6303 (output_constant_pool_2): Take a fixed_size_mode.
6305 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6306 Alan Hayward <alan.hayward@arm.com>
6307 David Sherwood <david.sherwood@arm.com>
6309 * doc/rtl.texi (vec_series): Document.
6310 (const): Say that the operand can be a vec_series.
6311 * rtl.def (VEC_SERIES): New rtx code.
6312 * rtl.h (const_vec_series_p_1): Declare.
6313 (const_vec_series_p): New function.
6314 * emit-rtl.h (gen_const_vec_series): Declare.
6315 (gen_vec_series): Likewise.
6316 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
6317 (gen_vec_series): Likewise.
6318 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
6319 * simplify-rtx.c (simplify_unary_operation): Handle negations
6321 (simplify_binary_operation_series): New function.
6322 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
6323 (test_vector_ops_series): New function.
6324 (test_vector_ops): Call it.
6325 * config/powerpcspe/altivec.md (altivec_lvsl): Use
6326 gen_const_vec_series.
6327 (altivec_lvsr): Likewise.
6328 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
6330 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6331 Alan Hayward <alan.hayward@arm.com>
6332 David Sherwood <david.sherwood@arm.com>
6334 * doc/rtl.texi (const): Update description of address constants.
6335 Say that vector constants are allowed too.
6336 * common.md (E, F): Use CONSTANT_P instead of checking for
6338 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
6339 checking for CONST_VECTOR.
6340 * expmed.c (make_tree): Use build_vector_from_val for a CONST
6342 * expr.c (expand_expr_real_2): Check for vector modes instead
6343 of checking for CONST_VECTOR.
6344 * rtl.h (const_vec_p): New function.
6345 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
6346 (unwrap_const_vec_duplicate): Handle them here too.
6348 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6349 David Malcolm <dmalcolm@redhat.com>
6350 Alan Hayward <alan.hayward@arm.com>
6351 David Sherwood <david.sherwood@arm.com>
6353 * rtl.h (vec_duplicate_p): New function.
6354 * selftest-rtl.c (assert_rtx_eq_at): New function.
6355 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
6356 (assert_rtx_eq_at): Declare.
6357 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
6358 * selftest-run-tests.c (selftest::run_tests): Call it.
6359 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
6360 (simplify_unary_operation_1): Recursively handle vector duplicates.
6361 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
6363 (simplify_subreg): Handle subregs of vector duplicates.
6364 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
6365 (selftest::simplify_rtx_c_tests): New functions.
6367 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6368 Alan Hayward <alan.hayward@arm.com>
6369 David Sherwood <david.sherwood@arm.com>
6371 * emit-rtl.h (gen_const_vec_duplicate): Declare.
6372 (gen_vec_duplicate): Likewise.
6373 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
6375 (gen_const_vector): ...here.
6376 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
6377 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
6378 whose elements are all equal.
6379 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
6380 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
6381 (simplify_relational_operation): Likewise.
6382 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
6384 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
6385 (aarch64_expand_vector_init): Likewise.
6386 * config/arm/arm.c (neon_vdup_constant): Likewise.
6387 (neon_expand_vector_init): Likewise.
6388 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
6389 (arm_block_set_unaligned_vect): Likewise.
6390 (arm_block_set_aligned_vect): Likewise.
6391 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
6392 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
6393 (expand_vec_perm_even_odd_pack): Likewise.
6394 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
6395 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
6396 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
6397 gen_const_vec_duplicate.
6398 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
6399 * config/mips/mips.c (mips_gen_const_int_vector): Use
6400 gen_const_vec_duplicate.
6401 (mips_expand_vector_init): Use CONST0_RTX.
6402 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
6403 (define_split): Use gen_const_vec_duplicate.
6404 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
6405 (define_split): Use gen_const_vec_duplicate.
6406 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
6407 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
6408 * config/spu/spu.c (spu_const): Likewise.
6410 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6411 Alan Hayward <alan.hayward@arm.com>
6412 David Sherwood <david.sherwood@arm.com>
6414 * combine.c (can_change_dest_mode): Reject changes in
6415 REGMODE_NATURAL_SIZE.
6417 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
6419 * configure.ac (--enable-libssp): New.
6420 (gcc_cv_libc_provides_ssp): Check for explicit setting before
6421 trying to determine target-specific default. Adjust indentation.
6422 * configure: Regenerated.
6423 * doc/install.texi (Configuration): Expand --disable-libssp
6426 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
6428 config/i386/i386.c (ix86_expand_epilogue): Correct stack
6431 2017-10-31 Martin Jambor <mjambor@suse.cz>
6434 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
6436 2017-10-31 David Malcolm <dmalcolm@redhat.com>
6438 * auto-profile.c (autofdo_source_profile::read): Use
6439 UNKNOWN_LOCATION rather than 0.
6440 * diagnostic-core.h (warning_at_rich_loc): Rename to...
6441 (warning_at): ...this overload.
6442 (warning_at_rich_loc_n): Rename to...
6443 (warning_n): ...this overload.
6444 (error_at_rich_loc): Rename to...
6445 (error_at): ...this overload.
6446 (pedwarn_at_rich_loc): Rename to...
6447 (pedwarn): ...this overload.
6448 (permerror_at_rich_loc): Rename to...
6449 (permerror): ...this overload.
6450 (inform_at_rich_loc): Rename to...
6451 (inform): ...this overload.
6452 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
6453 (diagnostic_n_impl_richloc): Rename to...
6454 (diagnostic_n_impl): ...this rich_location *-based decl.
6455 (inform_at_rich_loc): Rename to...
6456 (inform): ...this, and add an assertion.
6457 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
6458 (warning_at_rich_loc): Rename to...
6459 (warning_at): ...this, and add an assertion.
6460 (warning_at_rich_loc_n): Rename to...
6461 (warning_n): ...this, and add an assertion.
6462 (warning_n): Update location_t-based implementation for removal of
6463 location_t-based diagnostic_n_impl.
6464 (pedwarn_at_rich_loc): Rename to...
6465 (pedwarn): ...this, and add an assertion.
6466 (permerror_at_rich_loc): Rename to...
6467 (permerror): ...this, and add an assertion.
6468 (error_n): Update for removal of location_t-based diagnostic_n_impl.
6469 (error_at_rich_loc): Rename to...
6470 (error_at): ...this, and add an assertion.
6471 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
6472 (driver::do_spec_on_infiles): Likewise.
6473 * substring-locations.c (format_warning_va): Update for renaming
6474 of inform_at_rich_loc.
6476 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
6478 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
6479 _Float<N>X built-in functions so that the variant without the
6480 "__builtin_" prefix is only enabled for the GNU C and Objective C
6481 languages when they are in non-strict ANSI/ISO mode.
6482 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
6483 * target.def (floatn_builtin_p): Add a target hook to control
6484 whether _Float<N> and _Float<N>X built-in functions without the
6485 "__builtin_" prefix are enabled, and return true for C and
6486 Objective C in the default hook. Include langhooks.h in
6488 * targhooks.h (default_floatn_builtin_p): Likewise.
6489 * targhooks.c (default_floatn_builtin_p): Likewise.
6490 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
6491 floatn_builtin_p target hook.
6492 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
6494 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
6495 Eric Botcazou <ebotcazou@adacore.com>
6497 PR rtl-optimization/81803
6498 * lra-constraints.c (curr_insn_transform): Also reload the whole
6499 register for a strict subreg no wider than a word if this is for
6500 a WORD_REGISTER_OPERATIONS target.
6502 2017-10-31 Jason Merrill <jason@redhat.com>
6504 * gdbinit.in: Skip over inlines from timevar.h.
6506 2017-10-31 Martin Liska <mliska@suse.cz>
6508 * doc/gcov.texi: Document new option.
6509 * gcov.c (print_usage): Likewise print it.
6510 (process_args): Support the argument.
6511 (format_count): New function.
6512 (format_gcov): Use the function.
6514 2017-10-31 Martin Liska <mliska@suse.cz>
6516 * gcov.c (struct name_map): do not use typedef.
6517 Define operator== and operator<.
6518 (name_search): Remove.
6519 (name_sort): Remove.
6520 (main): Do not allocate names.
6521 (process_file): Add vertical space.
6522 (generate_results): Use std::find.
6523 (release_structures): Do not release memory.
6524 (find_source): Use std::find.
6526 2017-10-31 Martin Liska <mliska@suse.cz>
6528 * gcov.c (struct line_info): Remove it's typedef.
6529 (line_info::line_info): Add proper ctor.
6530 (line_info::has_block): Do not use a typedef.
6531 (struct source_info): Do not use typedef.
6532 (circuit): Likewise.
6533 (get_cycles_count): Likewise.
6534 (output_intermediate_file): Iterate via vector iterator.
6535 (add_line_counts): Use std::vector methods.
6536 (accumulate_line_counts): Likewise.
6537 (output_lines): Likewise.
6539 2017-10-31 Martin Liska <mliska@suse.cz>
6541 * gcov.c (struct source_info): Remove typedef.
6542 (source_info::source_info): Add proper ctor.
6543 (accumulate_line_counts): Use struct, not it's typedef.
6544 (output_gcov_file): Likewise.
6545 (output_lines): Likewise.
6546 (main): Do not allocate an array.
6547 (output_intermediate_file): Use size of vector container.
6548 (process_file): Resize the vector.
6549 (generate_results): Do not preallocate, use newly added vector
6551 (release_structures): Do not release sources.
6552 (find_source): Use vector methods.
6553 (add_line_counts): Do not use typedef.
6555 2017-10-31 Martin Liska <mliska@suse.cz>
6557 * doc/gcov.texi: Document that.
6558 * gcov.c (add_line_counts): Mark lines with a non-executed
6560 (output_line_beginning): Handle such lines.
6561 (output_lines): Pass new argument.
6562 (output_intermediate_file): Print it in intermediate format.
6564 2017-10-31 Martin Liska <mliska@suse.cz>
6566 * color-macros.h: New file.
6567 * diagnostic-color.c: Factor out color related to macros to
6569 * doc/gcov.texi: Document -k option.
6570 * gcov.c (INCLUDE_STRING): Include string.h.
6571 (print_usage): Add -k option.
6572 (process_args): Parse it.
6573 (pad_count_string): New function.
6574 (output_line_beginning): Likewise.
6575 (DEFAULT_LINE_START): New macro.
6576 (output_lines): Support color output.
6578 2017-10-31 Martin Liska <mliska@suse.cz>
6580 PR gcov-profile/82633
6581 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
6582 their interaction with GCOV infrastructure.
6583 * configure.ac: Add -fkeep-{inline,static}-functions to
6585 * configure: Regenerate.
6587 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
6590 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
6592 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
6595 * config/rs6000/rs6000.md (allocate_stack): Force update interval
6596 into a register if it does not fit into an immediate offset field.
6598 2017-10-31 Olivier Hainque <hainque@adacore.com>
6600 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
6602 2017-10-31 Julia Koval <julia.koval@intel.com>
6604 * config.gcc: Add gfniintrin.h.
6605 * config/i386/gfniintrin.h: New.
6606 * config/i386/i386-builtin-types.def
6607 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
6608 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
6609 __builtin_ia32_vgf2p8affineinvqb_v32qi,
6610 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
6611 __builtin_ia32_vgf2p8affineinvqb_v16qi,
6612 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
6613 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
6614 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
6615 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
6616 V64QI_FTYPE_V64QI_V64QI_INT): New types.
6617 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
6618 * config/i386/immintrin.h: Include gfniintrin.h.
6619 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
6621 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
6623 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
6625 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
6627 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
6628 (arm_ashldi3_1bit): Remove pattern.
6629 (ashrdi3): Remove shift by 1 expansion.
6630 (arm_ashrdi3_1bit): Remove pattern.
6631 (lshrdi3): Remove shift by 1 expansion.
6632 (arm_lshrdi3_1bit): Remove pattern.
6633 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
6634 cost of ashldi3 by 1.
6635 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
6636 (<shift>di3_neon): Likewise.
6638 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6640 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
6641 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
6642 and (*aarch64_simd_mov<VQ:mode>).
6643 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
6644 pattern alternative.
6645 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
6646 attributes to match pattern alternative.
6648 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
6650 * config.gcc (powerpc*-*-*): Add emmintrin.h.
6651 * config/rs6000/emmintrin.h: New file.
6652 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
6654 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
6656 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
6657 * (movdi_vfp_cortexa8): Remove pattern.
6659 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6661 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
6663 (Specific, alpha*-dec-osf5.1): Remove.
6664 (Specific, mips-sgi-irix5): Remove.
6665 (Specific, mips-sgi-irix6): Remove.
6667 2017-10-30 Jakub Jelinek <jakub@redhat.com>
6670 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
6671 arguments to clear_bit_region_be.
6673 2017-10-30 Jim Wilson <wilson@tuliptree.org>
6675 * gimplify.c: Include memmodel.h.
6677 2017-10-30 Martin Jambor <mjambor@suse.cz>
6679 * omp-grid.c (grid_attempt_target_gridification): Also insert a
6680 condition whether loop should be executed at all.
6682 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
6684 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6685 gimple folding of vec_madd() intrinsics.
6686 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
6687 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
6688 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
6690 2017-10-30 Richard Biener <rguenther@suse.de>
6692 PR tree-optimization/82762
6694 2017-10-23 Richard Biener <rguenther@suse.de>
6696 PR tree-optimization/82129
6698 2017-08-01 Richard Biener <rguenther@suse.de>
6700 PR tree-optimization/81181
6701 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6702 (compute_antic): ... end of iteration here.
6704 2017-10-30 Joseph Myers <joseph@codesourcery.com>
6706 * doc/invoke.texi (C Dialect Options): Document -std=c17,
6707 -std=iso9899:2017 and -std=gnu17.
6708 * doc/standards.texi (C Language): Document C17 support.
6709 * doc/cpp.texi (Overview): Mention -std=c17.
6710 (Standard Predefined Macros): Document C11 and C17 values of
6711 __STDC_VERSION__. Do not refer to C99 support as incomplete.
6712 * doc/extend.texi (Inline): Do not list individual options for
6713 standards newer than C99.
6714 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
6716 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
6719 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
6721 * asan.c (asan_finish_file): Align asan globals array by shadow
6724 2017-10-30 Jakub Jelinek <jakub@redhat.com>
6727 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
6728 (struct store_immediate_info): Add bitregion_start and bitregion_end
6730 (store_immediate_info::store_immediate_info): Add brs and bre
6731 arguments and initialize bitregion_{start,end} from those.
6732 (struct merged_store_group): Add bitregion_start, bitregion_end,
6733 align_base and mask fields. Drop unnecessary struct keyword from
6734 struct store_immediate_info. Add do_merge method.
6735 (clear_bit_region_be): Use memset instead of loop storing zeros.
6736 (merged_store_group::do_merge): New method.
6737 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
6738 stores as long as the surrounding bitregions have no gaps.
6739 (merged_store_group::merge_overlapping): Use do_merge.
6740 (merged_store_group::apply_stores): Test that bitregion_{start,end}
6741 is byte aligned, rather than requiring that start and width are
6742 byte aligned. Drop unnecessary struct keyword from
6743 struct store_immediate_info. Allocate and populate also mask array.
6744 Make start of the arrays relative to bitregion_start rather than
6745 start and size them according to bitregion_{end,start} difference.
6746 (struct imm_store_chain_info): Drop unnecessary struct keyword from
6747 struct store_immediate_info.
6748 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
6749 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
6750 struct keyword from struct store_immediate_info.
6751 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
6752 between stores as long as the surrounding bitregions have no gaps.
6754 (struct split_store): Add orig non-static data member.
6755 (split_store::split_store): Initialize orig to false.
6756 (find_constituent_stmts): Return store_immediate_info *, non-NULL
6757 if there is exactly a single original stmt. Change stmts argument
6758 to pointer from reference, if NULL, don't push anything to it. Add
6759 first argument, use it to optimize skipping over orig stmts that
6760 are known to be before bitpos already. Simplify.
6761 (split_group): Return unsigned int count how many stores are or
6762 would be needed rather than a bool. Add allow_unaligned argument.
6763 Change split_stores argument from reference to pointer, if NULL,
6764 only do a dry run computing how many stores would be produced.
6765 Rewritten algorithm to use both alignment and misalign if
6766 !allow_unaligned and handle bitfield stores with gaps.
6767 (imm_store_chain_info::output_merged_store): Set start_byte_pos
6768 from bitregion_start instead of start. Compute allow_unaligned
6769 here, if true, do 2 split_group dry runs to compute which one
6770 produces fewer stores and prefer aligned if equal. Punt if
6771 new count is bigger or equal than original before emitting any
6772 statements, rather than during that. Remove no longer needed
6773 new_ssa_names tracking. Replace num_stmts with
6774 split_stores.length (). Use 32-bit stack allocated entries
6775 in split_stores auto_vec. Try to reuse original store lhs/rhs1
6776 if possible. Handle bitfields with gaps.
6777 (pass_store_merging::execute): Ignore bitsize == 0 stores.
6778 Compute bitregion_{start,end} for the stores and construct
6779 store_immediate_info with that. Formatting fixes.
6781 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
6784 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
6785 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
6787 2017-10-29 Jim Wilson <wilson@tuliptree.org>
6789 * gimplify.c: Include tm_p.h.
6791 * common.opt (gcoff): Re-add as ignored option.
6792 (gcoff1, gcoff2, gcoff3): Likewise.
6794 * Makefile.in (OBJS): Delete sdbout.o.
6795 (GTFILES): Delete $(srcdir)/sdbout.c.
6796 * debug.h: Delete sdb_debug_hooks.
6797 * final.c: Delete sdbout.h include.
6798 (final_scan_insn): Delete SDB_DEBUG check.
6799 (rest_of_clean_state): Likewise.
6800 * output.h: Delete sdb_begin_function_line.
6803 * toplev.c: Delete sdbout.h include.
6804 (process_options): Delete SDB_DEBUG check.
6805 * tree-core.h (tree_type_common): Delete pointer field of
6807 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
6808 TYPE_SYMTAB_POINTER.
6809 * tree.h (TYPE_SYMTAB_POINTER): Delete.
6810 (TYPE_SYMTAB_IS_POINTER): Delete.
6811 (TYPE_SYMTAB_IS_DIE): Renumber.
6812 * xcoffout.c: Refer to former sdbout.c file.
6813 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
6815 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
6816 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
6817 (Debugging Options): Delete -gcoff.
6818 (-gstabs): Delete SDB reference.
6820 (-gcoff@var{level}): Delete.
6821 * doc/passes.texi (Debugging information output): Delete SDB and
6822 sdbout.c references.
6823 * doc/tm.texi: Regenerate.
6824 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
6825 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
6827 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
6828 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
6829 (SDB_DEBUGGING_INFO): Delete.
6830 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
6831 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
6832 * target.def (output_source_filename): Delete COFF reference.
6834 * common.opt (gcoff): Delete.
6835 (gxcoff+): Update Negative chain.
6836 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
6838 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
6839 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
6840 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
6842 (expand_function_start): Change sdb reference to past tense.
6843 (expand_function_end): Change sdb reference to past tense.
6844 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
6845 * opts.c (debug_type_names): Delete coff entry.
6846 (common_handle_option): Delete OPT_gcoff case.
6847 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
6849 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
6850 * config/cris/cris.h: Delete SDB reference in comment.
6851 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
6852 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
6853 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
6854 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
6856 (ix86_expand_prologue): Likewise.
6857 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
6858 * config/ia64/ia64.h: Likewise.
6859 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
6860 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
6862 * config/mmix/mmix.h: Likewise.
6863 * config/nds32/nds32.c: Likewise.
6864 * config/stormy/storym16.h: Likewise.
6865 * config/visium/visium.h: Likewise.
6866 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
6868 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
6870 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
6871 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
6872 FRAME_POINTER_REGNUM point at high end of local var area.
6874 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
6876 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
6877 Move comment around. Do not reset best_edge for a copiable
6878 destination if the copy would cause a partition change.
6879 (better_edge_p): Remove redundant check.
6881 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6883 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
6885 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
6887 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
6888 for math functions that have _Float<N> and _Float<N>X variants.
6889 (mathfn_built_in_2): Add support for math functions that have
6890 _Float<N> and _Float<N>X variants.
6891 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
6892 (expand_builtin_mathfn_ternary): Add support for fma with
6893 _Float<N> and _Float<N>X variants.
6894 (expand_builtin): Likewise.
6895 (fold_builtin_3): Likewise.
6896 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
6897 create math function _Float<N> and _Float<N>X variants as external
6899 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
6900 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
6901 the __builtin_ prefix and if not strict ansi, without the prefix.
6902 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
6903 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
6904 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
6905 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
6906 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
6907 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
6908 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
6909 function signatures for fma _Float<N> and _Float<N>X variants.
6910 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
6911 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
6912 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
6913 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
6914 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
6915 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
6916 * gencfn-macros.c (print_case_cfn): Add support for math functions
6917 that have _Float<N> and _Float<N>X variants.
6918 (print_define_operator_list): Likewise.
6919 (fltfn_suffixes): Likewise.
6921 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
6922 for math functions that have _Float<N> and _Float<N>X variants.
6923 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
6924 and _Float<N>X variants.
6925 (COPYSIGN): Likewise.
6928 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
6929 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
6931 (integer_valued_read_call_p): Likewise.
6932 * fold-const-call.c (fold_const_call_ss): Likewise.
6933 (fold_const_call_sss): Add support for copysign, fmin, and fmax
6934 _Float<N> and _Float<N>X variants.
6935 (fold_const_call_ssss): Add support for fma _Float<N> and
6936 _Float<N>X variants.
6937 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
6938 support for copysign and fma _Float<N> and _Float<N>X variants.
6939 (backprop::process_builtin_call_use): Likewise.
6940 * tree-call-cdce.c (can_test_argument_range); Add support for
6941 sqrt _Float<N> and _Float<N>X variants.
6942 (edom_only_function): Likewise.
6943 (get_no_error_domain): Likewise.
6944 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
6945 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
6946 copysign _Float<N> and _Float<N>X variants.
6947 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
6948 handled by machine independent code.
6949 (FMAF128): Likewise.
6950 * doc/cpp.texi (Common Predefined Macros): Document defining
6951 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
6952 fma _Float<N> and _Float<N>X variants.
6954 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6957 * config/i386/i386-modes.def (CCFPU): Remove definition.
6958 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
6959 (ix86_cc_modes_compatible): Ditto.
6960 (ix86_expand_carry_flag_compare): Ditto.
6961 (ix86_expand_int_movcc): Ditto.
6962 (ix86_expand_int_addcc): Ditto.
6963 (ix86_reverse_condition): Ditto.
6964 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
6965 Return true/false for unordered/ordered fp comparisons.
6966 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
6967 (ix86_prepare_fp_compare_args): Update for rename.
6968 (ix86_expand_fp_compare): Update for rename. Generate unordered
6969 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
6970 (ix86_expand_sse_compare_and_jump): Ditto.
6971 * config/i386/predicates.md (fcmov_comparison_operator):
6972 Remove CCFPU mode handling.
6973 (ix86_comparison_operator): Ditto.
6974 (ix86_carry_flag_operator): Ditto.
6975 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
6976 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
6977 (*cmpu<mode>_cc_i387): Ditto.
6978 (FPCMP): Remove mode iterator.
6979 (unord): Remove mode attribute.
6980 (unord_subst): New define_subst transformation
6981 (unord): New define_subst attribute.
6983 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
6984 (*cmpi<unord>xf_i387): Ditto.
6985 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
6986 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
6987 using unord_subst transformation.
6988 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
6989 (round_saeonly): Also handle CCFP mode.
6990 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
6991 Remove UNSPEC_SAHF unspec handling.
6993 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6995 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
6997 2017-10-27 Jeff Law <law@redhat.com>
6999 * gimple-ssa-sprintf.c: Include domwalk.h.
7000 (class sprintf_dom_walker): New class, derived from dom_walker.
7001 (sprintf_dom_walker::before_dom_children): New function.
7002 (struct call_info): Moved into sprintf_dom_walker class
7003 (compute_formath_length, handle_gimple_call): Likewise.
7004 (sprintf_length::execute): Call the dominator walker rather
7005 than walking the statements.
7007 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
7008 gimple statement locations.
7009 (check_array_bounds): Corresponding changes. Get the statement's
7010 location directly from wi->stmt.
7012 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
7015 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
7017 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
7019 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
7020 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
7022 2017-10-27 Jakub Jelinek <jakub@redhat.com>
7025 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
7026 * config/i386/i386.c (maybe_get_pool_constant): Removed.
7027 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
7028 maybe_get_pool_constant.
7029 * config/i386/predicates.md (zero_extended_scalar_load_operand):
7032 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7034 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
7035 2.26 caveat. Update gas and gld versions.
7036 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
7039 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7041 * cgraph.h (set_malloc_flag): Declare.
7042 * cgraph.c (set_malloc_flag_1): New function.
7043 (set_malloc_flag): Likewise.
7044 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
7045 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
7047 (read_ipa_call_summary): Add support for reading is_return_callee.
7048 (write_ipa_call_summary): Stream is_return_callee.
7049 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
7050 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
7051 ipa-prop.h, ipa-fnsummary.h.
7052 (pure_const_names): Change to static.
7053 (malloc_state_e): Define.
7054 (malloc_state_names): Define.
7055 (funct_state_d): Add field malloc_state.
7056 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
7057 (check_retval_uses): New function.
7058 (malloc_candidate_p): Likewise.
7059 (analyze_function): Add support for malloc attribute.
7060 (pure_const_write_summary): Stream malloc_state.
7061 (pure_const_read_summary): Add support for reading malloc_state.
7062 (dump_malloc_lattice): New function.
7063 (propagate_malloc): New function.
7064 (warn_function_malloc): New function.
7065 (ipa_pure_const::execute): Call propagate_malloc and
7066 ipa_free_fn_summary.
7067 (pass_local_pure_const::execute): Add support for malloc attribute.
7068 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
7069 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
7071 2017-10-27 Martin Liska <mliska@suse.cz>
7073 PR gcov-profile/82457
7074 * doc/invoke.texi: Document that one needs a non-strict ISO mode
7075 for fork-like functions to be properly instrumented.
7077 2017-10-27 Richard Biener <rguenther@suse.de>
7080 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
7081 info when we redirected EH.
7083 2017-10-26 Michael Collison <michael.collison@arm.com>
7085 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
7087 (<optab>_trunchf<GPI:mode>2: New pattern.
7088 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
7089 * config/aarch64/iterators.md (wv): New mode attribute.
7090 (vf, VF): New mode attributes.
7091 (vgp, VGP): New mode attributes.
7092 (s): Update attribute with SImode and DImode prefixes.
7094 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
7096 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
7097 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
7098 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
7099 (nios2_option_overide): Initialize it. Don't allow R0-relative
7100 addressing with PIC.
7101 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
7102 (nios2_symbolic_constant_p): Likewise.
7103 (nios2_legitimate_address_p): Likewise.
7104 (nios2_r0rel_section_name_p): New.
7105 (nios2_symbol_ref_in_r0rel_data_p): New.
7106 (nios2_emit_move_sequence): Handle r0rel_constant_p.
7107 (r0rel_constant_p): New.
7108 (nios2_print_operand_address): Handle r0rel_constant_p.
7109 (nios2_cdx_narrow_form_p): Likewise.
7110 * config/nios2/nios2.opt (mr0rel-sec=): New option.
7111 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
7112 (Nios II Options): Document -mr0rel-sec.
7114 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
7116 * config/nios2/nios2.c: Include xregex.h.
7117 (nios2_gprel_sec_regex): New.
7118 (nios2_option_overide): Initialize it. Don't allow GP-relative
7119 addressing with PIC.
7120 (nios2_small_section_name_p): Check for regex match.
7121 * config/nios2/nios2.opt (mgprel-sec=): New option.
7122 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
7123 (Nios II Options): Document -mgprel-sec.
7125 2017-10-26 Jim Wilson <wilson@tuliptree.org>
7127 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
7129 2017-10-26 Tom de Vries <tom@codesourcery.com>
7131 PR tree-optimization/82707
7132 * gimple.c (gimple_copy): Fix unsharing of
7133 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
7135 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
7137 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
7138 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
7139 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
7140 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
7141 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
7142 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
7143 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
7144 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
7145 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
7146 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
7147 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
7148 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
7149 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
7150 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
7151 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
7152 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
7153 _mm512_mask_cmpunord_ps_mask): New intrinsics.
7155 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7157 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
7159 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
7160 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
7161 warning to rs6000.c. Remove the Undocumented flag, since it has
7163 (-mabi=ibmlongdouble): Likewise.
7164 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
7165 already set, set the default format for long double.
7166 (rs6000_debug_reg_global): Print whether long double is IBM or
7168 (rs6000_option_override_internal): Rework setting long double
7169 format. Only warn if the user is changing the long double default
7170 and they did not use -Wno-psabi.
7171 * doc/invoke.texi (PowerPC options): Update the documentation for
7172 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
7174 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7175 Alan Hayward <alan.hayward@arm.com>
7176 David Sherwood <david.sherwood@arm.com>
7178 * rtl.h (wider_subreg_mode): New function.
7179 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
7180 rather than an unsigned int *.
7181 * ira-color.c (regno_max_ref_width): Replace with...
7182 (regno_max_ref_mode): ...this new variable.
7183 (coalesced_pseudo_reg_slot_compare): Update accordingly.
7184 Use wider_subreg_mode.
7185 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
7186 rather than an unsigned int *.
7187 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
7188 (process_alt_operands): Likewise.
7189 (invariant_p): Likewise.
7190 * lra-spills.c (assign_mem_slot): Likewise.
7191 (add_pseudo_to_slot): Likewise.
7192 * lra.c (collect_non_operand_hard_regs): Likewise.
7193 (add_regs_to_insn_regno_info): Likewise.
7194 * reload1.c (regno_max_ref_width): Replace with...
7195 (regno_max_ref_mode): ...this new variable.
7196 (reload): Update accordingly. Update call to
7197 ira_sort_regnos_for_alter_reg.
7198 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
7199 (init_eliminable_invariants): Update to use regno_max_ref_mode.
7200 (scan_paradoxical_subregs): Likewise.
7202 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7204 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
7205 (aarch64_frame): Add emit_frame_chain boolean.
7206 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7207 Move eh_return case to aarch64_layout_frame.
7208 (aarch64_layout_frame): Initialize emit_frame_chain.
7209 (aarch64_expand_prologue): Use emit_frame_chain.
7211 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7213 * config/aarch64/aarch64.c (aarch64_layout_frame):
7214 Ensure LR is always stored at the bottom of the callee-saves.
7215 Remove rarely used frame layout which saves callee-saves at top of
7216 frame, so the store of LR can be used as a valid probe in all cases.
7218 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7220 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
7221 Improve unaligned TImode/TFmode base/offset split.
7223 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7224 Alan Hayward <alan.hayward@arm.com>
7225 David Sherwood <david.sherwood@arm.com>
7227 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
7228 * combine.c (find_single_use_1): Likewise.
7229 (expand_field_assignment): Likewise.
7230 (move_deaths): Likewise.
7231 * lra-constraints.c (simplify_operand_subreg): Likewise.
7232 (curr_insn_transform): Likewise.
7233 * lra.c (collect_non_operand_hard_regs): Likewise.
7234 (add_regs_to_insn_regno_info): Likewise.
7235 * rtlanal.c (reg_referenced_p): Likewise.
7236 (covers_regno_no_parallel_p): Likewise.
7238 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7240 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
7241 Don't print any bits outside the precision of the value.
7242 * wide-int.cc (test_printing): Add some new tests.
7244 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7246 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
7247 supports -xbrace_comment option.
7248 * configure: Regenerate.
7249 * config.in: Regenerate.
7250 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
7251 (ASM_CPU_SPEC): Use it.
7253 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7255 * target.def (static_rtx_alignment): New hook.
7256 * targhooks.h (default_static_rtx_alignment): Declare.
7257 * targhooks.c (default_static_rtx_alignment): New function.
7258 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
7259 * doc/tm.texi: Regenerate.
7260 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
7261 instead of targetm.constant_alignment. Remove call to
7263 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7264 (cris_preferred_mininum_alignment): New function, split out from...
7265 (cris_constant_alignment): ...here.
7266 (cris_static_rtx_alignment): New function.
7267 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
7269 (ix86_constant_alignment): ...here.
7270 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7271 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7272 (mmix_static_rtx_alignment): New function.
7273 * config/spu/spu.c (spu_static_rtx_alignment): New function.
7274 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7276 2017-10-26 Tamar Christina <tamar.christina@arm.com>
7279 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
7280 Add flag_trapping_math and flag_fp_int_builtin_inexact.
7282 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
7284 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
7285 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
7286 mark as a sign-extending load.
7287 (local_pic_load_u): Define.
7289 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
7292 * fold-const.c (operand_equal_for_comparison_p): Also return true
7293 if ARG0 is a simple variant of ARG1 with narrower precision.
7294 (fold_ternary_loc): Always pass unstripped operands to the predicate.
7296 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
7298 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
7300 * i386.h (processor_costs): Add gather_static, gather_per_elt,
7301 scatter_static, scatter_per_elt.
7302 * x86-tune-costs.h: Add new cost entries.
7304 2017-10-25 Richard Biener <rguenther@suse.de>
7306 * tree-ssa-sccvn.h (vn_eliminate): Declare.
7307 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
7308 class pass_fre): Move to ...
7309 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
7310 class pass_fre): ... here and adjust for statistics.
7312 2017-10-25 Jakub Jelinek <jakub@redhat.com>
7315 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
7316 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
7317 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
7318 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
7321 2017-10-25 Richard Biener <rguenther@suse.de>
7323 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
7324 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
7325 eliminate_push_avail, eliminate_insert): Move inside...
7326 (class eliminate_dom_walker): ... this class in preparation
7328 (fini_eliminate): Remove by merging with ...
7329 (eliminate): ... this function. Adjust for class changes.
7330 (pass_pre::execute): Remove fini_eliminate call.
7331 (pass_fre::execute): Likewise.
7333 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7336 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
7337 (VPERMI2, VPERMI2I): New mode iterators.
7338 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
7339 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
7341 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
7342 mode iterator. Remove 3 old define_insn patterns.
7343 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
7344 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
7345 VPERMI2 mode iterator, remove the other two expanders.
7346 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
7347 to use VPERMI2 mode iterator, add another alternative for vpermi2*
7348 instructions, remove the other two patterns.
7349 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
7350 mode iterator, remove the other two patterns.
7351 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
7352 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
7353 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
7354 and adjust argument order accordingly.
7355 (ix86_expand_vec_perm): Adjust caller.
7356 (expand_vec_perm_1): Likewise.
7357 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
7358 (expand_vec_perm_vpermt2_vpshub2): ... this.
7359 (ix86_expand_vec_perm_const_1): Adjust caller.
7360 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
7363 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
7364 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
7365 (vec_shl_<mode>): Remove unused expander.
7366 (avx512bw_<shift_insn><mode>3): New define_insn.
7367 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
7368 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
7370 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
7373 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
7376 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7378 PR rtl-optimization/82396
7379 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
7380 (autopref_multipass_init): Simplify initialization.
7381 (autopref_rank_data): Simplify sort order.
7382 * gcc/sched-int.h (autopref_multipass_data_): Remove
7383 multi_mem_insn_p, min_offset and max_offset.
7385 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7388 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7389 Check special value of flag_omit_frame_pointer.
7390 (aarch64_can_eliminate): Likewise.
7391 (aarch64_override_options_after_change_1): Simplify handling of
7392 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
7394 2017-10-24 Richard Biener <rguenther@suse.de>
7396 PR tree-optimization/82697
7397 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
7398 zero for conditional load and unconditional store.
7400 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7402 * doc/install.texi: Document bootstrap-cet.
7404 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7407 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
7408 ENDBR instruction at function entrance if function is only
7411 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7414 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
7415 patterns to better describe from which operation the CF is computed.
7416 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
7417 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
7418 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
7419 is 0, use _0 suffixed expanders instead of emitting a comparison
7422 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
7424 * config/i386/i386.md(*movsf_internal, *movdf_internal):
7425 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
7427 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
7430 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
7431 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
7432 * loop-iv.c (iv_get_reaching_def): Likewise.
7433 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
7434 variable is promoted and the partition contains undefined values.
7436 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7438 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
7440 (nios2_address_cost): Define.
7441 (nios2_legitimize_address): Recognize (exp + constant) directly.
7442 (TARGET_ADDRESS_COST): Define.
7444 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7446 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
7447 (nios2_symbolic_memory_operand_p): Declare.
7448 (nios2_split_large_constant): Declare.
7449 (nios2_split_symbolic_memory_operand): Declare.
7450 * config/nios2/nios2.c: Adjust includes.
7451 (nios2_symbolic_constant_allowed): New.
7452 (nios2_symbolic_constant_p): New.
7453 (nios2_plus_symbolic_constant_p): New.
7454 (nios2_valid_addr_expr_p): Recognize addresses involving
7456 (nios2_legitimate_address_p): Likewise, also LO_SUM.
7457 (nios2_symbolic_memory_operand_p): New.
7458 (nios2_large_constant_p): New.
7459 (nios2_split_large_constant): New.
7460 (nios2_split_plus_large_constant): New.
7461 (nios2_split_symbolic_memory_operand): New.
7462 (nios2_legitimize_address): Code refactoring. Handle addresses
7463 involving symbolic constants.
7464 (nios2_emit_move_sequence): Likewise.
7465 (nios2_print_operand): Improve error output.
7466 (nios2_print_operand_address): Handle LO_SUM.
7467 (nios2_cdx_narrow_form_p): Likewise.
7468 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
7469 operands involving symbolic constants.
7470 (movhi_internal, movsi_internal): Likewise.
7471 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
7472 (extendhisi2, extendqi<mode>2): Likewise.
7474 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7476 * tree-pass.h (PROP_rtl_split_insns): Define.
7477 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
7479 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7481 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
7483 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7486 * target.def (const_not_ok_for_debug_p): Default to
7487 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
7488 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
7489 * targhooks.c (default_const_not_ok_for_debug_p): New function.
7490 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
7491 which targetm.const_not_ok_for_debug_p returned true.
7492 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
7494 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
7496 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
7497 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
7498 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
7500 (ix86_const_not_ok_for_debug_p): New function.
7501 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
7502 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
7504 2017-10-23 David Malcolm <dmalcolm@redhat.com>
7507 * system.h: Conditionally include "unique-ptr.h" if
7508 INCLUDE_UNIQUE_PTR is defined.
7509 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
7510 of defining INCLUDE_UNIQUE_PTR before including "system.h".
7512 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
7514 * config/rl78/rl78.md: New define_expand "subdi3".
7516 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
7519 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
7520 DF_REF_INSN if DF_REF_INSN_INFO is false.
7522 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
7524 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
7525 xmm_move instead of sse_move.
7526 (sse_store_index): New function.
7527 (ix86_register_move_cost): Be more sensible about mismatch stall;
7528 model AVX moves correctly; make difference between sse->integer and
7530 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
7531 moves; make difference between SSE and AVX.
7532 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
7533 and zmm_move. Increase size of sse load and store tables;
7534 add unaligned load and store tables; add ssemmx_to_integer.
7535 * x86-tune-costs.h: Update all entries according to real
7536 move latencies from Agner Fog's manual and chip documentation.
7538 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7541 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
7542 * config/i386/constraints.md (Wf): New constraint.
7543 * config/i386/i386.md (UNSPEC_SBB): New unspec.
7544 (cmp<dwi>_doubleword): Removed.
7545 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
7546 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
7547 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
7548 expand with cmp<dwi>_doubleword. For LTU and GEU use
7549 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
7551 * common.opt (gcolumn-info): Enable by default.
7552 * doc/invoke.texi (gcolumn-info): Document new default.
7554 2017-10-23 Richard Biener <rguenther@suse.de>
7556 PR tree-optimization/82672
7557 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
7558 Fold the stmt if we propagated into it.
7560 2017-10-23 Richard Biener <rguenther@suse.de>
7562 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
7563 (bitmap_remove_expr_from_set): ... this. All callers call this
7564 for non-constant values.
7565 (bitmap_set_subtract): Rename to...
7566 (bitmap_set_subtract_expressions): ... this. Adjust and
7568 (bitmap_set_contains_value): Remove superfluous check.
7569 (bitmap_set_replace_value): Inline into single caller ...
7570 (bitmap_value_replace_in_set): ... here and simplify.
7571 (dependent_clean): Merge into ...
7572 (clean): ... this using an overload. Adjust.
7573 (prune_clobbered_mems): Adjust.
7574 (compute_antic_aux): Likewise.
7575 (compute_partial_antic_aux): Likewise.
7577 2017-10-23 Richard Biener <rguenther@suse.de>
7579 PR tree-optimization/82129
7581 2017-08-01 Richard Biener <rguenther@suse.de>
7583 PR tree-optimization/81181
7584 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7585 (compute_antic): ... end of iteration here.
7587 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7589 * target.def (starting_frame_offset): New hook.
7590 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
7591 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
7592 * doc/tm.texi.in: Regenerate.
7593 * hooks.h (hook_hwi_void_0): Declare.
7594 * hooks.c (hook_hwi_void_0): New function.
7595 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
7596 STARTING_FRAME_OFFSET.
7597 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
7598 * reload1.c (reload): Likewise.
7599 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
7600 instead of STARTING_FRAME_OFFSET.
7601 * function.c (try_fit_stack_local): Likewise.
7602 (assign_stack_local_1): Likewise
7603 (instantiate_virtual_regs): Likewise.
7604 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
7605 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
7606 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
7607 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
7608 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
7609 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
7610 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
7611 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
7612 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
7613 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
7614 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
7615 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
7616 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
7617 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
7618 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
7619 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
7620 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
7621 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
7622 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
7623 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
7624 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
7625 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
7626 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
7627 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
7628 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
7629 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
7630 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
7631 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
7632 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
7633 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
7634 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
7635 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
7636 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
7637 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
7638 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
7639 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
7640 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
7641 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
7642 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
7643 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
7644 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
7645 * config/avr/avr.c (avr_starting_frame_offset): Make static and
7646 return a HOST_WIDE_INT.
7647 (avr_builtin_setjmp_frame_value): Use it instead of
7648 STARTING_FRAME_OFFSET.
7649 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7650 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
7651 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
7653 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7654 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
7655 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
7656 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7657 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
7658 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
7659 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7660 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
7661 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
7662 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7663 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
7664 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
7666 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7667 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
7668 * config/mips/mips.c (mips_compute_frame_info): Refer to
7669 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
7670 (mips_starting_frame_offset): New function.
7671 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7672 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
7673 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
7674 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
7675 and return a HOST_WIDE_INT.
7676 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7677 (mmix_initial_elimination_offset): Refer to
7678 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
7679 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
7680 * config/pa/pa.c (pa_starting_frame_offset): New function.
7681 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
7682 (pa_expand_prologue): Likewise.
7683 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7684 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
7685 !FRAME_GROWS_DOWNWARD handling to...
7686 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7687 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
7688 !FRAME_GROWS_DOWNWARD handling to...
7689 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7690 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
7691 !FRAME_GROWS_DOWNWARD handling to...
7692 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7693 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
7695 (rs6000_starting_frame_offset): New function.
7696 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
7697 !FRAME_GROWS_DOWNWARD handling to...
7698 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7699 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
7700 !FRAME_GROWS_DOWNWARD handling to...
7701 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7702 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
7703 !FRAME_GROWS_DOWNWARD handling to...
7704 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7705 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
7706 (rs6000_starting_frame_offset): New function.
7707 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
7708 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
7709 * config/vax/vax.c (vax_starting_frame_offset): New function.
7710 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
7711 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7712 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
7713 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
7714 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7715 * system.h (STARTING_FRAME_OFFSET): Poison.
7717 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7719 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
7720 SCALAR_TYPE_MODE instead of TYPE_MODE.
7722 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7723 Alan Hayward <alan.hayward@arm.com>
7724 David Sherwood <david.sherwood@arm.com>
7726 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
7728 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7729 Alan Hayward <alan.hayward@arm.com>
7730 David Sherwood <david.sherwood@arm.com>
7732 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
7734 2017-10-23 Richard Biener <rguenther@suse.de>
7736 PR tree-optimization/82129
7737 * tree-ssa-pre.c (bitmap_set_and): Remove.
7738 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
7739 canonicalizing expressions in the set to those with lowest
7740 ID rather than taking that from the first edge.
7742 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7744 * combine.c (rtx_equal_for_field_assignment_p): Use
7745 byte_lowpart_offset.
7747 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7748 Alan Hayward <alan.hayward@arm.com>
7749 David Sherwood <david.sherwood@arm.com>
7751 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
7752 to a SUBREG_PROMOTED_VAR.
7754 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7755 Alan Hayward <alan.hayward@arm.com>
7756 David Sherwood <david.sherwood@arm.com>
7758 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
7759 (expand_debug_source_expr): Likewise.
7760 * combine.c (combine_simplify_rtx): Likewise.
7761 * cse.c (fold_rtx): Likewise.
7762 * optabs.c (expand_float): Likewise.
7763 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7764 (simplify_binary_operation_1): Likewise.
7766 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7767 Alan Hayward <alan.hayward@arm.com>
7768 David Sherwood <david.sherwood@arm.com>
7770 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
7771 (record_promoted_value): Likewise.
7772 * expr.c (expand_expr_real_2): Likewise.
7773 * ree.c (update_reg_equal_equiv_notes): Likewise.
7774 (combine_set_extension): Likewise.
7775 * rtlanal.c (low_bitmask_len): Likewise.
7776 * simplify-rtx.c (neg_const_int): 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 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
7784 * regcprop.c (maybe_mode_change): Likewise.
7785 * reload1.c (alter_reg): Likewise.
7787 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7789 * inchash.h (inchash::hash::add_wide_int): New function.
7790 * lto-streamer-out.c (hash_tree): Use it.
7792 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7794 * inchash.h (inchash::hash::add_wide_int): Rename to...
7795 (inchash::hash::add_hwi): ...this.
7796 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
7797 (polymorphic_call_target_hasher::hash): Likewise.
7798 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
7799 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
7800 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
7801 * lto-streamer-out.c (hash_tree): Likewise.
7802 * optc-save-gen.awk: Likewise.
7803 * tree.c (add_expr): Likewise.
7805 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
7808 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
7809 for ordered inequality comparisons even with TARGET_IEEE_FP.
7811 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
7814 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
7815 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
7816 Expand with cmp<dwi>_doubleword.
7818 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7820 * extend.texi: Add x86 specific to 'nocf_check' attribute.
7821 List CET intrinsics.
7822 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
7823 specific to -fcf-protection option.
7825 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7827 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
7828 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
7829 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
7830 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
7831 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
7832 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
7833 (extra_objs): Add cet.o for Linux/x86 targets.
7834 (tmake_file): Add i386/t-cet for Linux/x86 targets.
7835 * config/i386/cet.c: New file.
7836 * config/i386/cetintrin.h: Likewise.
7837 * config/i386/t-cet: Likewise.
7838 * config/i386/cpuid.h (bit_SHSTK): New.
7839 (bit_IBT): Likewise.
7840 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
7841 pass IBT and SHSTK bits.
7842 * config/i386/i386-builtin-types.def
7843 (VOID_FTYPE_UNSIGNED_PVOID): New.
7844 (VOID_FTYPE_UINT64_PVOID): Likewise.
7845 * config/i386/i386-builtin.def: Add CET intrinsics.
7846 * config/i386/i386-c.c (ix86_target_macros_internal): Add
7847 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
7848 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
7849 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
7851 * config/i386/i386.c (rest_of_insert_endbranch): New.
7852 (pass_data_insert_endbranch): Likewise.
7853 (pass_insert_endbranch): Likewise.
7854 (make_pass_insert_endbranch): Likewise.
7855 (ix86_notrack_prefixed_insn_p): Likewise.
7856 (ix86_target_string): Add -mibt, -mshstk flags.
7857 (ix86_option_override_internal): Add flag_cf_protection
7859 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
7860 (ix86_print_operand): Add 'notrack' prefix output.
7861 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
7862 (ix86_expand_builtin): Expand CET intrinsics.
7863 (x86_output_mi_thunk): Add 'endbranch' instruction.
7864 * config/i386/i386.h (TARGET_IBT): New.
7865 (TARGET_IBT_P): Likewise.
7866 (TARGET_SHSTK): Likewise.
7867 (TARGET_SHSTK_P): Likewise.
7868 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
7869 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
7870 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
7871 (builtin_setjmp_setup): New pattern.
7872 (builtin_longjmp): Likewise.
7873 (rdssp<mode>): Likewise.
7874 (incssp<mode>): Likewise.
7875 (saveprevssp): Likewise.
7876 (rstorssp): Likewise.
7877 (wrss<mode>): Likewise.
7878 (wruss<mode>): Likewise.
7879 (setssbsy): Likewise.
7880 (clrssbsy): Likewise.
7881 (nop_endbr): Likewise.
7882 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
7884 * config/i386/immintrin.h: Include <cetintrin.h>.
7885 * config/i386/linux-common.h
7886 (file_end_indicate_exec_stack_and_cet): New prototype.
7887 (TARGET_ASM_FILE_END): New.
7889 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7891 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
7892 latencies instead of having separate table; make difference between
7893 integer and float costs.
7894 * i386.h (processor_costs): Remove scalar_stmt_cost,
7895 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
7896 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
7898 * x86-tune-costs.h: Remove entries which has been removed in
7899 procesor_costs from all tables; make cond_taken_branch_cost
7900 and cond_not_taken_branch_cost COST_N_INSNS based.
7902 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7904 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
7906 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7908 * config/i386/i386.md (isa): Remove fma_avx512f.
7909 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
7910 <avx512>_fmadd_<mode>_mask3<round_name>,
7911 <avx512>_fmsub_<mode>_mask<round_name>,
7912 <avx512>_fmsub_<mode>_mask3<round_name>,
7913 <avx512>_fnmadd_<mode>_mask<round_name>,
7914 <avx512>_fnmadd_<mode>_mask3<round_name>,
7915 <avx512>_fnmsub_<mode>_mask<round_name>,
7916 <avx512>_fnmsub_<mode>_mask3<round_name>,
7917 <avx512>_fmaddsub_<mode>_mask<round_name>,
7918 <avx512>_fmaddsub_<mode>_mask3<round_name>,
7919 <avx512>_fmsubadd_<mode>_mask<round_name>,
7920 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
7921 (*vec_widen_umult_even_v16si<mask_name>,
7922 *vec_widen_smult_even_v16si<mask_name>): Likewise.
7923 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
7925 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7927 * extend.texi: Add 'nocf_check' documentation.
7928 * gimple.texi: Add second parameter to
7929 gimple_build_call_from_tree.
7930 * invoke.texi: Add -fcf-protection documentation.
7931 * rtl.texi: Add REG_CALL_NOTRACK documenation.
7933 2017-10-20 Richard Biener <rguenther@suse.de>
7935 PR tree-optimization/82473
7936 * tree-vect-loop.c (vectorizable_reduction): Properly get at
7937 the largest input type.
7939 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7941 * c-attribs.c (handle_nocf_check_attribute): New function.
7942 (c_common_attribute_table): Add 'nocf_check' handling.
7943 * gimple-parser.c: Add second argument NULL to
7944 gimple_build_call_from_tree.
7945 * attrib.c (comp_type_attributes): Check nocf_check attribute.
7946 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
7948 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
7949 * common.opt: Add fcf-protection flag.
7950 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
7951 * flag-types.h: Add enum cf_protection_level.
7952 * gimple.c (gimple_build_call_from_tree): Add second parameter.
7953 Add 'nocf_check' attribute propagation to gimple call.
7954 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
7955 (gimple_build_call_from_tree): Update prototype.
7956 (gimple_call_nocf_check_p): New function.
7957 (gimple_call_set_nocf_check): Likewise.
7958 * gimplify.c: Add second argument to gimple_build_call_from_tree.
7959 * ipa-icf.c: Add nocf_check attribute in statement hash.
7960 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
7961 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
7962 * toplev.c (process_options): Add flag_cf_protection handling.
7964 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7966 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
7968 2017-10-20 Richard Biener <rguenther@suse.de>
7970 PR tree-optimization/82603
7971 * tree-if-conv.c (predicate_mem_writes): Make sure to only
7972 remove false predicated stores.
7974 2017-10-20 Richard Biener <rguenther@suse.de>
7976 * graphite-isl-ast-to-gimple.c
7977 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
7978 Remove return value and simplify, dump copied stmt after lhs
7980 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
7981 Reduce dump verbosity.
7982 (gsi_insert_earliest): Likewise.
7983 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
7984 * graphite.c (print_global_statistics): Adjust dumping.
7985 (print_graphite_scop_statistics): Likewise.
7986 (print_graphite_statistics): Do not dump loops here.
7987 (graphite_transform_loops): But here.
7989 2017-10-20 Nicolas Roche <roche@adacore.com>
7991 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
7992 * configure: Regenerate.
7994 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7997 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
7998 functions when optimizing replace GIMPLE_RETURN stmts with
7999 calls to __builtin_unreachable ().
8002 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
8003 for -fsanitize=thread link of executables.
8004 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
8005 link of executables.
8008 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
8010 (<shift_insn><mode>3<mask_name>): Change the last of the 3
8011 define_insns for logical vector shifts to use VI248_AVX512BW
8012 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
8013 condition, useless isa and prefix attributes. Change the first
8014 2 of these define_insns to ...
8015 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
8016 define_insn for avx512vl.
8017 (<shift_insn><mode>3): ... and this, new define_insn without
8018 masking for non-avx512vl.
8021 * config/i386/sse.md (*andnot<mode>3,
8022 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
8023 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
8024 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
8025 not applied use empty suffix even for TARGET_AVX512VL.
8026 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
8027 is applied, supply evex,evex or evex,evex,evex instead of just
8030 2017-10-20 Julia Koval <julia.koval@intel.com>
8032 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
8033 (OPTION_MASK_ISA_GFNI_UNSET): New.
8034 (ix86_handle_option): Handle OPT_mgfni.
8035 * config/i386/cpuid.h (bit_GFNI): New.
8036 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
8037 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
8038 * config/i386/i386.c (ix86_target_string): Add -mgfni.
8039 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
8040 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
8041 * config/i386/i386.opt: Add mgfni.
8043 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
8045 * config/msp430/msp430.c (msp430_option_override): Disable
8046 -fdelete-null-pointer-checks.
8047 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
8049 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8051 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
8052 of x87 and SSE instructions.
8054 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8056 * asan.c (create_cond_insert_point): Do not update edge count.
8057 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
8058 (afdo_propagate_circuit): Likewise.
8059 (afdo_calculate_branch_prob): Likewise.
8060 (afdo_annotate_cfg): Likewise.
8061 * basic-block.h (struct edge_def): Remove count.
8062 (edge_def::count): New accessor.
8063 * bb-reorder.c (rotate_loop): Update.
8064 (find_traces_1_round): Update.
8065 (connect_traces): Update.
8066 (sanitize_hot_paths): Update.
8067 * cfg.c (unchecked_make_edge): Update.
8068 (make_single_succ_edge): Update.
8069 (check_bb_profile): Update.
8070 (dump_edge_info): Update.
8071 (update_bb_profile_for_threading): Update.
8072 (scale_bbs_frequencies_int): Update.
8073 (scale_bbs_frequencies_gcov_type): Update.
8074 (scale_bbs_frequencies_profile_count): Update.
8075 (scale_bbs_frequencies): Update.
8076 * cfganal.c (connect_infinite_loops_to_exit): Update.
8077 * cfgbuild.c (compute_outgoing_frequencies): Update.
8078 (find_many_sub_basic_blocks): Update.
8079 * cfgcleanup.c (try_forward_edges): Update.
8080 (try_crossjump_to_edge): Update
8081 * cfgexpand.c (expand_gimple_cond): Update
8082 (expand_gimple_tailcall): Update
8083 (construct_exit_block): Update
8084 * cfghooks.c (verify_flow_info): Update
8085 (redirect_edge_succ_nodup): Update
8086 (split_edge): Update
8087 (make_forwarder_block): Update
8088 (duplicate_block): Update
8089 (account_profile_record): Update
8090 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
8091 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
8092 * cfgloopmanip.c (scale_loop_profile): Update.
8094 (lv_adjust_loop_entry_edge): Update.
8095 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
8096 (force_nonfallthru_and_redirect): Update.
8097 (purge_dead_edges): Update.
8098 (rtl_flow_call_edges_add): Update.
8099 * cgraphunit.c (init_lowered_empty_function): Update.
8100 (cgraph_node::expand_thunk): Update.
8101 * gimple-pretty-print.c (dump_probability): Update.
8102 (dump_edge_probability): Update.
8103 * gimple-ssa-isolate-paths.c (isolate_path): Update.
8104 * haifa-sched.c (sched_create_recovery_edges): Update.
8105 * hsa-gen.c (convert_switch_statements): Update.
8106 * ifcvt.c (dead_or_predicable): Update.
8107 * ipa-inline-transform.c (inline_transform): Update.
8108 * ipa-split.c (split_function): Update.
8109 * ipa-utils.c (ipa_merge_profiles): Update.
8110 * loop-doloop.c (add_test): Update.
8111 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
8112 * lto-streamer-in.c (input_cfg): Update.
8113 (input_function): Update.
8114 * lto-streamer-out.c (output_cfg): Update.
8115 * modulo-sched.c (sms_schedule): Update.
8116 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
8117 * predict.c (maybe_hot_edge_p): Update.
8118 (unlikely_executed_edge_p): Update.
8119 (probably_never_executed_edge_p): Update.
8120 (dump_prediction): Update.
8121 (drop_profile): Update.
8122 (propagate_unlikely_bbs_forward): Update.
8123 (determine_unlikely_bbs): Update.
8124 (force_edge_cold): Update.
8125 * profile.c (compute_branch_probabilities): Update.
8126 * reg-stack.c (better_edge): Update.
8127 * shrink-wrap.c (handle_simple_exit): Update.
8128 * tracer.c (better_p): Update.
8129 * trans-mem.c (expand_transaction): Update.
8130 (split_bb_make_tm_edge): Update.
8131 * tree-call-cdce.c: Update.
8132 * tree-cfg.c (gimple_find_sub_bbs): Update.
8133 (gimple_split_edge): Update.
8134 (gimple_duplicate_sese_region): Update.
8135 (gimple_duplicate_sese_tail): Update.
8136 (gimple_flow_call_edges_add): Update.
8137 (insert_cond_bb): Update.
8138 (execute_fixup_cfg): Update.
8139 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
8140 * tree-complex.c (expand_complex_div_wide): Update.
8141 * tree-eh.c (lower_resx): Update.
8142 (unsplit_eh): Update.
8143 (cleanup_empty_eh_move_lp): Update.
8144 * tree-inline.c (copy_edges_for_bb): Update.
8145 (freqs_to_counts): Update.
8146 (copy_cfg_body): Update.
8147 * tree-ssa-dce.c (remove_dead_stmt): Update.
8148 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
8149 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
8150 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
8151 (unloop_loops): Update.
8152 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8153 * tree-ssa-loop-split.c (connect_loops): Update.
8154 (split_loop): Update.
8155 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
8156 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
8157 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
8158 * tree-ssa-reassoc.c (branch_fixup): Update.
8159 * tree-ssa-tail-merge.c (replace_block_by): Update.
8160 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
8161 (compute_path_counts): Update.
8162 (update_profile): Update.
8163 (recompute_probabilities): Update.
8164 (update_joiner_offpath_counts): Update.
8165 (estimated_freqs_path): Update.
8166 (freqs_to_counts_path): Update.
8167 (clear_counts_path): Update.
8168 (ssa_fix_duplicate_block_edges): Update.
8169 (duplicate_thread_path): Update.
8170 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
8171 (case_bit_test_cmp): Update.
8172 (collect_switch_conv_info): Update.
8173 (gen_inbound_check): Update.
8174 (do_jump_if_equal): Update.
8175 (emit_cmp_and_jump_insns): Update.
8176 * tree-tailcall.c (decrease_profile): Update.
8177 (eliminate_tail_call): Update.
8178 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
8179 (vect_do_peeling): Update.
8180 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8181 * ubsan.c (ubsan_expand_null_ifn): Update.
8182 (ubsan_expand_ptr_ifn): Update.
8183 * value-prof.c (gimple_divmod_fixed_value): Update.
8184 (gimple_mod_pow2): Update.
8185 (gimple_mod_subtract): Update.
8186 (gimple_ic): Update.
8187 (gimple_stringop_fixed_value): Update.
8189 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8192 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
8194 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
8196 PR rtl-optimization/82395
8197 * ira-color.c (allocno_priority_compare_func): Fix comparison step
8198 based on non_spilled_static_chain_regno_p.
8200 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8202 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
8203 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
8204 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
8206 2017-10-19 Martin Sebor <msebor@redhat.com>
8208 PR tree-optimization/82596
8209 * tree.c (array_at_struct_end_p): Handle STRING_CST.
8211 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8213 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
8214 (get_mem_refs_of_builtin_call): Likewise.
8215 * builtins.c (expand_builtin_apply): Adjust call to
8216 allocate_dynamic_stack_space.
8217 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
8218 the third argument to allocate_dynamic_stack_space, otherwise -1.
8219 (expand_builtin): Deal with all alloca variants.
8220 (is_inexpensive_builtin): Likewise.
8221 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
8222 * calls.c (special_function_p): Deal with all alloca variants.
8223 (initialize_argument_information): Adjust call to
8224 allocate_dynamic_stack_space.
8225 (expand_call): Likewise.
8226 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
8227 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
8228 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
8229 use it for the stack usage computation.
8230 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
8231 * function.c (gimplify_parameters): Call build_alloca_call_expr.
8232 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
8233 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
8234 (in_loop_p): Remove first argument and useless check.
8235 (pass_walloca::execute): Remove useless test and adjust call to above.
8236 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
8237 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
8238 (gimplify_call_expr): Deal with all alloca variants.
8239 * hsa-gen.c (gen_hsa_alloca): Likewise.
8240 (gen_hsa_insns_for_call): Likewise.
8241 * ipa-pure-const.c (special_builtin_state): Likewise.
8242 * tree-chkp.c (chkp_build_returned_bound): Likewise.
8243 * tree-object-size.c (alloc_object_size): Likewise.
8244 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
8245 (call_may_clobber_ref_p_1): Likewise.
8246 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
8247 (ccp_fold_stmt): Likewise.
8248 (optimize_stack_restore): Likewise.
8249 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
8250 (mark_all_reaching_defs_necessary_1): Likewise.
8251 (propagate_necessity): Likewise.
8252 (eliminate_unnecessary_stmts): Likewise.
8253 * tree.c (build_common_builtin_nodes): Build
8254 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
8255 (build_alloca_call_expr): New function.
8256 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
8257 (CASE_BUILT_IN_ALLOCA): Likewise.
8258 (build_alloca_call_expr): Declare.
8259 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
8261 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8264 * dwarf2out.c (new_die_raw): New static inline function.
8265 (new_die): Use it to create the DIE.
8266 (add_AT_external_die_ref): Likewise.
8267 (clone_die): Likewise.
8268 (clone_as_declaration): Likewise.
8269 (dwarf2out_vms_debug_main_pointer): Likewise.
8270 (base_type_die): Likewise. Remove early return for corner cases.
8271 Do not call add_pubtype on the DIE here.
8272 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
8273 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
8274 typedefs for base types with DW_AT_endianity. Make sure a DIE with
8275 native order exists for base types, attach the DIE manually and call
8276 add_pubtype on it. Do not equate a reverse order DIE to the type.
8278 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
8280 * config/arm/arm.c (align_ok_ldrd_strd): New function.
8281 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
8283 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
8285 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8287 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
8288 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
8289 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
8290 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
8291 * opts.c (sanitizer_opts): Add builtin.
8292 * ubsan.c (instrument_builtin): New function.
8293 (pass_ubsan::execute): Call it.
8294 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
8295 * doc/invoke.texi: Document -fsanitize=builtin.
8297 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
8298 builtins, store max (log2 (align), 0) into uchar field instead of
8299 align into uptr field.
8300 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
8301 store uchar 0 field instead of uptr 0 field.
8302 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
8303 instead of passing one address of struct with 2 locations pass
8304 two addresses of structs with 1 location each.
8305 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
8306 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
8307 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
8308 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
8309 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
8310 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
8311 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
8312 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
8314 2017-10-19 Martin Liska <mliska@suse.cz>
8317 * file-find.c (remove_prefix): Remove.
8318 * file-find.h (remove_prefix): Likewise.
8319 * gcc-ar.c: Remove smartness of lookup.
8321 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
8323 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
8324 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
8325 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
8327 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8330 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
8331 (setcc + and to xor + setcc): New peephole2.
8333 2017-10-19 Tom de Vries <tom@codesourcery.com>
8335 * doc/sourcebuild.texi (Test Directives, Variants of
8336 dg-require-support): Add dg-require-stack-size.
8338 2017-10-19 Martin Liska <mliska@suse.cz>
8341 * gimplify.c (gimplify_decl_expr): Do not instrument variables
8342 that have a large alignment.
8343 (gimplify_target_expr): Likewise.
8345 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
8347 PR rtl-optimization/82602
8348 * ira.c (rtx_moveable_p): Return false for volatile asm.
8350 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
8353 * config/i386/i386-modes.def (CCGZ): New CC mode.
8354 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
8355 * config/i386/predicates.md (ix86_comparison_operator):
8357 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
8358 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
8359 with double-word subtraction.
8360 (put_condition_code): Handle CCGZmode.
8362 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
8364 * wide-int.cc (debug (const wide_int &)): New.
8365 (debug (const wide_int *)): New.
8366 (debug (const widest_int &)): New.
8367 (debug (const widest_int *)): New.
8369 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
8372 * lra-constraints.c (curr_insn_transform): Use non-input operand
8373 instead of output one for matched reload.
8375 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8377 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
8378 (tree-ssa-loop-ivopts.h): New header file.
8379 (struct builtin_info): New fields.
8380 (classify_builtin_1): Compute and record base and offset parts for
8381 memset builtin partition by calling strip_offset.
8382 (offset_cmp, fuse_memset_builtins): New functions.
8383 (finalize_partitions): Fuse adjacent memset partitions by calling
8385 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
8386 Expose the interface.
8387 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
8389 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8391 PR tree-optimization/82574
8392 * tree-loop-distribution.c (find_single_drs): New parameter. Check
8393 that data reference must be executed exactly once per iteration
8394 against the outermost loop in nest.
8395 (classify_partition): Update call to above function.
8397 2017-10-18 Richard Biener <rguenther@suse.de>
8399 PR tree-optimization/82591
8400 * graphite.c (graphite_transform_loops): Move code gen message
8402 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8403 Here. Handle scop_to_isl_ast failing.
8404 (scop_to_isl_ast): Limit the number of ISL operations.
8406 2017-10-18 Richard Biener <rguenther@suse.de>
8408 * graphite-isl-ast-to-gimple.c
8409 (translate_isl_ast_to_gimple::set_rename): Simplify.
8410 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
8411 (graphite_copy_stmts_from_block): ... here.
8412 (copy_bb_and_scalar_dependences): Simplify.
8413 (add_parameters_to_ivs_params): Canonicalize.
8414 (generate_entry_out_of_ssa_copies): Simplify.
8415 * graphite-sese-to-poly.c (extract_affine_name): Simplify
8416 by passing in ISL dimension.
8417 (parameter_index_in_region_1): Rename to ...
8418 (parameter_index_in_region): ... this.
8419 (extract_affine): Adjust assert, pass down parameter index.
8420 (add_param_constraints): Use range-info when available.
8421 (build_scop_context): Adjust.
8422 * sese.c (new_sese_info): Adjust.
8423 (free_sese_info): Likewise.
8424 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
8425 Remove unused typedefs.
8426 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
8428 2017-10-18 Martin Liska <mliska@suse.cz>
8430 * combine.c (simplify_compare_const): Add gcc_fallthrough.
8432 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8434 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
8435 (s390_sched_init): Do not reset s390_sched_state if we entered the
8436 current basic block via a fallthru edge and all others are unlikely.
8438 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8440 * config/s390/s390.c (NUM_SIDES): New variable.
8441 (LONGRUNNING_THRESHOLD): New variable.
8442 (LATENCY_FACTOR): New variable.
8443 (s390_sched_score): Decrease score for long-running instructions on
8445 (s390_sched_variable_issue): Perform bookkeeping for long-running
8448 2017-10-18 Richard Biener <rguenther@suse.de>
8450 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
8451 Simplify with removal of the parameter rename map.
8452 (set_rename): Likewise.
8453 (should_copy_to_new_region): Likewise.
8454 (graphite_copy_stmts_from_block): Likewise.
8455 (copy_bb_and_scalar_dependences): Remove initialization of
8456 unused copied_bb_map.
8458 (copy_internal_parameters): Likewise.
8459 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
8460 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
8461 Use INTEGRAL_TYPE_P.
8462 (parameter_index_in_region_1): Rename to ...
8463 (assign_parameter_index_in_region): ... this. Assert we have
8464 a parameter we handle.
8465 (scan_tree_for_params): Adjust.
8466 * sese.h (parameter_rename_map_t): Remove.
8467 (struct sese_info_t): Remove unused parameter_rename_map and
8468 copied_bb_map members.
8469 * sese.c (new_sese_info): Adjust.
8470 (free_sese_info): Likewise.
8472 2017-10-18 Martin Liska <mliska@suse.cz>
8475 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
8476 on an abnormal edge.
8478 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
8480 * doc/invoke.texi (ffunction-sections and fdata-sections):
8483 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8485 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
8486 the use statement can throw internally.
8488 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8490 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
8491 any RTX present on the RHS of a SET.
8492 * compare-elim.c (try_eliminate_compare): Restore comment.
8494 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8496 * langhooks.h (struct lang_hooks): Document that tree_size langhook
8497 may be also called on tcc_type nodes.
8498 * langhooks.c (lhd_tree_size): Likewise.
8500 2017-10-17 David Malcolm <dmalcolm@redhat.com>
8502 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
8503 format_warning_at_substring.
8504 (maybe_warn): Convert source_range * param to a location_t. Pass
8505 UNKNOWN_LOCATION rather than NULL to fmtwarn.
8506 (format_directive): Remove code to extract source_ranges and
8507 source_range * in favor of just a location_t.
8508 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
8510 * substring-locations.c (format_warning_va): Convert
8511 source_range * param to a location_t.
8512 (format_warning_at_substring): Likewise.
8513 * substring-locations.h (format_warning_va): Likewise.
8514 (format_warning_at_substring): Likewise.
8516 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
8518 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
8520 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
8521 and scatter/gather ops.
8523 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
8524 vec_gather_load and vec_scatter_store.
8525 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
8526 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
8527 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
8528 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
8529 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
8530 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
8532 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
8534 * reg-stack.c (compare_for_stack_reg): Add bool argument.
8535 Detect FTST instruction and handle its register pops. Only pop
8536 second operand if can_pop_second_op is true.
8537 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
8538 set can_pop_second_op to false in the compare_for_stack_reg call.
8540 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
8541 output_fp_compare for stack register operands.
8542 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
8543 instructions here. Do not emit stack register pops here. Assert
8544 that FCOMPP pops next to top stack register. Rewrite function.
8546 2017-10-17 Nathan Sidwell <nathan@acm.org>
8549 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
8550 use DECL_ASSEMBLER_NAME_RAW.
8553 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
8556 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
8557 Wilco Dijkstra <wilco.dijkstra@arm.com>
8559 * builtins.c (expand_builtin_update_setjmp_buf): Add a
8560 converstion to Pmode from the buf_addr.
8562 2017-10-17 Richard Biener <rguenther@suse.de>
8564 * graphite-dependences.c (scop_get_reads_and_writes): Change
8565 output parameters to references.
8567 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
8569 PR 71026/tree-optimization
8570 * fold-const.c (distribute_real_division): Removed.
8571 (fold_binary_loc): Remove calls to distribute_real_divison.
8573 2017-10-17 Richard Biener <rguenther@suse.de>
8575 * graphite-scop-detection.c
8576 (scop_detection::stmt_has_simple_data_refs_p): Always use
8577 the full nest as region.
8578 (try_generate_gimple_bb): Likewise.
8579 * sese.c (scalar_evolution_in_region): Simplify now that
8580 SCEV can handle instantiation in regions.
8581 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
8582 in the non-loop part of a function if requested.
8584 2017-10-17 Richard Biener <rguenther@suse.de>
8586 PR tree-optimization/82563
8587 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
8589 (graphite_regenerate_ast_isl): Call it.
8590 * graphite-scop-detection.c (build_scops): Remove entry edge split.
8592 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8594 PR tree-optimization/82549
8595 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
8596 Formatting fixes. Instead of calling make_bit_field_ref with negative
8599 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
8601 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
8602 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
8603 _mm_maskz_reduce_ss): New.
8604 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
8605 __builtin_ia32_reducess_mask): Ditto..
8606 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
8607 * config/i386/sse.md (reduces<mode>): Renamed to ...
8608 (reduces<mode><mask_scalar_name>): ... this.
8609 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
8611 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
8612 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
8614 2017-10-16 David Malcolm <dmalcolm@redhat.com>
8616 * Makefile.in (OBJS): Add unique-ptr-tests.o.
8617 * selftest-run-tests.c (selftest::run_tests): Call
8618 selftest::unique_ptr_tests_cc_tests.
8619 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
8620 * unique-ptr-tests.cc: New file.
8622 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
8625 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8627 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8628 (make_hard_regno_born, make_hard_regno_dead): Update
8629 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
8631 2017-10-16 Jeff Law <law@redhat.com>
8633 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
8635 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
8637 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
8639 2017-10-16 Olivier Hainque <hainque@adacore.com>
8641 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
8642 with_cpu if we were configured for an e500v2 target cpu name.
8644 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
8646 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
8647 * doc/invoke.texi: Document +nodsp as a valid extension for
8650 2017-10-16 Martin Liska <mliska@suse.cz>
8652 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
8653 (test_set_range): Likewise.
8654 (test_range_functions): Rename to ...
8655 (test_bit_in_range): ... this.
8656 (sbitmap_c_tests): Add new test.
8658 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8660 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
8662 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
8663 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
8665 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8667 * config/aarch64/aarch64-builtins.c
8668 (aarch64_types_quadopu_lane_qualifiers): New.
8669 (TYPES_QUADOPU_LANE): New.
8670 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
8671 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
8672 (aarch64_<sur>dot_laneq<vsi2qi>): New.
8673 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
8674 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
8675 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
8676 (Vdottype, DOTPROD): New.
8677 (sur): Add SDOT and UDOT.
8679 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8681 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
8682 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
8683 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
8685 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
8686 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
8687 Enable TARGET_DOTPROD.
8688 (cortex-a75.cortex-a55): Likewise.
8689 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
8691 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8693 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
8694 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
8696 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
8698 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
8699 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
8700 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
8701 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
8702 * config/arm/types.md (neon_dot, neon_dot_q): New.
8703 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
8705 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8707 * config/arm/arm.h (TARGET_DOTPROD): New.
8708 * config/arm/arm.c (arm_arch_dotprod): New.
8709 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
8710 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
8711 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
8712 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
8713 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
8714 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
8715 * doc/invoke.texi (armv8.2-a): Document dotprod
8717 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
8719 * i386.c (ix86_vec_cost): New function.
8720 (ix86_rtx_costs): Handle vector operations better.
8721 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
8722 * x86-tune-costs.h: Add new costs to all tables.
8724 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
8726 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
8728 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
8729 divsd, sqrtss and sqrtsd
8730 * x86-tune-costs.h: Add new entries to all costs.
8731 (znver1_cost): Fix to match real instruction latencies.
8733 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8734 Michael Collison <michael.collison@arm.com>
8736 * compare-elim.c: Include emit-rtl.h.
8737 (can_merge_compare_into_arith): New function.
8738 (try_validate_parallel): Likewise.
8739 (try_merge_compare): Likewise.
8740 (try_eliminate_compare): Call the above when no previous clobber
8742 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
8745 2017-10-14 Jakub Jelinek <jakub@redhat.com>
8749 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
8750 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
8754 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
8755 to be any operand_equal_p operands. For & (B - 1) require
8756 B to be power of 2. Recognize
8757 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
8759 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8762 * optabs.c (expand_memory_blockage): Fix call of
8763 targetm.have_memory_blockage.
8765 2017-10-14 Jakub Jelinek <jakub@redhat.com>
8768 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
8769 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
8770 objects to extra_objs instead of overwriting it.
8772 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8774 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
8775 Use any_fp_register_operand as operand[3] predicate. Simplify
8776 equality test for operands[2] and operands[4] memory location.
8777 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
8778 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
8779 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
8780 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
8781 any_fp_register_operand as operand[1] predicate. Simplify
8782 equality test for operands[0] and operands[3] memory location.
8783 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
8784 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
8785 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
8787 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8789 * target-insns.def: Add memory_blockage.
8790 * optabs.c (expand_memory_blockage): New function.
8791 (expand_asm_memory_barrier): Rename ...
8792 (expand_asm_memory_blockage): ... to this.
8793 (expand_mem_thread_fence): Call expand_memory_blockage
8794 instead of expand_asm_memory_barrier.
8795 (expand_mem_singnal_fence): Ditto.
8796 (expand_atomic_load): Ditto.
8797 (expand_atomic_store): Ditto.
8798 * doc/md.texi (Standard Pattern Names For Generation):
8799 Document memory_blockage instruction pattern.
8801 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
8803 * config/rl78/rl78.c (rl78_emit_libcall): New function.
8804 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
8805 * config/rl78/rl78.md: New define_expand "adddi3".
8807 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
8809 * cfghooks.c (verify_flow_info): Disable check that all probabilities
8812 2017-10-13 Jeff Law <law@redhat.com>
8814 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
8816 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8819 * internal-fn.c (expand_mul_overflow): If both operands have
8820 the same highpart of -1 or 0 and the topmost bit of lowpart
8821 is different, overflow is if res <= 0 rather than res < 0.
8823 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
8825 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
8826 TARGET_P9_VECTOR code for unaligned_load case.
8828 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
8830 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
8832 2017-10-13 Nathan Sidwell <nathan@acm.org>
8834 * tree-core.h (tree_contains_struct): Make bool.
8835 * tree.c (tree_contains_struct): Likewise.
8836 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
8837 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
8838 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
8839 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
8840 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
8842 2017-10-13 Richard Biener <rguenther@suse.de>
8844 * graphite-isl-ast-to-gimple.c
8845 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
8846 parameters and dominance check.
8847 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
8848 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
8849 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
8850 Do not update SSA form here or do intermediate IL verification.
8851 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
8852 (graphite_initialize): Remove check on the number of loops in
8853 the function and inline into graphite_transform_loops.
8854 (graphite_finalize): Inline into graphite_transform_loops.
8855 (graphite_transform_loops): Perform SSA update and IL verification
8857 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
8859 2017-10-13 Richard Biener <rguenther@suse.de>
8861 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
8862 graphite_expression_type_precision): Avoid global constructor
8864 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
8865 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
8866 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
8867 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
8868 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
8869 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
8871 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
8874 * config/i386/i386.h (ix86_red_zone_size): New.
8875 * config/i386/i386.md (push peephole2s): Replace
8876 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
8878 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8879 Alan Hayward <alan.hayward@arm.com>
8880 David Sherwood <david.sherwood@arm.com>
8882 * combine.c (can_change_dest_mode): Reject changes in
8883 REGMODE_NATURAL_SIZE.
8885 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8886 Alan Hayward <alan.hayward@arm.com>
8887 David Sherwood <david.sherwood@arm.com>
8889 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
8890 (expand_debug_source_expr): Likewise.
8891 * combine.c (combine_simplify_rtx): Likewise.
8892 * cse.c (fold_rtx): Likewise.
8893 * fwprop.c (canonicalize_address): Likewise.
8894 * targhooks.c (default_shift_truncation_mask): Likewise.
8896 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8897 Alan Hayward <alan.hayward@arm.com>
8898 David Sherwood <david.sherwood@arm.com>
8900 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
8901 (widened_mode): Likewise.
8902 (expand_unop): Likewise.
8903 * ree.c (transform_ifelse): Likewise.
8904 (merge_def_and_ext): Likewise.
8905 (combine_reaching_defs): Likewise.
8906 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8908 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8909 Alan Hayward <alan.hayward@arm.com>
8910 David Sherwood <david.sherwood@arm.com>
8912 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
8913 * combine.c (gen_lowpart_for_combine): Likewise.
8914 * dwarf2out.c (rtl_for_decl_location): Likewise.
8915 * final.c (alter_subreg): Likewise.
8916 * rtlhooks.c (gen_lowpart_general): Likewise.
8917 (gen_lowpart_if_possible): Likewise.
8919 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8920 Alan Hayward <alan.hayward@arm.com>
8921 David Sherwood <david.sherwood@arm.com>
8923 * calls.c (expand_call): Use subreg_lowpart_offset.
8924 * cse.c (cse_insn): Likewise.
8925 * regcprop.c (copy_value): Likewise.
8926 (copyprop_hardreg_forward_1): Likewise.
8928 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8931 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
8932 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
8933 =Q constraints to +Q and into insn condition add check
8934 that operands[0] and operands[1] are equal.
8935 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
8936 =Q constraints to +Q and into insn condition add check
8937 that operands[0] is equal to either operands[1] or operands[2].
8940 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
8941 instead of handling MINUS_EXPR twice (once for each argument),
8942 canonicalize operand order and handle just once, use rtype where
8943 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
8946 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
8947 any values of __C while still being pattern recognizable as a simple
8950 2017-10-13 Richard Biener <rguenther@suse.de>
8952 PR tree-optimization/82451
8954 2017-10-02 Richard Biener <rguenther@suse.de>
8956 PR tree-optimization/82355
8957 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
8958 a mapping for the enclosing loop but avoid generating one for
8960 (copy_bb_and_scalar_dependences): Remove premature codegen
8961 error on PHIs in blocks duplicated into multiple places.
8962 * graphite-scop-detection.c
8963 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
8964 in the region use it as loop and nest to analyze the DR in.
8965 (try_generate_gimple_bb): Likewise.
8966 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
8967 (add_loop_constraints): For blocks in a loop not in the region
8968 create a dimension with a single iteration.
8969 * sese.h (gbb_loop_at_index): Remove assert.
8971 * cfgloop.c (loop_preheader_edge): For the loop tree root
8972 return the single successor of the entry block.
8973 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8974 Reset the SCEV hashtable and niters.
8975 * graphite-scop-detection.c
8976 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
8977 assert that we only have POLYNOMIAL_CHREC that vary in loops
8978 contained in the region.
8979 (scop_detection::graphite_can_represent_expr): Adjust.
8980 (scop_detection::stmt_has_simple_data_refs_p): For loops
8981 not in the region set loop to NULL. The nest is now the
8982 entry edge to the region.
8983 (try_generate_gimple_bb): Likewise.
8984 * sese.c (scalar_evolution_in_region): Adjust for
8985 instantiate_scev change.
8986 * tree-data-ref.h (graphite_find_data_references_in_stmt):
8987 Make nest parameter the edge into the region.
8988 (create_data_ref): Likewise.
8989 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
8990 entry edge into a region and adjust instantiate_scev calls.
8991 (create_data_ref): Likewise.
8992 (graphite_find_data_references_in_stmt): Likewise.
8993 (find_data_references_in_stmt): Pass the loop preheader edge
8994 from the nest argument.
8995 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
8996 parameter the edge into the region.
8997 (instantiate_parameters): Use the loop preheader edge as entry.
8998 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
9000 (get_instantiated_value_entry): Make instantiate_below parameter
9001 the edge into the region.
9002 (instantiate_scev_name): Likewise. Adjust dominance checks,
9003 when we cannot use loop-based instantiation instantiate by
9004 walking use-def chains.
9005 (instantiate_scev_poly): Adjust.
9006 (instantiate_scev_binary): Likewise.
9007 (instantiate_scev_convert): Likewise.
9008 (instantiate_scev_not): Likewise.
9009 (instantiate_array_ref): Remove.
9010 (instantiate_scev_3): Likewise.
9011 (instantiate_scev_2): Likewise.
9012 (instantiate_scev_1): Likewise.
9013 (instantiate_scev_r): Do not blindly handle N-operand trees.
9014 Do not instantiate array-refs. Handle all constants and invariants.
9015 (instantiate_scev): Make instantiate_below parameter
9016 the edge into the region.
9017 (resolve_mixers): Use the loop preheader edge for the region
9018 parameter to instantiate_scev_r.
9019 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
9021 2017-10-13 Richard Biener <rguenther@suse.de>
9023 PR tree-optimization/82525
9024 * graphite-isl-ast-to-gimple.c
9025 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
9027 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
9028 Fail code generation when we cannot represent the isl integer.
9029 (binary_op_to_tree): Elide modulo operations that are no-ops
9030 in the type we code generate. Remove now superfluous code
9033 2017-10-13 Richard Biener <rguenther@suse.de>
9035 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
9036 (scop_detection::harmful_loop_in_region): Remove premature
9037 IV type restriction.
9038 (scop_detection::graphite_can_represent_scev): We can handle
9039 pointer IVs just fine.
9041 2017-10-13 Alan Modra <amodra@gmail.com>
9043 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
9044 "Clobbers and Scratch Registers". Add paragraph on
9045 alternative to clobbers for scratch registers and OpenBLAS
9048 2017-10-13 Alan Modra <amodra@gmail.com>
9050 * doc/extend.texi (Clobbers): Correct vax example. Delete old
9051 example of a memory input for a string of known length. Move
9052 commentary out of table. Add a number of new examples
9053 covering array memory inputs.
9055 2017-10-12 Martin Liska <mliska@suse.cz>
9057 PR tree-optimization/82493
9058 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
9059 (test_range_functions): New function.
9060 (sbitmap_c_tests): Likewise.
9061 * selftest-run-tests.c (selftest::run_tests): Run new tests.
9062 * selftest.h (sbitmap_c_tests): New function.
9064 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
9066 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
9068 * config/rs6000/amo.h: Fix spacing issue.
9070 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9073 * config/i386/i386.md (*ashl<mode>3_mask_1,
9074 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
9075 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
9078 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
9080 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
9081 (profile_probability): Set max_probability
9082 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
9084 * profile-count.c (profile_probability::differs_from_p): Do not
9085 rely on max_probaiblity == 10000
9087 2017-10-12 Jeff Law <law@redhat.com>
9089 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
9092 2017-10-12 Martin Sebor <msebor@redhat.com>
9096 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
9097 (handle_alias_pairs): Call it.
9098 * common.opt (-Wattribute-alias): New option.
9099 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
9100 * doc/invoke.texi (-Wattribute-alias): Document.
9102 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
9105 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9107 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9109 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9110 (make_hard_regno_born, make_hard_regno_dead): Update
9111 bb_killed_pseudos and bb_gen_pseudos.
9113 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
9115 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
9117 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
9119 * config/alpha/alpha.c (alpha_split_conditional_move):
9120 Use std::swap instead of manually swapping.
9121 (alpha_stdarg_optimize_hook): Ditto.
9122 (alpha_canonicalize_comparison): Ditto.
9124 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9126 * tree-loop-distribution.c (struct builtin_info): New struct.
9127 (struct partition): Refactor fields into struct builtin_info.
9128 (partition_free): Free struct builtin_info.
9129 (build_size_arg_loc, build_addr_arg_loc): Delete.
9130 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
9131 information from struct builtin_info.
9132 (find_single_drs): New function refactored from classify_partition.
9133 Also moved builtin validity checks to this function.
9134 (compute_access_range, alloc_builtin): New functions.
9135 (classify_builtin_st, classify_builtin_ldst): New functions.
9136 (classify_partition): Refactor code into functions find_single_drs,
9137 classify_builtin_st and classify_builtin_ldst.
9138 (distribute_loop): Don't do runtime alias check when distributing
9140 (find_seed_stmts_for_distribution): New function.
9141 (pass_loop_distribution::execute): Refactor code finding seed
9142 stmts into above function. Support distribution for the innermost
9143 two-level loop nest. Adjust dump information.
9145 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9147 * tree-loop-distribution.c: Adjust the general comment.
9148 (NUM_PARTITION_THRESHOLD): New macro.
9149 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
9150 (classify_partition): Skip builtin pattern of loop nest's inner loop.
9151 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
9152 in call to build_partition_graph.
9153 (finalize_partitions): New parameter. Make loop distribution more
9154 conservative by fusing more partitions.
9155 (distribute_loop): Don't do runtime alias check in case of loop nest
9157 (find_seed_stmts_for_distribution): New function.
9158 (prepare_perfect_loop_nest): New function.
9159 (pass_loop_distribution::execute): Refactor code finding seed stmts
9160 and loop nest into above functions. Support loop nest distribution.
9161 Adjust dump information accordingly.
9163 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9165 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
9166 and set PTYPE_SEQUENTIAL for merged partition.
9168 2017-10-12 Richard Biener <rguenther@suse.de>
9170 PR tree-optimization/69728
9172 2017-09-19 Richard Biener <rguenther@suse.de>
9174 PR tree-optimization/69728
9175 * graphite-sese-to-poly.c (schedule_error): New global.
9176 (add_loop_schedule): Handle empty domain by failing the
9178 (build_original_schedule): Handle schedule_error.
9180 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
9181 domain by returning an unchanged schedule.
9183 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9185 * genrecog.c (validate_pattern): For VEC_SELECT verify that
9186 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
9188 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
9190 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
9193 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9196 * expr.c (store_field): Don't optimize away bitsize == 0 store
9197 from CALL_EXPR with addressable return type.
9199 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9201 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
9202 * config/rs6000/rs6000.md (sel): Delete mode attribute.
9203 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
9204 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
9205 TARGET_ISEL instead of TARGET_ISEL<sel>.
9207 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
9209 * config/rs6000/rs6000.c
9210 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
9212 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9214 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
9216 (reg_or_cint_operand): Fix comment.
9217 (reg_or_zero_operand): New predicate.
9218 * config/rs6000/rs6000-protos.h (output_isel): Delete.
9219 * config/rs6000/rs6000.c (output_isel): Delete.
9220 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
9221 instead of reg_or_cint_operand. Output instruction directly (not via
9223 (isel_unsigned_<mode>): Ditto.
9224 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
9225 gpc_reg_operand. Add an instruction alternative for this. Output
9226 instruction directly.
9227 (*isel_reversed_unsigned_<mode>): Ditto.
9229 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9231 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
9232 (TARGET_CANONICALIZE_COMPARISON): Define.
9234 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
9237 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
9238 Check whether the dest is REG before adding REG_EQUIV note.
9240 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9243 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9245 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9246 (make_hard_regno_born, make_hard_regno_dead): Update
9247 bb_killed_pseudos and bb_gen_pseudos.
9249 2017-10-11 Nathan Sidwell <nathan@acm.org>
9251 * incpath.h (enum incpath_kind): Name enum, prefix values.
9252 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
9253 * incpath.c (heads, tails): Use INC_MAX.
9254 (add_env_var_paths, add_standard_paths): Use incpath_kind.
9255 (merge_include_chains, split_quote_chain,
9256 register_include_chains): Update incpath_kind names.
9257 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
9258 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
9260 (add_framework_path, darwin_register_objc_includes): Likewise.
9261 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
9263 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9265 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
9266 Do not use float_operator operator predicate.
9267 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
9268 * config/i386/predicates.md (float_operator): Remove predicate.
9270 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9272 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
9273 (*jccxf_i387): Ditto.
9274 (*jcc<mode>_i387): Ditto.
9275 (*jccu<mode>_i387): Ditto.
9276 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
9277 (*jcc_*_i387 splitters): Remove.
9278 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
9279 * config/i386/i386.c (ix86_split_fp_branch): Remove.
9280 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
9283 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9285 * profile-count.h (slow_safe_scale_64bit): New function.
9286 (safe_scale_64bit): New inline.
9287 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
9288 * profile-count.c: Include wide-int.h
9289 (slow_safe_scale_64bit): New.
9291 2017-10-11 Nathan Sidwell <nathan@acm.org>
9293 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
9294 HAS_DECL_ASSEMBLER_NAME_P.
9295 * gimple-expr.c (gimple_decl_printable_name: Check
9296 HAS_DECL_ASSEMBLER_NAME_P too.
9297 * ipa-utils.h (type_in_anonymous_namespace_p): Check
9298 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
9299 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
9300 * passes.c (rest_of_decl_compilation): Check
9301 HAS_DECL_ASSEMBLER_NAME_P too.
9302 * recog.c (verify_changes): Likewise.
9303 * tree-pretty-print.c (dump_decl_name): Likewise.
9304 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
9306 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
9307 (SET_DECL_ASSEMBLER_NAME): Use it.
9308 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
9309 (COPY_DECL_ASSEMBLER_NAME): Likewise.
9310 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
9312 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9314 * config.gcc (i386, x86_64): Add extra objects.
9315 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
9316 (ix86_min_insn_size): Declare.
9317 (ix86_issue_rate): Declare.
9318 (ix86_adjust_cost): Declare.
9319 (ia32_multipass_dfa_lookahead): Declare.
9320 (ix86_macro_fusion_p): Declare.
9321 (ix86_macro_fusion_pair_p): Declare.
9322 (ix86_bd_has_dispatch): Declare.
9323 (ix86_bd_do_dispatch): Declare.
9324 (ix86_core2i7_init_hooks): Declare.
9325 (ix86_atom_sched_reorder): Declare.
9326 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
9327 (COSTS_N_BYTES): Move to x86-tune-costs.h.
9328 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
9329 (rip_relative_addr_p): Rename to ...
9330 (ix86_rip_relative_addr_p): ... this one; export.
9331 (memory_address_length): Update.
9332 (ix86_issue_rate): Move to x86-tune-sched.c.
9333 (ix86_flags_dependent): Move to x86-tune-sched.c.
9334 (ix86_agi_dependent): Move to x86-tune-sched.c.
9335 (exact_dependency_1): Move to x86-tune-sched.c.
9336 (exact_store_load_dependency): Move to x86-tune-sched.c.
9337 (ix86_adjust_cost): Move to x86-tune-sched.c.
9338 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
9339 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
9340 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
9341 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
9342 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
9343 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
9344 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
9345 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
9346 (min_insn_size): Rename to ...
9347 (ix86_min_insn_size): ... this one; export.
9348 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
9349 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
9350 (core2i7_first_cycle_multipass_backtrack): Move to
9351 x86-tune-sched-core.c.
9352 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
9353 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
9354 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
9355 (ix86_avoid_jump_mispredicts): Update.
9356 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
9357 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
9358 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
9359 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
9360 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
9361 (MAX_INSN): Move to ix86-tune-sched-bd.c.
9362 (MAX_IMM): Move to ix86-tune-sched-bd.c.
9363 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
9364 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
9365 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
9366 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
9367 (MAX_STORE): Move to ix86-tune-sched-bd.c.
9368 (BIG): Move to ix86-tune-sched-bd.c.
9369 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
9370 (enum insn_path): Move to ix86-tune-sched-bd.c.
9371 (get_mem_group): Move to ix86-tune-sched-bd.c.
9372 (is_cmp): Move to ix86-tune-sched-bd.c.
9373 (dispatch_violation): Move to ix86-tune-sched-bd.c.
9374 (is_branch): Move to ix86-tune-sched-bd.c.
9375 (is_prefetch): Move to ix86-tune-sched-bd.c.
9376 (init_window): Move to ix86-tune-sched-bd.c.
9377 (allocate_window): Move to ix86-tune-sched-bd.c.
9378 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
9379 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
9380 (process_end_window): Move to ix86-tune-sched-bd.c.
9381 (allocate_next_window): Move to ix86-tune-sched-bd.c.
9382 (find_constant): Move to ix86-tune-sched-bd.c.
9383 (get_num_immediates): Move to ix86-tune-sched-bd.c.
9384 (has_immediate): Move to ix86-tune-sched-bd.c.
9385 (get_insn_path): Move to ix86-tune-sched-bd.c.
9386 (get_insn_group): Move to ix86-tune-sched-bd.c.
9387 (count_num_restricted): Move to ix86-tune-sched-bd.c.
9388 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
9389 (add_insn_window): Move to ix86-tune-sched-bd.c.
9390 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
9391 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
9392 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
9393 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
9394 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
9395 (do_dispatch): Move to ix86-tune-sched-bd.c.
9396 (has_dispatch): Move to ix86-tune-sched-bd.c.
9397 * config/i386/t-i386: Add new object files.
9398 * config/i386/x86-tune-costs.h: New file.
9399 * config/i386/x86-tune-sched-atom.c: New file.
9400 * config/i386/x86-tune-sched-bd.c: New file.
9401 * config/i386/x86-tune-sched-core.c: New file.
9402 * config/i386/x86-tune-sched.c: New file.
9404 2017-10-11 Liu Hao <lh_mouse@126.com>
9406 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
9407 the generic version below instead.
9408 (should_colorize): Recognize Windows consoles as terminals
9410 * pretty-print.c [__MINGW32__] (write_all): New function.
9411 [__MINGW32__] (find_esc_head): Likewise.
9412 [__MINGW32__] (find_esc_terminator): Likewise.
9413 [__MINGW32__] (eat_esc_sequence): Likewise.
9414 [__MINGW32__] (mingw_ansi_fputs): New function that handles
9416 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
9419 2017-10-11 Richard Biener <rguenther@suse.de>
9421 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
9422 Properly call analyze_scalar_evolution with the loop of the stmt.
9424 2017-10-11 Richard Biener <rguenther@suse.de>
9426 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
9427 * tree-core.h (tree_base): Add chrec_var union member.
9428 * tree.h (CHREC_VAR): Remove.
9429 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
9430 * tree-chrec.h (build_polynomial_chrec): Adjust.
9431 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
9432 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
9434 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
9436 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
9437 * match.pd: ... here.
9438 ((T) X == (T) Y): Relax condition.
9440 2017-10-11 Bin Cheng <bin.cheng@arm.com>
9442 PR tree-optimization/82472
9443 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
9445 (break_alias_scc_partitions): Update postorder number.
9447 2017-10-11 Martin Liska <mliska@suse.cz>
9450 * opts.c (parse_no_sanitize_attribute): Do not use error_value
9452 * opts.h (parse_no_sanitize_attribute): Remove last argument.
9454 2017-10-11 Martin Liska <mliska@suse.cz>
9456 * print-rtl.c (print_insn): Move declaration of idbuf
9457 to same scope as name.
9459 2017-10-11 Martin Liska <mliska@suse.cz>
9464 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9467 2017-10-11 Martin Liska <mliska@suse.cz>
9470 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9473 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
9475 * config.gcc (default_gnu_indirect_function): Default to yes for
9476 arm*-*-linux* with glibc.
9478 2017-10-11 Richard Biener <rguenther@suse.de>
9480 * tree-scalar-evolution.c (get_scalar_evolution): Handle
9481 default-defs and types we do not want to analyze.
9482 (interpret_loop_phi): Replace unreachable code with an assert.
9483 (compute_scalar_evolution_in_loop): Remove and inline ...
9484 (analyze_scalar_evolution_1): ... here, replacing condition with
9485 what makes the intent clearer. Remove handling of cases
9486 get_scalar_evolution now handles.
9488 2017-10-10 Jim Wilson <wilson@tuliptree.org>
9490 PR rtl-optimization/81434
9491 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
9492 comment for main loop. In sched_group_found if, also add checks for
9493 pass and min_cost_group.
9495 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
9497 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
9498 (rs6000_insn_cost): New function.
9499 * config/rs6000/rs6000.md (cost): New attribute.
9501 2017-10-10 Jakub Jelinek <jakub@redhat.com>
9502 H.J. Lu <hongjiu.lu@intel.com>
9506 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
9507 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
9508 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
9509 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
9510 to OPTION_MASK_ISA_AVX512VL - builtins that have both
9511 OPTION_MASK_ISA_MMX and some other bit set require both
9512 mmx and the ISAs without the mmx bit.
9513 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
9514 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
9515 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
9516 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
9517 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
9518 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
9519 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
9520 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
9521 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
9522 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
9523 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
9524 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
9525 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
9526 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
9527 Add OPTION_MASK_ISA_MMX.
9529 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
9531 * config.gcc (armv7*-*-freebsd*): New target.
9532 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
9534 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
9536 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
9537 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
9540 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
9542 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
9543 a template parameter.
9544 (WIDE_INT_REF_FOR): Update accordingly.
9545 * tree.h (wi::int_traits <const_tree>): Delete.
9546 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
9547 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
9548 (wi::tree_to_wide_ref): New typedef.
9549 (wi::to_wide): New function.
9550 * calls.c (get_size_range): Use wi::to_wide when operating on
9552 * cgraph.c (cgraph_node::create_thunk): Likewise.
9553 * config/i386/i386.c (ix86_data_alignment): Likewise.
9554 (ix86_local_alignment): Likewise.
9555 * dbxout.c (stabstr_O): Likewise.
9556 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
9557 * expr.c (const_vector_from_tree): Likewise.
9558 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
9559 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
9560 (fold_negate_expr_1, int_const_binop_1, const_binop)
9561 (fold_convert_const_int_from_real, optimize_bit_field_compare)
9562 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
9563 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
9564 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
9565 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
9566 (fold_not_const, round_up_loc): Likewise.
9567 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
9568 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
9569 (alloca_call_type): Likewise.
9570 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
9571 * godump.c (go_output_typedef): Likewise.
9572 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
9573 * internal-fn.c (get_min_precision): Likewise.
9574 * ipa-cp.c (ipcp_store_vr_results): Likewise.
9575 * ipa-polymorphic-call.c
9576 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
9577 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
9578 (ipa_modify_call_arguments): Likewise.
9579 * match.pd: Likewise.
9580 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
9581 * print-tree.c (print_node_brief, print_node): Likewise.
9582 * stmt.c (expand_case): Likewise.
9583 * stor-layout.c (layout_type): Likewise.
9584 * tree-affine.c (tree_to_aff_combination): Likewise.
9585 * tree-cfg.c (group_case_labels_stmt): Likewise.
9586 * tree-data-ref.c (dr_analyze_indices): Likewise.
9587 (prune_runtime_alias_test_list): Likewise.
9588 * tree-dump.c (dequeue_and_dump): Likewise.
9589 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
9590 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
9591 * tree-pretty-print.c (dump_generic_node): Likewise.
9592 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
9593 (simple_iv_with_niters): Likewise.
9594 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
9595 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
9596 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
9597 * tree-ssa-loop-niter.c (split_to_var_and_offset)
9598 (refine_value_range_using_guard, number_of_iterations_ne_max)
9599 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
9600 (get_cst_init_from_scev, record_nonwrapping_iv)
9601 (scev_var_range_cant_overflow): Likewise.
9602 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
9603 * tree-ssa-pre.c (compute_avail): Likewise.
9604 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
9605 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
9606 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
9607 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
9608 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
9609 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
9610 (dump_case_nodes, try_switch_expansion): Likewise.
9611 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
9612 (vect_do_peeling): Likewise.
9613 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
9614 * tree-vect-stmts.c (vectorizable_load): Likewise.
9615 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
9616 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
9617 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
9618 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
9619 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
9620 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
9621 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
9622 (evrp_dom_walker::before_dom_children): Likewise.
9623 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
9624 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
9625 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
9626 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
9627 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
9628 (get_range_pos_neg): Likewise.
9629 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
9630 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
9631 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
9632 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
9633 * config/avr/avr.c (avr_fold_builtin): Likewise.
9634 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
9635 * config/msp430/msp430.c (msp430_attr): Likewise.
9636 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
9637 * config/powerpcspe/powerpcspe-c.c
9638 (altivec_resolve_overloaded_builtin): Likewise.
9639 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
9640 (rs6000_expand_ternop_builtin): Likewise.
9641 * config/rs6000/rs6000-c.c
9642 (altivec_resolve_overloaded_builtin): Likewise.
9643 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
9644 (rs6000_expand_ternop_builtin): Likewise.
9645 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
9647 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9649 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
9650 when copying loop nest with only one inner loop.
9652 2017-10-10 Richard Biener <rguenther@suse.de>
9654 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
9655 blocks if SCEV is active.
9656 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
9658 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
9659 (scev_initialize): Assert we are not yet initialized.
9661 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9663 * tree-loop-distribution.c (generate_loops_for_partition): Remove
9664 inner loop's exit stmt by making it always exit the loop, otherwise
9665 we would generate an infinite empty loop.
9667 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9669 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
9670 renaming variables in new preheader if it's deleted.
9672 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9674 * tree-loop-distribution.c (struct partition): Remove unused field
9675 loops of the structure.
9676 (partition_alloc, partition_free): Ditto.
9677 (build_rdg_partition_for_vertex): Ditto.
9679 2017-10-09 Jeff Law <law@redhat.com>
9681 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
9682 return type to match prototype and documentation.
9684 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9686 * config/rs6000/rs6000.c (processor_costs): Move to ...
9687 * config/rs6000/rs6000.h: ... here.
9688 (rs6000_cost): Declare.
9690 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
9692 * except.c (setjmp_fn): New global variable.
9693 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
9694 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
9695 if DONT_USE_BUILTIN_SETJMP is defined.
9697 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9699 * target.def (insn_cost): New hook.
9700 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
9701 * doc/tm.texi: Regenerate.
9702 * rtlanal.c (insn_cost): Use the new hook.
9704 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9706 * combine.c (combine_validate_cost): Compute the new insn_cost,
9707 not just pattern_cost.
9708 (try_combine): Adjust comment.
9710 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9712 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
9714 * combine.c (uid_insn_cost): Adjust comment.
9715 (combine_validate_cost): Adjust comment. Use pattern_cost instead
9717 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
9718 * dse.c (find_shift_sequence): Ditto.
9719 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
9720 (bb_valid_for_noce_process_p): Use pattern_cost.
9721 * rtl.h (insn_rtx_cost): Delete.
9722 (pattern_cost): New prototype.
9723 (insn_cost): New prototype.
9724 * rtlanal.c (insn_rtx_cost): Rename to...
9725 (pattern_cost): ... this.
9728 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
9730 * config/i386/i386.md (*jcc_2): Remove insn pattern.
9731 (*jcc<mode>_0_r_i387): Ditto.
9732 (*jccxf_r_i387): Ditto.
9733 (*jcc<mode>_r_i387): Ditto.
9734 (*jccu<mode>_r_i387): Ditto.
9735 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
9736 (*jcc): Rename from *jcc_1.
9738 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9740 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
9741 deferred rescans after the lvx/stvx recombination pre-pass.
9743 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
9745 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
9746 memory operation instruction support.
9747 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
9748 (rs6000-ibm-aix[789]*): Likewise.
9749 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
9750 Document new functions.
9752 2017-10-09 Richard Biener <rguenther@suse.de>
9754 PR tree-optimization/82397
9755 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
9756 equality only for semantically equal trees.
9758 2017-10-09 Richard Biener <rguenther@suse.de>
9760 PR tree-optimization/82449
9761 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
9762 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
9763 allow constant addresses.
9764 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
9767 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9769 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
9772 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9775 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
9778 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9781 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
9783 2017-10-09 Jakub Jelinek <jakub@redhat.com>
9786 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
9787 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
9789 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
9791 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
9792 (WI_BINARY_PREDICATE_RESULT): Likewise.
9793 (wi::binary_traits::operator_result): New type.
9794 (wi::binary_traits::predicate_result): Likewise.
9795 (generic_wide_int::operator~, unary generic_wide_int::operator-)
9796 (generic_wide_int::operator==, generic_wide_int::operator!=)
9797 (generic_wide_int::operator&, generic_wide_int::and_not)
9798 (generic_wide_int::operator|, generic_wide_int::or_not)
9799 (generic_wide_int::operator^, generic_wide_int::operator+
9800 (binary generic_wide_int::operator-, generic_wide_int::operator*):
9802 (operator~, unary operator-, operator==, operator!=, operator&)
9803 (operator|, operator^, operator+, binary operator-, operator*): New
9805 * expr.c (get_inner_reference): Use wi::bit_and_not.
9806 * fold-const.c (fold_binary_loc): Likewise.
9807 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
9808 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
9809 (bit_value_binop): Likewise.
9810 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
9811 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
9812 (extract_range_from_binary_expr_1): Likewise.
9813 (masked_increment): Likewise.
9814 (simplify_bit_ops_using_ranges): Likewise.
9816 2017-10-09 Martin Jambor <mjambor@suse.cz>
9819 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
9820 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
9821 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
9823 (hsa_fixup_mov_insn_type): New function.
9824 (hsa_op_with_type::get_in_type): Use it.
9825 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
9826 immediates in an assert.
9827 (hsa_op_with_type::extend_int_to_32bit): New method.
9828 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
9829 types. Convert to dest type if necessary.
9830 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
9831 (reg_for_gimple_ssa): Pass false as min32int to
9832 hsa_type_for_scalar_tree_type.
9833 (gen_hsa_addr): Fixup type when creating addresable temporary.
9834 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
9835 (gen_hsa_unary_operation): Extend operands and convert to dest type if
9836 necessary. Call hsa_fixup_mov_insn_type.
9837 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
9838 extend operands and convert to dest type if necessary.
9839 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
9840 to dest type if necessary.
9841 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
9842 if conversion nt necessary and size matches.
9843 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
9844 to dest type if necessary.
9845 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
9846 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
9848 (gen_hsa_clrsb): Likewise.
9849 (gen_hsa_ffs): Likewise.
9850 (gen_hsa_divmod): Extend operands and convert to dest type if
9852 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
9854 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
9856 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
9857 Remove empty default arguments. Use a brace block as output
9859 (conditional return): Ditto.
9861 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
9862 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
9864 (group_ending_nop): Ditto.
9865 (doloop_end): Ditto.
9866 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
9867 (splitters for those): Ditto.
9869 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
9871 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
9872 a conditional jump (and the compare for it) so that pc_rtx is the
9874 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
9875 for the deleted and renamed ctr<mode>_internal[234] patterns.
9876 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
9877 Delete second conditional return pattern.
9878 (ctr<mode>_internal2): Delete this second bdnz pattern.
9879 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
9880 (ctr<mode>_internal4): Delete this second bdz pattern.
9882 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9884 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
9885 (always_initialized_rtx_for_ssa_name_p): New predicate.
9886 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
9887 (finish_out_of_ssa): Free new field of SA.
9888 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
9889 * tree-ssa-coalesce.c: Include tree-ssa.h.
9890 (get_parm_default_def_partitions): Remove extern keyword.
9891 (get_undefined_value_partitions): New function.
9892 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
9893 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
9895 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
9897 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9899 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
9901 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9903 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
9904 for targets that preffer 128bit.
9906 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9908 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
9910 2017-10-08 Olivier Hainque <hainque@adacore.com>
9912 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
9913 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
9915 (thumb_set_return_address): Likewise.
9917 2017-10-08 Olivier Hainque <hainque@adacore.com>
9919 * common/config/arm/arm-common.c (arm_except_unwind_info):
9920 Handle DWARF2_UNWIND_INFO.
9922 2017-10-07 Michael Collison <michael.collison@arm.com>
9924 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
9927 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
9929 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
9930 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
9931 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
9932 defined, force the creation of a new block for a dispatch label.
9934 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
9936 * invoke.texi (Wsuggest-attribute=cold): Document.
9937 * common.opt (Wsuggest-attribute=cold): New
9938 * ipa-pure-const.c (warn_function_cold): New function.
9939 * predict.c (compute_function_frequency): Use it.
9940 * predict.h (warn_function_cold): Declare.
9942 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
9944 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
9947 2017-10-06 Martin Liska <mliska@suse.cz>
9949 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
9950 keyword for member functions.
9951 (struct sanopt_tree_couple): New struct.
9952 (struct sanopt_tree_couple_hash): New function.
9953 (struct sanopt_ctx): Add new hash_map.
9954 (has_dominating_ubsan_ptr_check): New function.
9955 (record_ubsan_ptr_check_stmt): Likewise.
9956 (maybe_optimize_ubsan_ptr_ifn): Likewise.
9957 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
9958 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
9960 2017-10-06 Sudakshina Das <sudi.das@arm.com>
9963 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
9964 aarch64_simd_valid_immediate on CONST_VECTORs.
9965 (aarch64_reg_or_bic_imm): Likewise.
9967 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
9969 PR rtl-optimization/82396
9970 * haifa-sched.c (ready_sort_real): Disable qsort checking.
9972 2017-10-06 Sebastian Pop <sebpop@gmail.com>
9974 * graphite-dependences.c (scop_get_reads): Move code to...
9975 (scop_get_must_writes): Move code to...
9976 (scop_get_may_writes): Move code to...
9977 (scop_get_reads_and_writes): ... here.
9978 (scop_get_dependences): Call scop_get_reads_and_writes.
9980 2017-10-06 Jakub Jelinek <jakub@redhat.com>
9982 PR tree-optimization/82434
9983 * fold-const.h (can_native_encode_type_p,
9984 can_native_encode_string_p): Remove.
9985 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
9986 don't encode anything, just return what would be otherwise returned.
9987 (native_encode_fixed, native_encode_complex, native_encode_vector):
9989 (native_encode_string): Likewise. Inline by hand
9990 can_native_encode_string_p.
9991 (can_native_encode_type_p): Remove.
9992 (can_native_encode_string_p): Remove.
9993 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
9994 STRING_CSTs using can_native_encode_string_p, test all
9995 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
9996 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
9997 argument from native_encode_expr.
9998 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
9999 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
10002 2017-10-06 Richard Biener <rguenther@suse.de>
10004 PR tree-optimization/82397
10005 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
10006 operand_equal_p but rely on data_ref_compare_tree for detecting
10008 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
10009 to match up with dr_group_sort_cmp.
10011 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10014 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
10016 * config/s390/s390-builtin-types.def: Regenerate.
10018 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10021 * config/s390/s390-builtin-types.def: Regenerate.
10022 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
10023 Change flag from B_VXE to B_VX.
10024 (s390_vec_min_dbl): Remove B_VXE flag.
10026 2017-10-06 Richard Biener <rguenther@suse.de>
10028 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
10029 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
10030 translate_isl_ast_to_gimple::is_valid_rename,
10031 translate_isl_ast_to_gimple::get_rename,
10032 translate_isl_ast_to_gimple::get_def_bb_for_const,
10033 translate_isl_ast_to_gimple::get_new_name,
10034 translate_isl_ast_to_gimple::collect_all_ssa_names,
10035 translate_isl_ast_to_gimple::copy_loop_phi_args,
10036 translate_isl_ast_to_gimple::collect_all_ssa_names,
10037 translate_isl_ast_to_gimple::copy_loop_phi_args,
10038 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
10039 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
10040 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
10041 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
10042 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
10043 translate_isl_ast_to_gimple::copy_cond_phi_args,
10044 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
10045 translate_isl_ast_to_gimple::edge_for_new_close_phis,
10046 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
10047 translate_isl_ast_to_gimple::rename_uses,
10048 translate_isl_ast_to_gimple::rename_all_uses): Remove.
10049 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
10050 (set_rename_for_each_def): Likewise.
10051 (graphite_copy_stmts_from_block): Handle debug stmt resetting
10052 here. Handle rewriting SCEV analyzable uses here.
10053 (copy_bb_and_scalar_dependences): Generate code for PHI
10055 (graphite_regenerate_ast_isl): Adjust.
10056 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
10057 (add_write, add_read): New functions.
10058 (build_cross_bb_scalars_def): Use it and simplify.
10059 (build_cross_bb_scalars_use): Likewise.
10060 (graphite_find_cross_bb_scalar_vars): Inline into...
10061 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
10062 simulating out-of-SSA. Compute liveout and add dependencies.
10063 (build_scops): Force an empty entry block.
10064 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
10066 (sese_build_liveouts): Declare.
10067 (sese_trivially_empty_bb_p): Likewise.
10068 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
10069 compute liveout and debug_liveout.
10070 (sese_bad_liveouts_use): Remove.
10071 (sese_reset_debug_liveouts_bb): Likewise.
10072 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
10073 (sese_build_liveouts): Build liveout and debug_liveout and store
10075 (new_sese_info): Adjust.
10076 (free_sese_info): Likewise.
10077 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
10078 do not build liveout here.
10079 (move_sese_in_condition): Adjust region entry.
10080 (scev_analyzable_p): Match up with chrec_apply requirements.
10081 (sese_trivially_empty_bb_p): New.
10082 * tree-into-ssa.c (get_reaching_def): Properly support generating
10083 default-defs for incremental rewrite of anonymous names.
10085 2017-10-06 Richard Biener <rguenther@suse.de>
10087 * graphite-sese-to-poly.c (extract_affine): For casts increasing
10088 precision do not perform modulo reduction.
10090 2017-10-06 Richard Biener <rguenther@suse.de>
10092 PR tree-optimization/82436
10093 * tree-vect-slp.c (vect_supported_load_permutation_p): More
10094 conservatively choose the vectorization factor when checking
10095 whether we can perform the required load permutation.
10096 (vect_transform_slp_perm_load): Assert when we may not fail.
10098 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
10100 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
10101 message for incompatible -msdata=* and -mcall-* options.
10103 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10105 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
10106 rate for post-reload scheduling.
10108 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10110 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
10112 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10114 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
10115 to improve monte carlo in scimark.
10117 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10119 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
10120 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
10121 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
10122 pentium4_cost, nocona_cost): Set reassociation width to 1.
10123 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
10124 width to 2 for fp operations and 1 otherwise.
10125 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
10127 (atom_cost): Set reassociation width to 2.
10128 (slm_cost, generic_cost): Set fp reassociation width
10129 to 2 and 1 otherwise.
10130 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
10131 (core_cost): Set fp reassociation width to 4 and vector to 2.
10132 (ix86_reassociation_width): Rewrite using cost table; special case
10133 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
10134 and TARGET_AVX128_OPTIMAL.
10135 * config/i386/i386.h (processor_costs): Add
10136 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
10137 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
10138 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
10139 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
10140 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
10141 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
10143 2017-10-05 Nathan Sidwell <nathan@acm.org>
10145 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
10147 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10149 * config/arm/arm.c (arm_test_fpu_data): New.
10150 (arm_run_selftests): Call arm_test_fpu_data.
10152 2017-10-04 Nathan Sidwell <nathan@acm.org>
10154 * toplev.c (toplev::main): Remove excess parens on pretty_printer
10156 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
10158 2017-10-04 Sudakshina Das <sudi.das@arm.com>
10160 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
10161 check type for aarch64_simd_valid_immediate.
10162 (aarch64_output_simd_mov_immediate): Update prototype.
10163 (aarch64_simd_valid_immediate): Update prototype.
10164 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
10165 support for ORR-immediate.
10166 (and<mode>3): modified pattern to add support for BIC-immediate.
10167 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
10168 now checks for valid immediate for BIC and ORR based on new enum
10170 (aarch64_output_simd_mov_immediate): Function now used to output
10171 BIC/ORR imm as well based on new enum argument.
10172 * config/aarch64/constraints.md (Do): New vector immediate constraint.
10174 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
10175 (aarch64_reg_or_bic_imm): Likewise.
10177 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10179 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
10180 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
10182 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10186 PR rtl-optimization/82396
10187 * haifa-sched.c (autopref_multipass_init): Simplify
10189 (autopref_rank_data): Simplify sort order.
10190 * sched-int.h (autopref_multipass_data_): Remove
10191 multi_mem_insn_p, min_offset and max_offset.
10193 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10195 * doc/sourcebuild.texi: Document vect_peeling_profitable.
10197 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10199 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
10200 vect_doubleint_cvt.
10202 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10204 * doc/sourcebuild.texi: Document vect_long_mult.
10206 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
10208 PR tree-optimization/82413
10209 * fold-const.c (build_range_check): Use widest_int when comparing
10210 the maximum ETYPE value with HIGH.
10212 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10214 PR rtl-optimization/82396
10215 * haifa-sched.c (autopref_multipass_init): Simplify
10217 (autopref_rank_data): Simplify sort order.
10218 * sched-int.h (autopref_multipass_data_): Remove
10219 multi_mem_insn_p, min_offset and max_offset.
10221 2017-10-04 Jakub Jelinek <jakub@redhat.com>
10223 PR tree-optimization/82381
10224 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
10225 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
10228 PR tree-optimization/82374
10229 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
10230 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
10231 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
10232 current_function_decl to the new decl.
10234 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
10236 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
10237 helper macro for IEEE float128 hardware built-in functions.
10238 (SQRTF128_ODD): Add built-in functions with the round-to-odd
10240 (TRUNCF128_ODD): Likewise.
10241 (ADDF128_ODD): Likewise.
10242 (SUBF128_ODD): Likewise.
10243 (MULF128_ODD): Likewise.
10244 (DIVF128_ODD): Likewise.
10245 (FMAF128_ODD): Likewise.
10246 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
10247 UNSPEC_TRUNC_ROUND_TO_ODD.
10248 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
10249 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
10250 floating point round to odd instructions.
10251 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
10252 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
10253 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
10254 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
10255 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
10256 (trunc<mode>sf2_hw): Change the truncate with round to odd
10257 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
10258 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
10259 to odd hardware instructions.
10260 (sub<mode>3_odd): Likewise.
10261 (mul<mode>3_odd): Likewise.
10262 (div<mode>3_odd): Likewise.
10263 (sqrt<mode>2_odd): Likewise.
10264 (fma<mode>4_odd): Likewise.
10265 (fms<mode>4_odd): Likewise.
10266 (nfma<mode>4_odd): Likewise.
10267 (nfms<mode>4_odd): Likewise.
10268 (trunc<mode>df2_odd): Change the truncate with round to odd
10269 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
10271 * doc/extend.texi (PowerPC built-in functions): Update documentation
10272 for existing IEEE float128-bit built-in functions. Add built-in
10273 functions that generate the IEEE 128-bit floating point round to
10276 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
10278 PR rtl-optimization/77729
10279 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
10280 to (X&(C1&~C2))|C2 transformations.
10282 2017-10-03 Martin Jambor <mjambor@suse.cz>
10284 PR tree-optimization/82363
10285 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
10286 mismatch, mark lacc written regardless of racc.
10288 2017-10-03 Jakub Jelinek <jakub@redhat.com>
10290 PR tree-optimization/82381
10291 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
10292 stmt_to_insert nor wheather SSA_NAMEs are default defs.
10293 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
10294 fallthrough into reassoc_stmt_dominates_stmt_p.
10297 * combine.c (combine_instructions): Don't combine in unreachable
10300 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
10303 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
10304 function to not use the have_cpu variable. Do not set cpu_index,
10305 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
10306 or the default cpu.
10307 (rs6000_valid_attribute_p): Remove duplicate initializations of
10308 old_optimize and func_optimize.
10309 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
10310 (rs6000_activate_target_options): Make global.
10311 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
10314 2017-10-02 Jakub Jelinek <jakub@redhat.com>
10316 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
10317 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
10318 Set *poffset to 0 and *psize and *pmax_size to -1 if
10319 *poffset + *psize overflows in HOST_WIDE_INT.
10321 PR tree-optimization/82387
10322 PR tree-optimization/82388
10323 PR tree-optimization/82389
10324 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
10325 instead of live_bytes non-NULL.
10327 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
10330 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
10333 2017-10-02 Richard Biener <rguenther@suse.de>
10335 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
10336 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
10337 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
10339 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10341 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
10342 requested precision matches the type's.
10343 * calls.c (alloc_max_size): Calculate the new candidate size as
10344 a widest_int and use wi::to_widest when comparing it with the
10345 current candidate size.
10346 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
10347 zero rather than integer_zero_node.
10348 * match.pd: Check for a no-op conversion before using wi::add
10349 rather than after. Use tree_to_uhwi when summing small shift
10350 counts into an unsigned int.
10352 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10353 Alan Hayward <alan.hayward@arm.com>
10354 David Sherwood <david.sherwood@arm.com>
10357 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
10358 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
10359 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
10360 POINTER_AND_FP_REGS.
10362 2017-10-02 Richard Biener <rguenther@suse.de>
10364 PR tree-optimization/82355
10365 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
10366 a mapping for the enclosing loop but avoid generating one for
10367 the loop tree root.
10368 (copy_bb_and_scalar_dependences): Remove premature codegen
10369 error on PHIs in blocks duplicated into multiple places.
10370 * graphite-scop-detection.c
10371 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
10372 in the region use it as loop and nest to analyze the DR in.
10373 (try_generate_gimple_bb): Likewise.
10374 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
10375 (add_loop_constraints): For blocks in a loop not in the region
10376 create a dimension with a single iteration.
10377 * sese.h (gbb_loop_at_index): Remove assert.
10379 2017-10-01 Kevin Buettner <kevinb@redhat.com>
10381 * omp-expand.c (adjust_context_scope): New function.
10382 (expand_parallel_call): Call adjust_context_scope.
10384 2017-10-01 Jeff Law <law@redhat.com>
10386 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
10387 dom_opt_dom_walker class with direct access to private members.
10388 Add comments. Call test_for_singularity.
10389 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
10390 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
10391 m_dummy_cond anymore.
10392 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
10394 (pass_dominator:execute): Build the dummy_cond here and pass it
10395 to the dom_opt_dom_walker ctor.
10396 (test_for_singularity): New function.
10398 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
10399 Maya Rashish <coypu@sdf.org>
10401 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
10402 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
10403 (alpha*-*-netbsd*) Use nbsd_tm_file.
10404 (arm*-*-netbsdelf*) Likewise.
10405 (i[34567]86-*-netbsdelf*) Likewise.
10406 (x86_64-*-netbsd*) Likewise.
10407 (mips*-*-netbsd*) Likewise.
10408 (powerpc-*-netbsd*) Likewise.
10409 (sh*-*-netbsd*) Likewise.
10410 (sparc-*-netbsdelf*) Likewise.
10411 (sparc64-*-netbsd*) Likewise.
10412 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
10414 (vax-*-netbsdelf*) Likewise.
10415 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
10416 (UINT_FAST8_TYPE) Likewise.
10417 (INT_FAST16_TYPE) Check CHAR_FAST16.
10418 (UINT_FAST16_TYPE) Likewise.
10420 2017-09-30 Jakub Jelinek <jakub@redhat.com>
10423 * config/i386/i386.md
10424 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
10425 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
10426 *divmodsi4_zext_2): New define_insn_and_split.
10427 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
10428 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
10429 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
10430 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
10431 New define_insn_and_split.
10432 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
10433 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
10434 operands[1] having DImode when mode is SImode.
10436 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
10437 always SImode for DIV and MOD in REG_EQUAL notes.
10439 2017-09-29 Yury Gribov <tetra2005@gmail.com>
10441 PR middle-end/82319
10442 * match.pd: Fix handling of NaNs in pattern.
10444 2017-09-29 Jeff Law <law@redhat.com>
10446 * sbitmap.c (bitmap_bit_in_range_p): New function.
10447 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
10448 * tree-ssa-dse.c (live_bytes_read): New function.
10449 (dse_classify_store): Ignore reads of dead bytes.
10451 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
10452 typos and whitespace errors.
10453 * config/i386/predicates.md (address_no_seg_operand): Likewise.
10454 * config/s390/s390.c (s390_emit_prologue): Likewise.
10456 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10459 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
10460 with a symbol for LRA.
10462 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10464 PR rtl-optimization/82338
10465 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
10467 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
10469 * genmodes.c (calc_wider_mode): Suppress qsort macro.
10470 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
10471 (qsort_chk): Declare.
10472 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
10473 (qsort_chk): New function.
10475 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10477 PR tree-optimization/82337
10478 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
10479 phi definition if the PHI result appears in an abnormal PHI.
10480 (find_basis_for_base_expr): Don't record a basis if the LHS of the
10481 basis appears in an abnormal PHI.
10483 2017-09-29 Richard Biener <rguenther@suse.de>
10485 * graphite-isl-ast-to-gimple.c
10486 (translate_isl_ast_to_gimple::set_codegen_error): New function.
10487 (binary_op_to_tree): Use it.
10488 (get_rename_from_scev): Likewise.
10489 (copy_loop_phi_nodes): Likewise.
10490 (copy_bb_and_scalar_dependences): Likewise.
10491 (translate_pending_phi_nodes): Likewise.
10493 2017-09-29 Jakub Jelinek <jakub@redhat.com>
10496 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
10497 for movabsq $(i32 << shift), r64.
10499 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
10501 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
10502 index when encoding %esp as %rsp to avoid 0x67 prefix.
10504 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
10506 * config/i386/i386.md (*movsf_internal, *movdf_internal):
10507 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
10509 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
10511 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
10512 Extensions with more than 16 double VFP registers.
10513 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
10514 to_clear_mask and all code related to it. Replace the remaining
10515 entry by a sbitmap and adapt code accordingly.
10517 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
10519 * brig-builtins.def: Change pure attributes to const.
10521 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10523 * config.gcc (default_gnu_indirect_function): Default to yes for
10524 sparc*-*-linux* with glibc.
10526 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10528 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
10529 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
10530 when creating .init_array and .fini_array sections with priority
10533 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
10536 * config/aarch64/aarch64.c
10537 (aarch64_builtin_support_vector_misalignment): Always return false
10538 when misalignment is unknown.
10540 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
10542 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
10543 this function to return false if the definition used by the swap
10544 instruction is artificial, or if the memory address from which the
10545 constant value is loaded is not represented by a base address held
10546 in a register or if the base address register is a frame or stack
10547 pointer. Additionally, return false if the base address of the
10548 loaded constant is a SYMBOL_REF but is not considered to be a
10550 (replace_swapped_load_constant): New function.
10551 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
10552 loaded constant vector with a load of a swapped constant vector.
10554 2017-09-27 Carl Love <cel@us.ibm.com>
10556 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
10557 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
10558 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
10561 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
10563 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
10564 first, always call autopref_rank_data otherwise.
10566 2017-09-27 Richard Biener <rguenther@suse.de>
10568 * graphite-scop-detection.c (find_scop_parameters): Move
10569 loop bound handling ...
10570 (gather_bbs::before_dom_children): ... here, avoiding the need
10571 to build scop_info->loop_nest.
10572 (record_loop_in_sese): Remove.
10573 * sese.h (sese_info_t::loop_nest): Remove.
10574 * sese.c (new_sese_info): Do not allocate loop_nest.
10575 (free_sese_info): Do not free loop_nest.
10577 2017-09-27 Jakub Jelinek <jakub@redhat.com>
10580 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
10581 lhs from calls if the lhs has addressable type.
10583 2017-09-27 Richard Biener <rguenther@suse.de>
10585 * graphite.h (scop::max_alias_set): New member.
10586 * graphite-scop-detection.c: Remove references to non-existing
10587 --param in comments.
10588 (build_alias_sets): Record the maximum alias set used for drs.
10589 (build_scops): Support zero as unlimited for
10590 --param graphite-max-arrays-per-scop.
10591 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
10592 and inline into ...
10593 (build_poly_sr_1): ... here. Compute alias set based on the
10594 maximum alias set used for drs rather than
10595 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
10597 2017-09-27 Richard Biener <rguenther@suse.de>
10599 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
10600 --param loop-block-tile-size=0 to disable tiling.
10602 2017-09-27 Richard Biener <rguenther@suse.de>
10604 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
10605 (graphite-max-nb-scop-params): Document special value zero.
10606 * domwalk.h (dom_walker::STOP): New symbolical constant.
10607 (dom_walker::dom_walker): Add optional parameter for bb to
10609 (dom_walker::~dom_walker): Declare.
10610 (dom_walker::before_dom_children): Document STOP return value.
10611 (dom_walker::m_user_bb_to_rpo): New member.
10612 (dom_walker::m_bb_to_rpo): Likewise.
10613 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
10614 mapping here if not provided by the user.
10615 (dom_walker::~dom_walker): Free bb to RPO mapping if not
10616 provided by the user.
10617 (dom_walker::STOP): Define.
10618 (dom_walker::walk): Do not compute bb to RPO mapping here.
10619 Support STOP return value from before_dom_children to stop
10621 * graphite-optimize-isl.c (optimize_isl): If the schedule
10622 is the same still generate code if -fgraphite-identity
10623 or -floop-parallelize-all are given.
10624 * graphite-scop-detection.c: Include cfganal.h.
10625 (gather_bbs::gather_bbs): Get and pass through bb to RPO
10627 (gather_bbs::before_dom_children): Return STOP for BBs
10629 (build_scops): Compute bb to RPO mapping and pass it to
10630 the domwalk. Treat --param graphite-max-nb-scop-params=0
10631 as not limiting the number of params.
10632 * graphite.c (graphite_initialize): Remove limit on the
10633 number of basic-blocks in a function.
10634 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
10635 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
10636 default value of 10.
10638 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10640 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
10641 Adjust code to eliminate needing to do the shift right 32-bits
10642 operation after XSCVDPSPN.
10644 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10646 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
10647 ((X / Y) != 0 -> X >= Y): Likewise.
10649 2017-09-26 Carl Love <cel@us.ibm.com>
10651 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
10652 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
10653 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
10654 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
10655 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
10656 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
10657 definitions and overloading.
10658 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
10659 statement for P9V_BUILTIN_XST_LEN_R.
10660 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
10661 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
10662 define_expand and define_insn for the instructions and builtins.
10663 * doc/extend.texi: Update the built-in documentation file for the new
10664 built-in functions.
10665 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
10666 define_insn for the instructions
10668 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
10671 * gcc/config/netbsd-protos.h: New file.
10672 * gcc/config/netbsd.c: New file.
10673 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
10674 * gcc/config/t-netbsd: New file.
10675 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
10676 (tmake_file) Add t-netbsd.
10677 (extra_objs) Add netbsd.o.
10679 2017-09-26 Janus Weil <janus@gcc.gnu.org>
10683 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
10685 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10687 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
10688 sign extension from a vector register to a GPR by doing a 32-bit
10689 direct move and then an EXTSW.
10690 (extendsi<mode>2 splitter): Likewise.
10691 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
10692 right or vector extract after doing XSCVDPSPN. Use
10693 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
10695 (movdi_from_sf_zero_ext): Likewise.
10696 (reload_gpr_from_vsxsf): Likewise.
10697 (p8_mfvsrd_4_disf): Delete, no longer used.
10698 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
10699 then needing to move the SFmode to a GPR to use the XSCVDPSP
10700 instruction instead of FRSP and XSCVDPSPN.
10701 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
10702 it is adjacent to the other XSCVSPDP insns.
10703 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
10704 SFmode to be in traditional Altivec registers.
10705 (vsx_xscvdpspn): Eliminate useless alternative constraint.
10706 (vsx_xscvspdpn): Likewise.
10707 (vsx_xscvspdpn_scalar): Likewise.
10709 2017-09-26 Martin Jambor <mjambor@suse.cz>
10711 * tree-sra.c (compare_access_positions): Put integral types first,
10712 stabilize sorting of integral types, remove conditions putting
10713 non-full-precision integers last.
10714 (sort_and_splice_var_accesses): Disable scalarization if a
10715 non-integert would be represented by a non-full-precision integer.
10717 2017-09-26 Joseph Myers <joseph@codesourcery.com>
10719 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
10720 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
10721 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
10722 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
10723 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
10724 conditionals inside the function instead of around it. Call
10725 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
10726 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
10728 2017-09-26 Richard Biener <rguenther@suse.de>
10730 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
10732 (scop_detection::build_scop_breadth): ... this. Removed.
10733 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
10734 (scop_detection::harmful_stmt_in_bb): Likewise.
10735 (scop_detection::graphite_can_represent_stmt): Likewise.
10736 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
10737 (scop_detection::can_represent_loop): Remove recursion, fold in ...
10738 (scop_detection::can_represent_loop_1): ... this. Removed.
10739 (scop_detection::harmful_loop_in_region): Simplify after inlining
10740 the above and remove more quadraticness.
10741 (build_scops): Adjust.
10742 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
10745 2017-09-26 Jakub Jelinek <jakub@redhat.com>
10748 * config/i386/i386.c (ix86_print_operand_address_as): Only test
10749 REGNO (base) == SP_REG if base is a REG.
10751 PR middle-end/35691
10752 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
10753 if it is different SSA_NAME.
10754 (optimize_range_tests_cmp_bitwise): New function.
10755 (optimize_range_tests): Call it.
10757 2017-09-26 Richard Biener <rguenther@suse.de>
10759 PR tree-optimization/82321
10760 * graphite.c (canonicalize_loop_closed_ssa): Properly check
10761 for the def being inside the loop.
10763 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10765 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
10767 * config/s390/s390-builtins.def: Fix constraint on op4.
10769 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10771 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
10772 independent expanders.
10773 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
10774 ("vec_ordered", "vec_unordered"): New expanders.
10776 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10778 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
10781 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10783 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
10784 vec_unpacks_lo_v16qi.
10785 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
10787 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10789 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
10790 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
10791 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
10793 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10795 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
10797 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
10799 ("*vec_slb<mode>"): New insn pattern.
10800 ("vec_shr_<mode>"): New expander.
10801 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
10802 and force the shift count operand to V16QImode.
10803 ("vec_srb<mode>"): Set shift count mode to V16QI.
10805 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10807 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
10808 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
10809 ("vec_widen_smult_hi_<mode>"): New expander definitions.
10811 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
10814 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
10816 2017-09-26 Richard Biener <rguenther@suse.de>
10818 PR tree-optimization/82320
10819 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
10822 2017-09-25 Jeff Law <law@redhat.com>
10824 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
10825 prototype for new argument.
10826 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
10827 mostly extracted from rs6000_emit_allocate_stack.
10828 (rs6000_emit_probe_stack_range_stack_clash): New function.
10829 (rs6000_emit_allocate_stack): Call
10830 rs6000_emit_probe_stack_range_stack_clash as needed.
10831 (rs6000_emit_probe_stack_range): Add additional argument
10832 to call to gen_probe_stack_range{si,di}.
10833 (output_probe_stack_range): New.
10834 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
10835 (output_probe_stack_range_stack_clash): New.
10836 (rs6000_emit_prologue): Emit notes into dump file as requested.
10837 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
10838 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
10839 Add additional operand and pass it to output_probe_stack_range.
10841 2017-09-25 Bin Cheng <bin.cheng@arm.com>
10843 PR tree-optimization/82163
10844 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
10845 (checking_verify_loop_closed_ssa): New parameter.
10846 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
10847 (check_loop_closed_ssa_stmt): Delete.
10848 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
10849 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
10850 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
10853 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
10855 * brig-builtins.def: Treat HSAIL barrier builtins as
10856 setjmp/longjump style functions.
10858 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10860 * target.def (constant_alignment): New hook.
10861 * defaults.h (CONSTANT_ALIGNMENT): Delete.
10862 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
10863 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
10864 * doc/tm.texi: Regenerate.
10865 * targhooks.h (default_constant_alignment): Declare.
10866 (constant_alignment_word_strings): Likewise.
10867 * targhooks.c (default_constant_alignment): New function.
10868 (constant_alignment_word_strings): Likewise.
10869 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
10870 instead of CONSTANT_ALIGNMENT.
10871 * varasm.c (align_variable, get_variable_align, build_constant_desc)
10872 (force_const_mem): Likewise.
10873 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
10874 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
10875 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
10876 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10877 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
10879 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
10880 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10881 constant_alignment_word_strings.
10882 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
10883 (CONSTANT_ALIGNMENT): Likewise.
10884 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10885 (arm_constant_alignment): New function.
10886 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
10887 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10888 constant_alignment_word_strings.
10889 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
10890 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10891 constant_alignment_word_strings.
10892 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
10893 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10894 (cris_constant_alignment): New function.
10895 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
10896 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10897 (epiphany_constant_alignment): New function.
10898 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
10899 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10900 constant_alignment_word_strings.
10901 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
10902 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10903 constant_alignment_word_strings.
10904 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
10905 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10906 constant_alignment_word_strings.
10907 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
10908 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
10909 * config/i386/i386.c (ix86_constant_alignment): Make static.
10910 Use the same interface as the target hook.
10911 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10912 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
10913 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10914 constant_alignment_word_strings.
10915 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
10916 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
10917 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10918 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
10919 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10920 constant_alignment_word_strings.
10921 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
10922 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10923 constant_alignment_word_strings.
10924 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
10925 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10926 constant_alignment_word_strings.
10927 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
10928 * config/microblaze/microblaze.c (microblaze_constant_alignment):
10930 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10931 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
10932 * config/mips/mips.c (mips_constant_alignment): New function.
10933 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10934 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
10935 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
10936 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10937 (mmix_constant_alignment): Make static. Use the same interface
10938 as the target hook.
10939 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
10940 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10941 constant_alignment_word_strings.
10942 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
10943 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10944 constant_alignment_word_strings.
10945 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
10946 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10947 constant_alignment_word_strings.
10948 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
10949 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10950 (rs6000_constant_alignment): New function.
10951 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
10952 * config/riscv/riscv.c (riscv_constant_alignment): New function.
10953 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10954 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
10955 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10956 (rs6000_constant_alignment): New function.
10957 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
10958 * config/s390/s390.c (s390_constant_alignment): New function.
10959 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10960 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
10961 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10962 constant_alignment_word_strings.
10963 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
10964 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10965 (sparc_constant_alignment): New function.
10966 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
10967 * config/spu/spu.c (spu_constant_alignment): New function.
10968 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10969 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
10970 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10971 constant_alignment_word_strings.
10972 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
10973 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10974 constant_alignment_word_strings.
10975 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
10976 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10977 constant_alignment_word_strings.
10978 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
10979 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10980 (visium_constant_alignment): New function.
10981 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
10982 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10983 (xtensa_constant_alignment): New function.
10984 * system.h (CONSTANT_ALIGNMENT): Poison.
10986 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
10988 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10989 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
10990 (rs6000_builtin_valid_without_lhs): New helper function.
10991 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10992 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
10994 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10996 * target.h (vec_perm_indices): Use unsigned short rather than
10998 (auto_vec_perm_indices): Likewise.
10999 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
11000 Use unsigned int rather than unsigned char.
11001 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
11003 2017-09-25 Richard Biener <rguenther@suse.de>
11005 * cfgloop.h (sort_sibling_loops): Declare.
11006 * cfgloop.c (sort_sibling_loops_cmp): New helper.
11007 (sort_sibling_loops): New function sorting the sibling loop list
11009 * graphite.c (graphite_transform_loops): Sort sibling loops.
11011 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
11013 * target.def (vec_perm_const_ok): Change sel parameter to
11015 * optabs-query.c (can_vec_perm_p): Update accordingly.
11016 * doc/tm.texi: Regenerate.
11017 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
11018 auto_vec_perm_indices and remove separate nelt field.
11019 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
11020 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
11021 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
11022 (aarch64_expand_vec_perm_const): Update accordingly.
11023 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
11024 to vec_perm_indices.
11025 * config/arm/arm.c (expand_vec_perm_d): Change perm to
11026 auto_vec_perm_indices and remove separate nelt field.
11027 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
11028 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
11029 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
11031 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
11032 to vec_perm_indices.
11033 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
11034 sel to vec_perm_indices.
11035 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
11036 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
11037 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
11039 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
11042 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
11045 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
11046 on the FUNCTION_DECL function context if it has a DIE that is a
11049 2017-09-25 Richard Biener <rguenther@suse.de>
11051 PR tree-optimization/82285
11052 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
11055 2017-09-25 Tom de Vries <tom@codesourcery.com>
11059 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
11062 2017-09-25 Richard Biener <rguenther@suse.de>
11064 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
11065 ISL errors other than isl_error_quota happen. Dump if the
11066 schedule is the same.
11067 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
11068 errors instead of aborting inside ISL.
11070 2017-09-25 Iain Sandoe <iain@codesourcery.com>
11073 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
11074 of libgcc_eh for m64.
11075 * config/i386/darwin64.h: Likewise.
11077 2017-09-25 Richard Biener <rguenther@suse.de>
11079 PR middle-end/82144
11080 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
11081 attribute for incomplete types nor twice for complete ones.
11083 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
11086 * config/i386/i386.c (ix86_print_operand_address_as): Encode
11087 %esp as %rsp to avoid 0x67 prefix if there is no index or base
11090 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
11093 * config/i386/i386.opt (mprefer-avx256): Use
11094 ix86_target_flags variable.
11095 * config/i386/i386.c (ix86_target_string): Move
11096 -mprefer-avx256 to flag2_opts.
11098 2017-09-22 Jakub Jelinek <jakub@redhat.com>
11100 PR middle-end/35691
11101 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
11102 and x != -1 | y != -1 into (x & y) != -1.
11104 2017-09-22 Steve Ellcey <sellcey@cavium.com>
11106 * config.gcc: Add new case statement to set
11107 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
11108 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
11109 s390x-*-linux* case statements. Added aarch64 to the list of
11110 supported architectures.
11112 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11114 PR tree-optimization/82289
11115 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
11116 STMT_VINFO_RELEVANT_P.
11118 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11119 Alan Hayward <alan.hayward@arm.com>
11120 David Sherwood <david.sherwood@arm.com>
11122 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
11123 for VR_RANGE only; don't allow VR_ANTI_RANGE.
11124 (extract_range_from_binary_expr_1): Don't call
11125 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
11127 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11128 Alan Hayward <alan.hayward@arm.com>
11129 David Sherwood <david.sherwood@arm.com>
11131 * target.def (preferred_vector_alignment): New hook.
11132 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
11134 * doc/tm.texi: Regenerate.
11135 * targhooks.h (default_preferred_vector_alignment): Declare.
11136 * targhooks.c (default_preferred_vector_alignment): New function.
11137 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
11139 (DR_TARGET_ALIGNMENT): New macro.
11140 (aligned_access_p): Update commentary.
11141 (vect_known_alignment_in_bytes): New function.
11142 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
11144 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
11145 Calculate the misalignment based on the target alignment rather than
11147 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
11148 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
11149 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
11150 the target alignment, rather than masking the element misalignment
11151 with the number of elements in a vector. Also use the target
11152 alignment when calculating the maximum number of peels.
11153 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
11154 instead of TYPE_ALIGN_UNIT.
11155 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
11156 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
11157 (vect_create_addr_base_for_vector_ref): Update call accordingly.
11158 (vect_create_data_ref_ptr): Likewise.
11159 (vect_setup_realignment): Realign by ANDing with
11160 -DR_TARGET_MISALIGNMENT.
11161 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
11162 the number of peels based on DR_TARGET_ALIGNMENT.
11163 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
11164 with the guaranteed alignment boundary when deciding whether
11166 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
11167 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
11168 (ensure_base_align): Remove stmt_info parameter. Get the
11169 target base alignment from DR_TARGET_ALIGNMENT.
11170 (vectorizable_store): Update call accordingly. Interpret
11171 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
11173 (vectorizable_load): Likewise.
11175 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11176 Alan Hayward <alan.hayward@arm.com>
11177 David Sherwood <david.sherwood@arm.com>
11179 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
11180 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
11181 (vect_enhance_data_refs_alignment): Likewise.
11183 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11185 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
11186 error. Only quit immediately if parsing is complete.
11187 (BEGIN): Initialize fatal_err and parse_done.
11188 (begin fpu, end fpu): Check number of arguments.
11189 (begin arch, end arch): Likewise.
11190 (begin cpu, end cpu): Likewise.
11191 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
11192 (optalias): Likewise.
11194 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11196 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
11197 * config/arm/arm-isa.h: Delete. Move definitions to ...
11198 * arm-cpus.in: ... here. Use new feature and fgroup values.
11199 * config/arm/arm.c (arm_option_override): Use lower case for feature
11201 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
11202 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
11203 * config/arm/parsecpu.awk (END): Add new command 'isa'.
11205 (print_isa_bits_for): New function.
11206 (gen_isa): New function.
11207 (gen_comm_data): Use print_isa_bits_for.
11208 (define feature): New keyword.
11209 (define fgroup): New keyword.
11210 * config/arm/t-arm (TM_H): Remove.
11211 (GTM_H): Add arm-isa.h.
11212 (arm-isa.h): Add rule to generate file.
11213 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
11214 case for feature bit names.
11216 2017-09-22 Richard Biener <rguenther@suse.de>
11218 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
11220 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
11221 print of no dependency loops ...
11222 * graphite.c (graphite_transform_loops): ... here.
11223 (canonicalize_loop_closed_ssa_form): Work from inner to outer
11225 (same_close_phi_node, remove_duplicate_close_phi,
11226 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
11227 (canonicalize_loop_closed_ssa): ... here and simplify.
11228 * graphite-optimize-isl.c: Include tree-vectorizer.h.
11229 (optimize_isl): Use dump_printf_loc to tell when we stopped
11230 optimizing because of an ISL timeout.
11232 2017-09-22 Richard Biener <rguenther@suse.de>
11234 PR tree-optimization/82291
11235 * tree-if-conv.c (predicate_mem_writes): Make sure to
11236 remove writes in blocks predicated with false.
11238 2017-09-22 Richard Biener <rguenther@suse.de>
11240 * sese.c: Include cfganal.h.
11241 (if_region_set_false_region): Remove.
11242 (create_if_region_on_edge): Likewise.
11243 (move_sese_in_condition): Re-implement without destroying
11246 2017-09-22 Richard Biener <rguenther@suse.de>
11248 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
11249 Verify both BBs contain loop PHI nodes before dispatching to
11250 copy_loop_phi_args.
11251 (graphite_regenerate_ast_isl): Do not recompute dominators,
11252 do not verify three times. Restructure for clarity.
11253 * graphite-scop-detection.c (same_close_phi_node,
11254 remove_duplicate_close_phi, make_close_phi_nodes_unique,
11255 defined_in_loop_p, canonicalize_loop_closed_ssa,
11256 canonicalize_loop_closed_ssa_form): Simplify, remove excess
11257 checking and SSA rewrite, move to ...
11258 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
11259 (graphite_initialize): Do not pass in ctx, do not reset the
11260 SCEV cache, compute only dominators.
11261 (graphite_transform_loops): Allocate ISL ctx after
11262 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
11263 Maintain post-dominators only around build_scops.
11264 * sese.c (if_region_set_false_region): Make static. Free
11265 and recompute dominators.
11266 (move_sese_in_condition): Assert we don't get called with
11267 post-dominators computed.
11268 * sese.h (if_region_set_false_region): Remove.
11270 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
11272 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
11273 mode attribute for TARGET_AVX512VL.
11275 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
11277 * config/i386/i386.opt (mprefer-avx256): New option.
11278 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
11280 (ix86_preferred_simd_mode): Return 256-bit AVX modes
11281 for TARGET_PREFER_AVX256.
11282 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
11284 2017-09-21 Jeff Law <law@redhat.com>
11286 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
11287 Fix dump output if the only stack space is for pushed registers.
11289 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11291 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
11294 2017-09-21 Martin Sebor <msebor@redhat.com>
11297 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
11298 code (in C++) or code that triggers warnings.
11300 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
11302 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
11304 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
11306 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
11307 * sched-rgn.c: Ditto.
11308 * sel-sched-ir.c: Ditto.
11310 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
11312 * toplev.h (set_random_seed): Adjust return type.
11313 * toplev.c (init_local_tick): Move eager initialization of random_seed
11314 to get_random_seed. Adjust comment.
11315 (init_random_seed): Inline to get_random_seed, delete.
11316 (get_random_seed): Initialize random_seed lazily.
11317 (set_random_seed): Do not return previous value.
11318 (print_switch_value): Do not call get_random_seed.
11320 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
11322 * cgraph.c (delete_function_version): New, broken out from...
11323 (cgraph_node::delete_function_version): ...here. Rename to
11324 cgraph_node::delete_function_version_by_decl. Update all uses.
11325 (cgraph_node::remove): Call delete_function_version.
11327 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11330 * tree-inline.c (expand_call_inline): Emit clobber stmts for
11331 VAR_DECLs to which addressable non-volatile parameters are mapped
11332 and for id->retvar after the return value assignment. Clear
11333 id->retval and id->retbnd after inlining.
11335 2017-09-21 Richard Biener <rguenther@suse.de>
11337 PR tree-optimization/82276
11338 PR tree-optimization/82244
11339 * tree-vrp.c (build_assert_expr_for): Set
11340 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
11342 (remove_range_assertions): Revert earlier change.
11344 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
11347 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
11349 2017-09-21 Richard Biener <rguenther@suse.de>
11351 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11352 Restore valid IL after code generation errors.
11353 * graphite.c (graphite_transform_loops): Diagnose code
11354 generation issues as MSG_MISSED_OPTIMIZATION and continue
11355 with processing SCOPs.
11357 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11358 Alan Hayward <alan.hayward@arm.com>
11359 David Sherwood <david.sherwood@arm.com>
11361 * calls.c (compute_argument_addresses): Use simplify_gen_binary
11362 rather than choosing between plus_constant and gen_rtx_<CODE>.
11363 * expr.c (emit_push_insn): Likewise.
11364 (expand_expr_real_2): Likewise.
11366 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11367 Alan Hayward <alan.hayward@arm.com>
11368 David Sherwood <david.sherwood@arm.com>
11370 * loop-unroll.c (split_iv): Call copy_rtx on the step.
11372 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11373 Alan Hayward <alan.hayward@arm.com>
11374 David Sherwood <david.sherwood@arm.com>
11376 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
11377 calling tree_to_uhwi.
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 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
11384 INTEGER_CST rather than a negative test for ADDR_EXPR.
11386 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11387 Alan Hayward <alan.hayward@arm.com>
11388 David Sherwood <david.sherwood@arm.com>
11390 * tree-vrp.c (extract_range_from_binary_expr_1): Check
11391 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
11393 2017-09-21 Richard Biener <rguenther@suse.de>
11395 PR tree-optimization/71351
11396 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
11397 graphite_create_new_loop_guard): Remove, fold remaining parts
11399 (translate_isl_ast_node_for): ... here and simplify.
11401 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11404 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
11405 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
11406 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
11407 alternative always use QI mode, for -Os imov (=R,R) alternative
11408 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
11411 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11412 Jeff Law <law@redhat.com>
11414 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
11415 (allocate_stack_space): New function, partially extracted from
11416 s390_emit_prologue.
11417 (s390_emit_prologue): Track offset to most recent stack probe.
11418 Code to allocate space moved into allocate_stack_space.
11419 Dump actions when no stack is allocated.
11420 (s390_prologue_plus_offset): New function.
11421 (s390_emit_stack_probe): Likewise.
11423 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
11425 * common.opt (Wa, Wl, Wp, g, gz=): Add
11427 (gno-column-info): Remove.
11428 (gcolumn-info): Drop RejectNegative.
11429 (gno-): New prefix.
11430 (gno-record-gcc-switches): Remove.
11431 (grecord-gcc-switches): Drop RejectNegative.
11432 (gno-split-dwarf): Remove.
11433 (gsplit-dwarf): Drop RejectNegative.
11434 (gno-strict-dwarf): Remove.
11435 (gstrict-dwarf): Drop RejectNegative.
11436 * config/darwin.opt (gfull, gused): Add RejectNegative.
11437 * dwarf2out.c (gen_producer_string): Drop
11438 gno-record-gcc-switches handler.
11439 * optc-gen.awk: Add g to prefixes with negative forms.
11440 * opts-common.c (remapping_prefix_p): New.
11441 (find_opt): Check it.
11442 (generate_canonical_option): Test g prefix.
11443 (option_map): Add -gno- mapping.
11444 (add_misspelling_candidates): Check remapping_prefix_p.
11446 2017-09-20 Jeff Law <law@redhat.com>
11448 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
11449 thinko in stack clash protection support.
11451 * explow.c (compute_stack_clash_protection_loop_data): Use
11452 CONST_INT_P instead of explicit test. Verify object is a
11453 CONST_INT_P before looking at INTVAL.
11454 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
11455 instead of explicit test.
11457 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
11460 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
11461 address instead of to r1 and r11.
11463 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
11465 * config.gcc: Support "knm".
11466 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
11467 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
11469 * config/i386/i386.c (m_KNM): Define.
11470 (processor_target_table): Add "knm".
11472 (ix86_option_override_internal): Add "knm".
11473 (ix86_issue_rate): Add PROCESSOR_KNM.
11474 (ix86_adjust_cost): Ditto.
11475 (ia32_multipass_dfa_lookahead): Ditto.
11476 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
11477 (fold_builtin_cpu): Add M_INTEL_KNM.
11478 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
11479 (processor_type): Add PROCESSOR_KNM.
11480 * config/i386/x86-tune.def: Add m_KNM.
11481 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
11483 2017-09-20 Richard Biener <rguenther@suse.de>
11485 PR tree-optimization/80213
11486 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
11487 are allowed in empty BBs as well.
11488 (canonicalize_loop_closed_ssa): Also look for other complex
11490 (scop_detection::get_sese): Include the loop-closed PHI block
11492 (scop_detection::merge_sese): Remove code adding extra blocks.
11493 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
11494 (build_scops): Assert the final returned scop is invalid.
11496 2017-09-20 Richard Biener <rguenther@suse.de>
11498 PR tree-optimization/82264
11499 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
11501 (vn_phi_lookup): Likewise.
11502 (vn_phi_insert): Likewise.
11504 2017-09-20 Jakub Jelinek <jakub@redhat.com>
11506 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
11507 that fits into uhwi or shwi, add DW_AT_const_value regardless
11508 of early_dwarf without going through RTL, using add_AT_unsigned
11511 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
11512 (DEBUG_LTO_ABBREV_SECTION): Likewise.
11513 (DEBUG_LTO_MACINFO_SECTION): Likewise.
11514 (DEBUG_MACRO_SECTION): Likewise.
11515 (DEBUG_LTO_MACRO_SECTION): Likewise.
11516 (DEBUG_STR_DWO_SECTION): Likewise.
11517 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
11518 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
11519 (DEBUG_LTO_DWO_LINE_SECTION): Define.
11520 (DEBUG_LTO_LINE_STR_SECTION): Define.
11521 (init_sections_and_labels): Initialize debug_line_str_section
11522 variable. Initialize debug_loc_section for -gdwarf-5 to
11523 DEBUG_LOCLISTS_SECTION. Formatting fixes.
11525 2017-09-20 Richard Biener <rguenther@suse.de>
11527 * graphite-sese-to-poly.c (extract_affine): Properly handle
11528 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
11530 2017-09-20 Richard Biener <rguenther@suse.de>
11532 PR tree-optimization/81373
11533 * graphite-scop-detection.c (build_cross_bb_scalars_def):
11534 Force SESE live-out defs to be handled even if they are
11537 2017-09-19 Jeff Law <law@redhat.com>
11539 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
11540 nothing for stack adjustments with REG_STACK_CHECK.
11541 * sched-deps.c (parse_add_or_inc): Reject insns with
11542 REG_STACK_CHECK from dependency breaking.
11543 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
11544 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
11545 * reg-notes.def (STACK_CHECK): New note.
11547 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
11548 (ix86_expand_prologue): Dump stack clash info as needed.
11549 Call ix86_adjust_stack_and_probe_stack_clash as needed.
11551 * function.c (dump_stack_clash_frame_info): New function.
11552 * function.h (dump_stack_clash_frame_info): Prototype.
11553 (enum stack_clash_probes): New enum.
11555 * config/alpha/alpha.c (alpha_expand_prologue): Also check
11556 flag_stack_clash_protection.
11557 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
11558 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
11559 (arm_frame_pointer_required): Likewise.
11560 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
11561 (ia64_expand_prologue): Likewise.
11562 * config/mips/mips.c (mips_expand_prologue): Likewise.
11563 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
11564 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11565 (sparc_flat_expand_prologue): Likewise.
11566 * config/spu/spu.c (spu_expand_prologue): Likewise.
11568 * explow.c: Include "params.h".
11569 (anti_adjust_stack_and_probe_stack_clash): New function.
11570 (get_stack_check_protect): Likewise.
11571 (compute_stack_clash_protection_loop_data): Likewise.
11572 (emit_stack_clash_protection_loop_start): Likewise.
11573 (emit_stack_clash_protection_loop_end): Likewise.
11574 (allocate_dynamic_stack_space): Use get_stack_check_protect.
11575 Use anti_adjust_stack_and_probe_stack_clash.
11576 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
11577 (emit_stack_clash_protection_loop_start): Likewise.
11578 (emit_stack_clash_protection_loop_end): Likewise.
11579 * rtl.h (get_stack_check_protect): Prototype.
11580 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
11581 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
11582 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
11584 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
11586 * doc/tm.texi: Rebuilt.
11587 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
11588 get_stack_check_protect.
11589 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
11590 * config/arm/arm.c (arm_expand_prologue): Likewise.
11591 (arm_frame_pointer_required): Likewise.
11592 * config/i386/i386.c (ix86_expand_prologue): Likewise.
11593 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
11594 * config/mips/mips.c (mips_expand_prologue): Likewise.
11595 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
11596 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
11597 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11598 (sparc_flat_expand_prologue): Likewise.
11600 * common.opt (-fstack-clash-protection): New option.
11601 * flag-types.h (enum stack_check_type): Note difference between
11602 -fstack-check= and -fstack-clash-protection.
11603 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
11604 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
11605 * toplev.c (process_options): Issue warnings/errors for cases
11606 not handled with -fstack-clash-protection.
11607 * doc/invoke.texi (-fstack-clash-protection): Document new option.
11608 (-fstack-check): Note additional problem with -fstack-check=generic.
11609 Note that -fstack-check is primarily for Ada and refer users
11610 to -fstack-clash-protection for stack-clash-protection.
11611 Document new params for stack clash protection.
11613 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
11615 * config/i386/i386.c (ix86_split_long_move): Do not handle
11616 address used for LEA in a special way.
11618 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
11620 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
11622 2017-09-19 Martin Sebor <msebor@redhat.com>
11625 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
11626 of incompatible types.
11628 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
11630 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11631 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
11632 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11633 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
11635 2017-09-19 Richard Biener <rguenther@suse.de>
11637 PR tree-optimization/82244
11638 * tree-vrp.c (remove_range_assertions): Do not propagate
11639 a constant to abnormals but replace the assert with a copy.
11641 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
11643 PR rtl-optimization/57878
11644 PR rtl-optimization/68988
11645 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
11646 avoidance test involving non_reload_pseudos. Move frequency test
11647 below the general fragmentation avoidance test.
11649 2017-09-19 Richard Biener <rguenther@suse.de>
11651 PR tree-optimization/69728
11652 * graphite-sese-to-poly.c (schedule_error): New global.
11653 (add_loop_schedule): Handle empty domain by failing the
11655 (build_original_schedule): Handle schedule_error.
11657 2017-09-19 Richard Biener <rguenther@suse.de>
11659 * graphite-scop-detection.c (scop_detection::can_represent_loop):
11660 Do not iterate to sibling loops but only to siblings of inner
11663 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
11666 * config/m68k/m68k.md (moveq feeding equality comparison): Check
11667 that the registers are different.
11669 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
11671 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
11672 to processor_model and "amdfam17h" to arch_names_table.
11673 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
11675 2017-09-18 Jakub Jelinek <jakub@redhat.com>
11678 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
11680 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11681 Alan Hayward <alan.hayward@arm.com>
11682 David Sherwood <david.sherwood@arm.com>
11684 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
11686 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
11688 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
11689 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
11690 vect_schedule_slp_instance.
11691 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
11692 Update call to vect_slp_analyze_node_operations. Simplify return
11694 (vect_slp_analyze_bb_1): Update call accordingly.
11695 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
11696 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
11697 (vect_schedule_slp): Update call accordingly.
11699 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11700 Alan Hayward <alan.hayward@arm.com>
11701 David Sherwood <david.sherwood@arm.com>
11703 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
11704 with types that aren't in fact scalar.
11706 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11708 * tree-vect-slp.c (vect_record_max_nunits): New function,
11710 (vect_build_slp_tree_1): ...here.
11711 (vect_build_slp_tree_2): Call it for phis too.
11713 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11715 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
11716 to vect_get_vec_def_for_operand when getting the mask operand.
11718 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11719 Alan Hayward <alan.hayward@arm.com>
11720 David Sherwood <david.sherwood@arm.com>
11722 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
11725 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11726 Alan Hayward <alan.hayward@arm.com>
11727 David Sherwood <david.sherwood@arm.com>
11729 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
11730 calculation for vector booleans.
11732 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11733 Alan Hayward <alan.hayward@arm.com>
11734 David Sherwood <david.sherwood@arm.com>
11736 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
11738 (vect_transform_stmt): ...here.
11739 (vect_analyze_stmt): Use it instead of calling
11740 vectorizable_live_operation directly.
11742 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
11744 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
11745 non-SIMT targets in acc vector loops.
11747 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
11749 * configure.ac: Add arc and check if assembler supports gdwarf2.
11750 * configure: Regenerate.
11752 2017-09-18 Richard Biener <rguenther@suse.de>
11754 PR tree-optimization/82220
11755 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
11756 epilogue niters from the min_profitable_iters compute.
11758 2017-09-18 Jakub Jelinek <jakub@redhat.com>
11761 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
11762 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
11763 (ix86_init_pic_reg): Revert 2017-09-01 changes.
11765 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
11768 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
11769 switching to a new text section.
11771 2017-09-18 Richard Biener <rguenther@suse.de>
11773 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
11775 (build_alias_set): Reject aliases with no access function.
11777 2017-09-18 Richard Biener <rguenther@suse.de>
11779 PR tree-optimization/79622
11780 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
11782 (build_cross_bb_scalars_use): Likewise.
11784 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
11786 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
11788 2017-09-18 Alan Modra <amodra@gmail.com>
11791 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
11792 stack_pointer_rtx for count 0. Update comments. Break up
11793 large rtl expression.
11795 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
11797 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
11798 Increase to 20 bytes.
11799 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
11800 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
11801 or avx version of the stub.
11803 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
11806 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
11807 compute the minimum stack alignment. Also update preferred stack
11808 boundary for leaf functions.
11810 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
11812 PR tree-optimization/82228
11813 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
11816 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
11818 * common/config/nds32/nds32-common.c
11819 (nds32_option_optimization_table): Refine formatting.
11820 (nds32_option_optimization_table): Use -fsched-pressure and
11821 -fomit-frame-pointer for specific optimization level.
11823 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
11825 * config/nds32/nds32.c: Refine formatting and comments.
11826 * config/nds32/nds32.h: Likewise.
11827 * config/nds32/nds32.md: Likewise.
11828 * config/nds32/nds32-cost.c: Likewise.
11829 * config/nds32/nds32-isr.c: Likewise.
11830 * config/nds32/nds32-md-auxiliary.c: Likewise.
11831 * config/nds32/nds32-multiple.md: Likewise.
11832 * config/nds32/nds32-predicates.c: Likewise.
11834 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
11835 Jakub Jelinek <jakub@redhat.com>
11837 Add support for -std=c++2a.
11838 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
11840 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
11842 2017-09-15 Steve Ellcey <sellcey@cavium.com>
11845 * doc/extend.texi (Common Function Attributes): Add
11846 references to ARM, AArch64, and S/390 specific attributes.
11847 (Function Specific Option Pragmas): Add AArch64 and S/390
11848 to list of back ends that support the target pragma.
11850 2017-09-15 Nathan Sidwell <nathan@acm.org>
11852 * doc/standards.texi: Fix C++17 description. Update URLs for
11855 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
11857 * common.opt (Wcast-align=strict): New warning option.
11858 * doc/invoke.texi: Document -Wcast-align=strict.
11860 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
11862 * cgraph.h (cgraph_thunk_info): Add comments.
11863 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
11864 assert for VIRTUAL_* arguments stricter.
11866 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
11868 PR tree-optimization/71026
11869 * match.pd: Move RDIV patterns from fold-const.c
11870 * fold-const.c (distribute_real_division): Removed.
11871 (fold_binary_loc): Remove calls to distribute_real_divison.
11873 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11875 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
11876 c++1z and gnu++1z as deprecated. Change other references to
11877 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
11878 Change -Wc++1z-compat to -Wc++17-compat.
11879 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
11880 * dwarf2out.c (highest_c_language): Handle C++17.
11881 (gen_compile_unit_die): Likewise.
11883 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11885 PR rtl-optimization/82192
11886 * combine.c (make_extraction): Don't look through non-paradoxical
11887 SUBREGs or TRUNCATE if pos + len is or might be bigger than
11890 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11891 Alan Hayward <alan.hayward@arm.com>
11892 David Sherwood <david.sherwood@arm.com>
11894 * target.def (function_arg_offset): New hook.
11895 * targhooks.h (default_function_arg_offset): Declare.
11896 * targhooks.c (default_function_arg_offset): New function.
11897 * function.c (locate_and_pad_parm): Use
11898 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
11899 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
11900 (TARGET_FUNCTION_ARG_OFFSET): ...this.
11901 * doc/tm.texi: Regenerate.
11902 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
11903 * config/spu/spu.c (spu_function_arg_offset): New function.
11904 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
11905 * system.h (FUNCTION_ARG_OFFSET): Poison.
11907 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11908 Alan Hayard <alan.hayward@arm.com>
11909 David Sherwood <david.sherwood@arm.com>
11911 * target.def (truly_noop_truncation): New hook.
11912 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
11913 than TRULY_NOOP_TRUNCATION.
11914 * hooks.h (hook_bool_uint_uint_true): Declare.
11915 * hooks.c (hook_bool_uint_uint_true): New function.
11916 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
11917 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
11918 * doc/tm.texi: Regenerate.
11919 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
11920 rather than TRULY_NOOP_TRUNCATION in comments.
11921 (simplify_comparison): Likewise.
11922 (record_truncated_value): Likewise.
11923 * expmed.c (extract_bit_field_1): Likewise.
11924 (extract_split_bit_field): Likewise.
11925 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
11926 instead of TRULY_NOOP_TRUNCATION.
11927 * function.c (assign_parm_setup_block): Likewise.
11928 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
11929 * rtlhooks.c: Include target.h.
11930 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
11931 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
11932 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
11933 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
11934 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
11935 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
11936 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
11937 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
11938 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
11939 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
11940 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
11941 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
11942 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
11943 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
11944 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
11945 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
11946 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
11947 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
11948 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
11949 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
11950 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
11951 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
11952 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
11953 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
11954 * config/mips/mips.c (mips_truly_noop_truncation): New function.
11955 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11956 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11957 rather than TRULY_NOOP_TRUNCATION in comments.
11958 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
11959 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
11960 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
11961 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
11962 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
11963 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
11964 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
11965 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
11966 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
11967 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
11968 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
11969 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11970 rather than TRULY_NOOP_TRUNCATION in comments.
11971 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
11972 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
11973 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
11974 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
11975 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
11976 TRULY_NOOP_TRUNCATION condition.
11977 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
11978 (TRULY_NOOP_TRUNCATION): Delete.
11979 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
11980 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
11981 * config/spu/spu.c (spu_truly_noop_truncation): New function.
11982 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11983 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
11984 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
11985 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
11986 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11987 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11988 rather than TRULY_NOOP_TRUNCATION in comments.
11989 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
11990 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
11991 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
11992 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
11993 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
11994 * system.h (TRULY_NOOP_TRUNCATION): Poison.
11996 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
11999 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
12000 (*cmp_ior): Likewise.
12001 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
12002 (*ior_scc_scc_cmp): Likewise.
12003 (*and_scc_scc): Likewise.
12004 (*and_scc_scc_cmp): Likewise.
12006 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
12007 Alan Hayard <alan.hayward@arm.com>
12008 David Sherwood <david.sherwood@arm.com>
12010 * target.def (can_change_mode_class): New hook.
12011 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
12012 (hard_regno_nregs): Likewise.
12013 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
12014 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
12015 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
12016 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
12017 (LOAD_EXTEND_OP): Update accordingly.
12018 * doc/tm.texi: Regenerate.
12019 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
12020 CANNOT_CHANGE_MODE_CLASS.
12021 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
12022 (REG_CAN_CHANGE_MODE_P): ...this new macro.
12023 * combine.c (simplify_set): Update accordingly.
12024 * emit-rtl.c (validate_subreg): Likewise.
12025 * recog.c (general_operand): Likewise.
12026 * regcprop.c (mode_change_ok): Likewise.
12027 * reload1.c (choose_reload_regs): Likewise.
12028 (inherit_piecemeal_p): Likewise.
12029 * rtlanal.c (simplify_subreg_regno): Likewise.
12030 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
12031 instead of CANNOT_CHANGE_MODE_CLASS.
12032 (reload_cse_simplify_operands): Likewise.
12033 * reload.c (push_reload): Use targetm.can_change_mode_class
12034 instead of CANNOT_CHANGE_MODE_CLASS.
12035 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
12036 REG_CANNOT_CHANGE_MODE_P.
12037 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12038 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
12039 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12040 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12041 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12042 (arm_can_change_mode_class): New function.
12043 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
12044 than CANNOT_CHANGE_MODE_CLASS in comments.
12045 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12046 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
12047 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
12048 (ix86_can_change_mode_class): ...this new function, inverting the
12049 sense of the return value.
12050 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12051 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12052 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12053 (ia64_can_change_mode_class): New function.
12054 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12055 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
12056 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
12057 (m32c_can_change_mode_class): ...this new function, inverting the
12058 sense of the return value.
12059 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12060 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12061 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
12062 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
12063 (mips_can_change_mode_class): ...this new function, inverting the
12064 sense of the return value.
12065 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12066 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12067 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12068 (msp430_can_change_mode_class): New function.
12069 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12070 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
12071 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12072 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12073 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12074 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
12075 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12076 (pa_cannot_change_mode_class): Replace with...
12077 (pa_can_change_mode_class): ...this new function, inverting the
12078 sense of the return value.
12079 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
12080 than CANNOT_CHANGE_MODE_CLASS in comments.
12081 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12082 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
12083 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12084 (pdp11_cannot_change_mode_class): Replace with...
12085 (pdp11_can_change_mode_class): ...this new function, inverting the
12086 sense of the return value.
12087 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12088 * config/powerpcspe/powerpcspe-protos.h
12089 (rs6000_cannot_change_mode_class_ptr): Delete.
12090 * config/powerpcspe/powerpcspe.c
12091 (rs6000_cannot_change_mode_class_ptr): Delete.
12092 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12093 (rs6000_option_override_internal): Assign to
12094 targetm.can_change_mode_class instead of
12095 rs6000_cannot_change_mode_class_ptr.
12096 (rs6000_cannot_change_mode_class): Replace with...
12097 (rs6000_can_change_mode_class): ...this new function, inverting the
12098 sense of the return value.
12099 (rs6000_debug_cannot_change_mode_class): Replace with...
12100 (rs6000_debug_can_change_mode_class): ...this new function.
12101 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12102 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
12103 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12104 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12105 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
12107 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
12108 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12109 (rs6000_option_override_internal): Assign to
12110 targetm.can_change_mode_class instead of
12111 rs6000_cannot_change_mode_class_ptr.
12112 (rs6000_cannot_change_mode_class): Replace with...
12113 (rs6000_can_change_mode_class): ...this new function, inverting the
12114 sense of the return value.
12115 (rs6000_debug_cannot_change_mode_class): Replace with...
12116 (rs6000_debug_can_change_mode_class): ...this new function.
12117 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12118 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
12119 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
12120 (s390_can_change_mode_class): ...this new function, inverting the
12121 sense of the return value.
12122 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12123 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12124 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
12125 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12126 (sh_cannot_change_mode_class): Replace with...
12127 (sh_can_change_mode_class): ...this new function, inverting the
12128 sense of the return value.
12129 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12130 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12131 (sparc_can_change_mode_class): New function.
12132 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12133 * config/spu/spu.c (spu_can_change_mode_class): New function.
12134 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12135 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12136 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12137 (visium_can_change_mode_class): New function.
12138 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
12140 2017-09-15 Richard Biener <rguenther@suse.de>
12142 PR tree-optimization/82217
12143 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
12144 but not undefined case.
12146 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12149 * postreload.c (reload_cse_simplify_operands): Skip
12150 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
12152 2017-09-15 Richard Biener <rguenther@suse.de>
12154 PR tree-optimization/68823
12155 * graphite-scop-detection.c (build_alias_set): If we have a
12156 possible dependence check whether we can handle them by just
12157 looking at the DRs DR_ACCESS_FNs.
12158 (build_scops): If build_alias_set fails, fail the SCOP.
12160 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
12162 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
12163 to support float128 built-in functions that require the ISA 3.0
12165 (BU_FLOAT128_3_HW): Likewise.
12166 (SQRTF128): Add support for the IEEE 128-bit square root and fma
12167 built-in functions.
12168 (FMAF128): Likewise.
12170 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
12171 support for built-in functions that need the ISA 3.0 IEEE 128-bit
12172 floating point instructions.
12173 (rs6000_invalid_builtin): Likewise.
12174 (rs6000_builtin_mask_names): Likewise.
12175 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
12176 (RS6000_BTM_FLOAT128_HW): Likewise.
12177 (RS6000_BTM_COMMON): Likewise.
12178 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
12180 * doc/extend.texi (RS/6000 built-in functions): Document the
12181 IEEE 128-bit floating point square root and fused multiply-add
12182 built-in functions.
12184 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
12186 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
12187 reg (r2) isn't in the set of registers defined in the prologue.
12189 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12190 Alan Hayward <alan.hayward@arm.com>
12191 David Sherwood <david.sherwood@arm.com>
12193 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
12194 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
12195 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
12196 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
12197 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
12199 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
12200 max_vectorization_factor.
12201 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
12203 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12204 Alan Hayward <alan.hayward@arm.com>
12205 David Sherwood <david.sherwood@arm.com>
12207 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
12208 (vect_worthwhile_without_simd_p): Declare.
12209 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
12210 (vectorizable_reduction): Use it.
12211 * tree-vect-stmts.c (vectorizable_shift): Likewise.
12212 (vectorizable_operation): Likewise.
12214 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12215 Alan Hayward <alan.hayward@arm.com>
12216 David Sherwood <david.sherwood@arm.com>
12218 * tree-vectorizer.h (vect_get_num_copies): New function.
12219 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
12220 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12221 (vectorizable_induction): Likewise.
12222 (vectorizable_live_operation): Likewise.
12223 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12224 (vectorizable_bswap): Likewise.
12225 (vectorizable_call): Likewise.
12226 (vectorizable_conversion): Likewise.
12227 (vectorizable_assignment): Likewise.
12228 (vectorizable_shift): Likewise.
12229 (vectorizable_operation): Likewise.
12230 (vectorizable_store): Likewise.
12231 (vectorizable_load): Likewise.
12232 (vectorizable_condition): Likewise.
12233 (vectorizable_comparison): Likewise.
12234 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
12236 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12237 Alan Hayward <alan.hayward@arm.com>
12238 David Sherwood <david.sherwood@arm.com>
12240 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
12241 of vect_init_vector.
12243 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12244 Alan Hayward <alan.hayward@arm.com>
12245 David Sherwood <david.sherwood@arm.com>
12247 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
12248 an inline wrapper that provides a location.
12249 (gimple_build_vector): Likewise.
12250 * gimple-fold.c (gimple_build_vector_from_val): New function.
12251 (gimple_build_vector): Likewise.
12252 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
12253 functions to build the initial value. Always return a gimple value.
12254 (get_initial_defs_for_reduction): Likewise. Only compute
12256 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
12257 vect_init_vector on the results from get_initial_def(s)_for_reduction.
12258 (vectorizable_induction): Use gimple_build_vector rather than
12261 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12262 Alan Hayward <alan.hayward@arm.com>
12263 David Sherwood <david.sherwood@arm.com>
12265 * target.h (vec_perm_indices): New typedef.
12266 (auto_vec_perm_indices): Likewise.
12267 * optabs-query.h: Include target.h
12268 (can_vec_perm_p): Take a vec_perm_indices *.
12269 * optabs-query.c (can_vec_perm_p): Likewise.
12270 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
12271 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12272 * tree-vect-generic.c (lower_vec_perm): Likewise.
12273 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
12274 (vect_grouped_load_supported): Likewise.
12275 (vect_shift_permute_load_chain): Likewise.
12276 (vect_permute_store_chain): Use auto_vec_perm_indices.
12277 (vect_permute_load_chain): Likewise.
12278 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
12279 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
12280 Update uses of can_vec_perm_p.
12281 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
12282 mode with a number of elements. Take a vec_perm_indices *.
12283 (vect_create_epilog_for_reduction): Update accordingly.
12284 Use auto_vec_perm_indices.
12285 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
12286 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
12287 (vect_transform_slp_perm_load): Likewise.
12288 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
12289 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
12290 (vect_gen_perm_mask_checked): Likewise.
12291 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
12292 (vect_gen_perm_mask_checked): Likewise.
12293 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
12294 (vectorizable_store): Likewise.
12295 (vectorizable_load): Likewise.
12296 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
12297 (vectorizable_bswap): Likewise.
12299 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12300 Alan Hayward <alan.hayward@arm.com>
12301 David Sherwood <david.sherwood@arm.com>
12303 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
12304 * tree.c (build_vector): Likewise.
12305 (build_vector_from_ctor): Update accordingly.
12306 (build_vector_from_val): Likewise.
12307 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
12308 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12309 * tree-vect-generic.c (add_rshift): Likewise.
12310 (expand_vector_divmod): Likewise.
12311 (optimize_vector_constructor): Likewise.
12312 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
12313 (vect_transform_slp_perm_load): Likewise.
12314 (vect_schedule_slp_instance): Likewise.
12315 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
12316 (vectorizable_call): Likewise.
12317 (vect_gen_perm_mask_any): Likewise. Add elements in order.
12318 * expmed.c (make_tree): Likewise.
12319 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
12320 a vector passed to build_vector.
12321 (fold_convert_const): Likewise.
12322 (exact_inverse): Likewise.
12323 (fold_ternary_loc): Likewise.
12324 (fold_relational_const): Likewise.
12325 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
12326 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
12327 (const_unop): Likewise. Store the reduction accumulator in a
12328 variable rather than an array.
12329 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
12330 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
12331 the new vector, rather than constructing it after the input arrays.
12332 (native_interpret_vector): Use auto_vec<tree> when building
12333 a vector passed to build_vector. Add elements in order.
12334 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
12335 auto_vec<tree> when building a vector passed to build_vector.
12336 (vect_create_epilog_for_reduction): Likewise.
12337 (vectorizable_induction): Likewise.
12338 (get_initial_def_for_reduction): Likewise. Fix indentation of
12340 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
12342 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
12343 passed to build_vector.
12345 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12346 Alan Hayward <alan.hayward@arm.com>
12347 David Sherwood <david.sherwood@arm.com>
12349 * tree-core.h (tree_base::u): Add an "nelts" field.
12350 (tree_vector): Use VECTOR_CST_NELTS as the length.
12351 * tree.c (tree_size): Likewise.
12352 (make_vector): Initialize VECTOR_CST_NELTS.
12353 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
12354 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
12355 TYPE_VECTOR_SUBPARTS.
12356 * expr.c (const_vector_mask_from_tree): Consistently use "units"
12357 as the number of units, setting it from VECTOR_CST_NELTS.
12358 (const_vector_from_tree): Likewise.
12359 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
12360 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
12361 (fold_negate_expr_1): Likewise.
12362 (fold_convert_const): Likewise.
12363 (const_binop): Likewise. Differentiate the number of output and
12365 (const_unop): Likewise.
12366 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
12367 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
12368 in cases that did the opposite.
12370 2017-09-14 Richard Biener <rguenther@suse.de>
12372 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
12375 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
12377 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
12379 2017-09-14 Jakub Jelinek <jakub@redhat.com>
12382 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
12383 if and where to split a bb, except for splitting before debug insn
12384 sequences followed by non-label real insn. Delete debug insns
12385 in between basic blocks.
12387 * combine.c (make_compound_operation_int): Formatting fixes.
12389 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
12390 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12391 * config/netbsd.h (LINK_EH_SPEC): Likewise.
12392 * config/sol2.h (LINK_EH_SPEC): Likewise.
12393 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12394 * config/s390/linux.h (LINK_SPEC): Likewise.
12395 * config/freebsd.h (LINK_EH_SPEC): Likewise.
12396 * config/openbsd.h (LINK_EH_SPEC): Likewise.
12397 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12398 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
12399 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
12400 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12401 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
12402 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
12404 2017-09-13 Jakub Jelinek <jakub@redhat.com>
12406 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
12408 (ENDFILE_LINUX_SPEC): Likewise.
12409 (LINK_EH_SPEC): Likewise.
12410 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
12411 (LINK_OS_LINUX_SPEC32): Likewise.
12412 (LINK_OS_LINUX_SPEC64): Likewise.
12413 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
12414 (LINK_OS_LINUX_SPEC): Likewise.
12416 2017-09-13 Martin Liska <mliska@suse.cz>
12418 PR middle-end/82154
12419 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
12420 CASE_HIGH is NULL_TREE.
12422 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12423 Alan Hayward <alan.hayward@arm.com>
12424 David Sherwood <david.sherwood@arm.com>
12426 * target.def (secondary_memory_needed): New hook.
12427 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
12428 instead of SECONDARY_MEMORY_NEEDED.
12429 (secondary_memory_needed_mode): Likewise.
12430 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
12431 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
12432 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
12433 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
12434 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
12435 * doc/tm.texi: Regenerate.
12436 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
12437 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
12438 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12439 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
12440 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
12441 * config/i386/i386.c (inline_secondary_memory_needed): Put the
12442 mode argument first and change the reg_class arguments to reg_class_t.
12443 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
12444 Make static. Update the call to inline_secondary_memory_needed.
12445 (ix86_register_move_cost): Update the call to
12446 inline_secondary_memory_needed.
12447 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12448 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
12450 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
12451 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12453 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
12454 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
12455 * config/mips/mips.c (mips_secondary_memory_needed): Make static
12456 and match hook interface. Add comment from mips.h.
12457 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12458 * config/mmix/mmix.md (truncdfsf2): Refer to
12459 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12461 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
12462 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
12463 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12464 (pa_secondary_memory_needed): New function.
12465 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
12466 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
12467 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12468 (pdp11_secondary_memory_needed): Make static and match hook interface.
12469 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
12470 * config/powerpcspe/powerpcspe-protos.h
12471 (rs6000_secondary_memory_needed_ptr): Delete.
12472 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
12474 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12475 (rs6000_option_override_internal): Assign to
12476 targetm.secondary_memory_needed rather than
12477 rs6000_secondary_memory_needed_ptr.
12478 (rs6000_secondary_memory_needed): Match hook interface.
12479 (rs6000_debug_secondary_memory_needed): Likewise.
12480 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
12481 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
12482 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
12483 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12484 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
12485 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
12487 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
12488 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12489 (rs6000_option_override_internal): Assign to
12490 targetm.secondary_memory_needed rather than
12491 rs6000_secondary_memory_needed_ptr.
12492 (rs6000_secondary_memory_needed): Match hook interface.
12493 (rs6000_debug_secondary_memory_needed): Likewise.
12494 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
12495 * config/s390/s390.c (s390_secondary_memory_needed): New function.
12496 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12497 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
12498 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12499 (sparc_secondary_memory_needed): New function.
12500 * lra-constraints.c (check_and_process_move): Refer to
12501 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12503 (curr_insn_transform): Likewise.
12504 (process_alt_operands): Use targetm.secondary_memory_needed
12505 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12506 (check_secondary_memory_needed_p): Likewise.
12507 (choose_split_class): Likewise.
12508 * reload.c: Unconditionally include code that was previously
12509 conditional on SECONDARY_MEMORY_NEEDED.
12510 (push_secondary_reload): Use targetm.secondary_memory_needed
12511 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12512 (push_reload): Likewise.
12513 * reload1.c: Unconditionally include code that was previously
12514 conditional on SECONDARY_MEMORY_NEEDED.
12515 (choose_reload_regs): Use targetm.secondary_memory_needed
12516 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12517 (gen_reload): Likewise.
12518 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
12520 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12521 Alan Hayward <alan.hayward@arm.com>
12522 David Sherwood <david.sherwood@arm.com>
12524 * target.def (secondary_memory_needed_mode): New hook:
12525 * targhooks.c (default_secondary_memory_needed_mode): Declare.
12526 * targhooks.h (default_secondary_memory_needed_mode): New function.
12527 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
12528 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
12529 * doc/tm.texi: Regenerate.
12530 * lra-constraints.c (check_and_process_move): Use
12531 targetm.secondary_memory_needed_mode instead of
12532 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
12533 (curr_insn_transform): Likewise.
12534 * reload.c (get_secondary_mem): Likewise.
12535 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12536 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
12538 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12539 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12540 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
12541 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12542 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
12544 * config/powerpcspe/powerpcspe-protos.h
12545 (rs6000_secondary_memory_needed_mode): Delete.
12546 * config/powerpcspe/powerpcspe.c
12547 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12548 (rs6000_secondary_memory_needed_mode): Make static.
12549 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12550 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
12552 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12554 (rs6000_secondary_memory_needed_mode): Make static.
12555 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12556 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
12557 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12558 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12559 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12561 (sparc_secondary_memory_needed_mode): New function.
12562 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
12564 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
12566 * config/aarch64/constraints.md (Umq): New constraint.
12567 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
12569 (mov<mode>): Update condition.
12571 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12573 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
12574 when bitposition is the same.
12576 2017-09-13 Richard Biener <rguenther@suse.de>
12578 * dwarf2out.c (output_die_symbol): Remove.
12579 (output_die): Do not output a DIEs symbol.
12581 2017-09-13 Richard Biener <rguenther@suse.de>
12583 PR middle-end/82128
12584 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
12585 default-def to avoid breaking iterator update with the weird
12586 interaction with cgraph_update_edges_for_call_stmt_node.
12588 2017-09-13 Richard Biener <rguenther@suse.de>
12590 * tree-cfg.c (verify_gimple_assign_binary): Add verification
12591 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
12592 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
12593 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
12595 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
12597 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
12598 Disable pc relative literal load irrespective of
12599 TARGET_FIX_ERR_A53_84341 for default.
12601 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
12603 * config/sparc/sparc.c (output_return): Output the source location of
12604 the insn in the delay slot, if any.
12605 (output_sibcall): Likewise.
12607 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
12610 * common.opt (-static-pie): New alias.
12611 (shared): Negate static-pie.
12612 (-no-pie): Update help text.
12614 (static-pie): New option.
12615 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
12616 -static-pie support.
12617 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
12618 (LINK_EH_SPEC): Likewise.
12619 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12620 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
12621 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
12622 * gcc.c (LINK_COMMAND_SPEC): Likewise.
12623 (init_gcc_specs): Likewise.
12624 (init_spec): Likewise.
12625 (display_help): Update help message for -pie.
12626 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
12629 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
12631 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
12632 (movdi_aarch64): Likewise.
12633 (movti_aarch64): Likewise.
12635 2017-09-12 Simon Wright <simon@pushface.org>
12638 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
12639 calculation of the minor version, always output as 0.
12641 2017-09-12 Jakub Jelinek <jakub@redhat.com>
12644 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
12645 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
12646 on it early, rather than manual conversion late. For
12647 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
12648 instead of performing manual conversion.
12650 2017-09-12 Carl Love <cel@us.ibm.com>
12652 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
12653 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
12655 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12656 VMULOSW): Add definitions.
12657 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12658 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12659 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12660 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12661 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12663 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
12665 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
12667 (movti_aarch64): Likewise.
12668 (movdf_aarch64): Likewise.
12669 (movtf_aarch64): Likewise.
12670 (load_pairdi): Likewise.
12671 (store_pairdi): Likewise.
12672 (load_pairdf): Likewise.
12673 (store_pairdf): Likewise.
12674 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
12675 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
12676 (ldr_got_small_<mode>): Likewise.
12677 (ldr_got_small_28k_<mode>): Likewise.
12678 (ldr_got_tiny): Likewise.
12679 * config/aarch64/iterators.md (ldst_sz): New.
12680 (ldpstp_sz): Likewise.
12681 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
12683 (thunderx_load): Split load_8 to load_16.
12684 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
12686 (thunderx2t99_storepair_basic): Split store_8 to store_16.
12687 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
12688 (xgene1_store_pair): Split store_8 to store_16.
12689 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
12690 (falkor_st_0_st_sd): Split store_8 to store_16.
12692 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
12694 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
12695 and store1/2/3/4 to store_4/8/12/16.
12696 * config/aarch64/aarch64.md: Update for rename.
12697 * config/arm/arm.md: Likewise.: Likewise.
12698 * config/arm/arm.c: Likewise.
12699 * config/arm/thumb1.md: Likewise.
12700 * config/arm/thumb2.md: Likewise.
12701 * config/arm/vfp.md: Likewise.
12702 * config/arm/arm-generic.md: Likewise.
12703 * config/arm/arm1020e.md: Likewise.
12704 * config/arm/arm1026ejs.md: Likewise.
12705 * config/arm/arm1136jfs.md: Likewise.
12706 * config/arm/arm926ejs.md: Likewise.
12707 * config/arm/cortex-a15.md: Likewise.
12708 * config/arm/cortex-a17.md: Likewise.
12709 * config/arm/cortex-a5.md: Likewise.
12710 * config/arm/cortex-a53.md: Likewise.
12711 * config/arm/cortex-a57.md: Likewise.
12712 * config/arm/cortex-a7.md: Likewise.
12713 * config/arm/cortex-a8.md: Likewise.
12714 * config/arm/cortex-a9.md: Likewise.
12715 * config/arm/cortex-m4.md: Likewise.
12716 * config/arm/cortex-m7.md: Likewise.
12717 * config/arm/cortex-r4.md: Likewise.
12718 * config/arm/exynos-m1.md: Likewise.
12719 * config/arm/fa526.md: Likewise.
12720 * config/arm/fa606te.md: Likewise.
12721 * config/arm/fa626te.md: Likewise.
12722 * config/arm/fa726te.md: Likewise.
12723 * config/arm/fmp626.md: Likewise.
12724 * config/arm/iwmmxt.md: Likewise.
12725 * config/arm/ldmstm.md: Likewise.
12726 * config/arm/marvell-pj4.md: Likewise.
12727 * config/arm/xgene1.md: Likewise.
12728 * config/aarch64/thunderx.md: Likewise.
12729 * config/aarch64/thunderx2t99.md: Likewise.
12730 * config/aarch64/falkor.md: Likewise.
12732 2017-09-12 Martin Liska <mliska@suse.cz>
12734 * attribs.c (private_lookup_attribute): New function.
12735 * attribs.h (private_lookup_attribute): Declared here.
12736 (lookup_attribute): Called from this place.
12738 2017-09-12 Richard Biener <rguenther@suse.de>
12740 PR tree-optimization/82157
12741 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
12742 stmts with side-effects.
12744 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12745 Alan Hayward <alan.hayward@arm.com>
12746 David Sherwood <david.sherwood@arm.com>
12748 * target.def (hard_regno_nregs): New hook.
12749 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
12750 * targhooks.h (default_hard_regno_nregs): Declare.
12751 * targhooks.c (default_hard_regno_nregs): New function.
12752 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
12753 (TARGET_HARD_REGNO_NREGS): ...this hook.
12754 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
12755 (CLASS_MAX_NREGS): Likewise.
12756 * doc/tm.texi: Regenerate.
12757 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
12758 instead of HARD_REGNO_NREGS.
12759 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
12760 HARD_REGNO_NREGS in the comment.
12761 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
12762 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
12763 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
12764 Return an unsigned int.
12765 (TARGET_HARD_REGNO_NREGS): Redefine.
12766 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
12767 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
12768 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
12769 (arc_hard_regno_nregs): New function.
12770 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
12771 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
12772 (arm_hard_regno_nregs): New function.
12773 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
12774 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
12775 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
12776 (TARGET_HARD_REGNO_NREGS): Redefine.
12777 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
12778 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
12779 (HARD_REGNO_NREGS): Delete.
12780 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
12781 (cr16_hard_regno_nregs): New function.
12782 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
12783 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
12784 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
12785 (cris_hard_regno_nregs): New function.
12786 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
12787 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
12788 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
12789 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
12790 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
12791 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
12792 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
12793 (frv_hard_regno_nregs): Make static. Take and return an
12795 (frv_class_max_nregs): Remove outdated copy of documentation.
12796 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
12797 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
12798 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
12799 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
12800 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
12801 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
12802 (TARGET_HARD_REGNO_NREGS): Redefine.
12803 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
12804 (CLASS_MAX_NREGS): Update comment.
12805 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
12806 (ia64_hard_regno_nregs): New function.
12807 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
12808 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
12809 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
12810 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
12811 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
12813 (m32c_hard_regno_nregs): Likewise. Make static.
12814 (TARGET_HARD_REGNO_NREGS): Redefine.
12815 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
12816 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
12817 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
12818 (m68k_hard_regno_nregs): New function.
12819 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
12820 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
12821 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
12822 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
12823 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
12824 Take and return an unsigned int.
12825 (TARGET_HARD_REGNO_NREGS): Redefine.
12826 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
12827 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
12828 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
12829 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
12830 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
12831 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
12832 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
12833 (msp430_hard_regno_nregs): Make static. Take and return an
12835 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
12836 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
12837 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
12838 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
12839 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
12840 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
12841 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
12842 (TARGET_HARD_REGNO_NREGS): Redefine.
12843 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
12844 (PA_HARD_REGNO_NREGS): ...this.
12845 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
12846 (PA_HARD_REGNO_NREGS): ...this.
12847 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
12848 (pa_hard_regno_nregs): New function.
12849 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
12850 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
12851 (pdp11_hard_regno_nregs): New function.
12852 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
12853 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
12854 (rs6000_hard_regno_nregs_hook): New function.
12855 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
12856 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
12857 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
12858 Take and return an unsigned int. Move earlier in file.
12859 (TARGET_HARD_REGNO_NREGS): Redefine.
12860 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
12861 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
12862 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
12863 (rl78_hard_regno_nregs): Make static. Take and return an
12865 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
12866 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
12867 (rs6000_hard_regno_nregs_hook): New function.
12868 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
12869 * config/rx/rx.c (rx_hard_regno_nregs): New function.
12870 (TARGET_HARD_REGNO_NREGS): Redefine.
12871 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
12872 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
12873 instead of HARD_REGNO_NREGS.
12874 (s390_hard_regno_nregs): New function.
12875 (s390_hard_regno_mode_ok): Add comment from s390.h.
12876 (TARGET_HARD_REGNO_NREGS): Redefine.
12877 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
12878 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
12879 (sh_hard_regno_nregs): New function.
12880 (sh_pass_in_reg_p): Use it.
12881 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
12882 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
12883 (sparc_hard_regno_nregs): New function.
12884 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
12885 * config/spu/spu.c (spu_hard_regno_nregs): New function.
12886 (spu_function_arg_advance): Use it, supplying a valid register number.
12887 (TARGET_HARD_REGNO_NREGS): Redefine.
12888 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
12889 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
12890 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
12891 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
12892 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
12893 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
12894 (CLASS_MAX_NREGS): Remove copy of old documentation.
12895 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
12896 (visium_hard_regno_nregs): New function.
12897 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
12898 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
12899 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
12900 xtensa_hard_regno_nregs): New function.
12901 * system.h (HARD_REGNO_NREGS): Poison.
12903 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12905 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
12906 hard_regno_nregs instead of HARD_REGNO_NREGS.
12907 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
12908 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
12909 (c6x_expand_epilogue): Likewise.
12910 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
12911 (frv_read_iacc_argument): Likewise.
12912 * config/sh/sh.c: Include regs.h.
12913 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12914 (regs_used): Likewise.
12915 (output_stack_adjust): Likewise.
12916 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
12917 * expmed.c: Include regs.h.
12918 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12919 * ree.c: Include regs.h.
12920 (combine_reaching_defs): Use hard_regno_nregs instead of
12922 (add_removable_extension): Likewise.
12924 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12926 * regs.h (hard_regno_nregs): Turn into a function.
12927 (end_hard_regno): Update accordingly.
12928 * caller-save.c (setup_save_areas): Likewise.
12929 (save_call_clobbered_regs): Likewise.
12930 (replace_reg_with_saved_mem): Likewise.
12931 (insert_restore): Likewise.
12932 (insert_save): Likewise.
12933 * combine.c (can_change_dest_mode): Likewise.
12934 (move_deaths): Likewise.
12935 (distribute_notes): Likewise.
12936 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
12937 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
12938 (rs6000_split_multireg_move): Likewise.
12939 (rs6000_register_move_cost): Likewise.
12940 (rs6000_memory_move_cost): Likewise.
12941 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
12942 (rs6000_split_multireg_move): Likewise.
12943 (rs6000_register_move_cost): Likewise.
12944 (rs6000_memory_move_cost): Likewise.
12945 * cselib.c (cselib_reset_table): Likewise.
12946 (cselib_lookup_1): Likewise.
12947 * emit-rtl.c (set_mode_and_regno): Likewise.
12948 * function.c (aggregate_value_p): Likewise.
12949 * ira-color.c (setup_profitable_hard_regs): Likewise.
12950 (check_hard_reg_p): Likewise.
12951 (calculate_saved_nregs): Likewise.
12952 (assign_hard_reg): Likewise.
12953 (improve_allocation): Likewise.
12954 (calculate_spill_cost): Likewise.
12955 * ira-emit.c (modify_move_list): Likewise.
12956 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
12957 (ira_hard_reg_in_set_p): Likewise.
12958 * ira.c (setup_reg_mode_hard_regset): Likewise.
12959 (clarify_prohibited_class_mode_regs): Likewise.
12960 (check_allocation): Likewise.
12961 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12962 (lra_setup_reg_renumber): Likewise.
12963 (setup_try_hard_regno_pseudos): Likewise.
12964 (spill_for): Likewise.
12965 (assign_hard_regno): Likewise.
12966 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
12967 * lra-constraints.c (in_class_p): Likewise.
12968 (lra_constraint_offset): Likewise.
12969 (simplify_operand_subreg): Likewise.
12970 (lra_constraints): Likewise.
12971 (split_reg): Likewise.
12972 (split_if_necessary): Likewise.
12973 (invariant_p): Likewise.
12974 (inherit_in_ebb): Likewise.
12975 * lra-lives.c (process_bb_lives): Likewise.
12976 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
12977 (get_hard_regs): Likewise.
12978 (do_remat): Likewise.
12979 * lra-spills.c (assign_spill_hard_regs): Likewise.
12980 * mode-switching.c (create_pre_exit): Likewise.
12981 * postreload.c (reload_combine_recognize_pattern): Likewise.
12982 * recog.c (peep2_find_free_register): Likewise.
12983 * regcprop.c (kill_value_regno): Likewise.
12984 (set_value_regno): Likewise.
12985 (copy_value): Likewise.
12986 (maybe_mode_change): Likewise.
12987 (find_oldest_value_reg): Likewise.
12988 (copyprop_hardreg_forward_1): Likewise.
12989 * regrename.c (check_new_reg_p): Likewise.
12990 (regrename_do_replace): Likewise.
12991 * reload.c (push_reload): Likewise.
12992 (combine_reloads): Likewise.
12993 (find_dummy_reload): Likewise.
12994 (operands_match_p): Likewise.
12995 (find_reloads): Likewise.
12996 (find_equiv_reg): Likewise.
12997 (reload_adjust_reg_for_mode): Likewise.
12998 * reload1.c (count_pseudo): Likewise.
12999 (count_spilled_pseudo): Likewise.
13000 (find_reg): Likewise.
13001 (clear_reload_reg_in_use): Likewise.
13002 (free_for_value_p): Likewise.
13003 (allocate_reload_reg): Likewise.
13004 (choose_reload_regs): Likewise.
13005 (reload_adjust_reg_for_temp): Likewise.
13006 (emit_reload_insns): Likewise.
13007 (delete_output_reload): Likewise.
13008 * rtlanal.c (subreg_get_info): Likewise.
13009 * sched-deps.c (sched_analyze_reg): Likewise.
13010 * sel-sched.c (init_regs_for_mode): Likewise.
13011 (mark_unavailable_hard_regs): Likewise.
13012 (choose_best_reg_1): Likewise.
13013 (verify_target_availability): Likewise.
13014 * valtrack.c (dead_debug_insert_temp): Likewise.
13015 * var-tracking.c (track_loc_p): Likewise.
13016 (emit_note_insn_var_location): Likewise.
13017 * varasm.c (make_decl_rtl): Likewise.
13018 * reginfo.c (choose_hard_reg_mode): Likewise.
13019 (init_reg_modes_target): Refer directly to
13020 this_target_regs->x_hard_regno_nregs.
13022 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13024 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
13025 instead of hard_regno_nregs.
13027 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13029 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
13030 end_hard_regno instead of hard_regno_nregs.
13031 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
13032 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
13033 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
13034 * ira-color.c (improve_allocation): Likewise.
13035 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13036 * lra-lives.c (mark_regno_live): Likewise.
13037 (mark_regno_dead): Likewise.
13038 * lra-remat.c (operand_to_remat): Likewise.
13039 * lra.c (collect_non_operand_hard_regs): Likewise.
13040 * postreload.c (reload_combine_note_store): Likewise.
13041 (move2add_valid_value_p): Likewise.
13042 * reload.c (regno_clobbered_p): Likewise.
13044 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13046 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
13048 * config/v850/v850.c (v850_reorg): Likewise.
13049 * reload.c (refers_to_regno_for_reload_p): Likewise.
13050 (find_equiv_reg): Likewise.
13051 * reload1.c (reload_reg_reaches_end_p): Likewise.
13053 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13055 * caller-save.c (add_used_regs): Use REG_NREGS instead of
13057 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
13058 * config/arm/arm.c (output_move_neon): Likewise.
13059 (arm_attr_length_move_neon): Likewise.
13060 (neon_split_vcombine): Likewise.
13061 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
13062 (c6x_mark_reg_written): Likewise.
13063 (c6x_dwarf_register_span): Likewise.
13064 * config/i386/i386.c (ix86_save_reg): Likewise.
13065 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
13066 (rws_access_reg): Likewise.
13067 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
13068 * mode-switching.c (create_pre_exit): Likewise.
13069 * ree.c (combine_reaching_defs): Likewise.
13070 (add_removable_extension): Likewise.
13071 * regcprop.c (find_oldest_value_reg): Likewise.
13072 (copyprop_hardreg_forward_1): Likewise.
13073 * reload.c (reload_inner_reg_of_subreg): Likewise.
13074 (push_reload): Likewise.
13075 (combine_reloads): Likewise.
13076 (find_dummy_reload): Likewise.
13077 (reload_adjust_reg_for_mode): Likewise.
13078 * reload1.c (find_reload_regs): Likewise.
13079 (forget_old_reloads_1): Likewise.
13080 (reload_reg_free_for_value_p): Likewise.
13081 (reload_adjust_reg_for_temp): Likewise.
13082 (emit_reload_insns): Likewise.
13083 (delete_output_reload): Likewise.
13084 * sel-sched.c (choose_best_reg_1): Likewise.
13085 (choose_best_pseudo_reg): Likewise.
13087 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13088 Alan Hayward <alan.hayward@arm.com>
13089 David Sherwood <david.sherwood@arm.com>
13091 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
13092 * target.def (slow_unaligned_access): New hook.
13093 * targhooks.h (default_slow_unaligned_access): Declare.
13094 * targhooks.c (default_slow_unaligned_access): New function.
13095 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
13096 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
13097 * doc/tm.texi: Regenerate.
13098 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
13099 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
13100 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
13102 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
13103 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
13105 (rs6000_slow_unaligned_access): New function.
13106 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
13107 (expand_block_compare): Likewise.
13108 (expand_strn_compare): Likewise.
13109 (rs6000_rtx_costs): Likewise.
13110 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
13111 (riscv_slow_unaligned_access): Likewise.
13112 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
13113 (riscv_slow_unaligned_access_p): ...this and make static.
13114 (riscv_option_override): Update accordingly.
13115 (riscv_slow_unaligned_access): New function.
13116 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
13117 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
13118 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
13119 (rs6000_slow_unaligned_access): New function.
13120 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
13121 (rs6000_rtx_costs): Likewise.
13122 * config/rs6000/rs6000-string.c (expand_block_compare)
13123 (expand_strn_compare): Use targetm.slow_unaligned_access instead
13124 of SLOW_UNALIGNED_ACCESS.
13125 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
13126 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
13127 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
13128 of SLOW_UNALIGNED_ACCESS.
13129 * expmed.c (simple_mem_bitfield_p): Likewise.
13130 * expr.c (alignment_for_piecewise_move): Likewise.
13131 (emit_group_load_1): Likewise.
13132 (emit_group_store): Likewise.
13133 (copy_blkmode_from_reg): Likewise.
13134 (emit_push_insn): Likewise.
13135 (expand_assignment): Likewise.
13136 (store_field): Likewise.
13137 (expand_expr_real_1): Likewise.
13138 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
13139 * lra-constraints.c (simplify_operand_subreg): Likewise.
13140 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13141 * gimple-ssa-store-merging.c: Likewise in block comment at start
13143 * tree-ssa-strlen.c: Include target.h.
13144 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
13145 of SLOW_UNALIGNED_ACCESS.
13146 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
13148 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13150 PR rtl-optimization/82185
13151 * expmed.c (emit_store_flag_int): Only test tem if it has been
13154 2017-09-12 Richard Biener <rguenther@suse.de>
13156 PR middle-end/82149
13157 * match.pd ((FTYPE) N CMP CST): Fix typo.
13159 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
13161 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
13163 (mips_near_type_p): Add 'short_call' attribute as a synonym
13165 * doc/extend.texi (short_call): Document new function attribute.
13167 2017-09-12 Jakub Jelinek <jakub@redhat.com>
13170 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
13171 assertion check that in the condition.
13172 (get_atomic_generic_size): Likewise. Before testing if parameter
13173 has pointer type, if it has array type, call for C++
13174 default_conversion to perform array-to-pointer conversion.
13176 2017-09-12 Richard Biener <rguenther@suse.de>
13178 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
13179 for operations we cannot scalarize.
13181 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
13183 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
13184 vectors heap vectors. Clean up comments.
13185 Make visited_bbs a reference.
13186 (profitable_jump_thread_path): Make GC
13187 vectors heap vectors. Clean up comments.
13189 (convert_and_register_jump_thread_path): Make GC vectors heap
13191 (check_subpath_and_update_thread_path): Same. Clean up comments.
13192 Make visited_bbs a reference.
13193 (handle_phi): Abstract common code to to
13194 register_jump_thread_path_if_profitable.
13195 Rename VAR_BB to DEF_BB.
13197 Make GC vectors heap vectors.
13198 Make visited_bbs a reference.
13199 (handle_assignment): Same.
13200 (register_jump_thread_path_if_profitable): New.
13201 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
13203 Make GC vectors heap vectors. Clean up comments.
13204 Make visited_bbs a reference.
13205 (find_jump_threads_backwards): Make visited_bbs live in the stack.
13206 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
13209 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
13212 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
13213 words of E_DImode object are reachable by xtensa_uimm8x4 access.
13215 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
13217 Revert r251800 and r251799.
13219 2017-09-11 Martin Jambor <mjambor@suse.cz>
13222 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
13223 arguments in advance.
13224 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
13225 use it to find predecessor edges.
13226 (naive_outof_ssa): Collect vector of predecessors.
13228 2017-09-08 Jason Merrill <jason@redhat.com>
13230 PR c++/70029 - ICE with ref-qualifier and -flto
13231 * langhooks.h (struct lang_hooks_for_types): Add
13232 copy_lang_qualifiers.
13233 * attribs.c (build_type_attribute_qual_variant): Use it.
13234 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
13236 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
13237 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
13239 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
13242 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
13243 (*mulsi3_sp64): New instruction.
13244 (mulsi3): New expander.
13246 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
13248 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
13250 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13252 * sancov.c: Include memmodel.h.
13254 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
13257 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
13260 2017-09-07 Carl Love <cel@us.ibm.com>
13262 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
13263 the sldi instruction.
13265 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
13267 * sancov.c: Include tm_p.h.
13269 2017-09-07 Jakub Jelinek <jakub@redhat.com>
13272 * output.h (switch_to_other_text_partition): New declaration.
13273 * varasm.c (switch_to_other_text_partition): New function.
13274 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
13275 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
13276 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
13277 to the other text partition before emitting LCL label and switch back
13278 after emitting the word after it.
13280 2017-09-07 Richard Biener <rguenther@suse.de>
13282 * passes.def (pass_split_crit_edges): Remove instance before PRE.
13283 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
13284 critical edges here, after loop init.
13285 (pass_data_pre): Remove PROP_no_crit_edges flags.
13286 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
13287 for valueization of call args to avoid leaking VN_TOP.
13288 (visit_use): Assert we do not visit default defs.
13289 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
13290 Use error_mark_node to more easily detect leaking VN_TOP.
13291 All default-defs are varying, not VN_TOP. Mark them visited.
13292 (run_scc_vn): Make code match comment.
13294 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13296 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
13297 OPTION_MASK_FLOAT128_KEYWORD.
13298 (POWERPC_MASKS): Likewise.
13299 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
13300 support for the -mfloat128-type option, and make -mfloat128
13301 default on PowerPC Linux systems. Define or undefine
13302 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
13303 Define __float128 to be __ieee128 if IEEE 128-bit support is
13304 enabled, or undefine it.
13305 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
13306 Delete defining __FLOAT128_TYPE__.
13307 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
13308 -mfloat128-type option and make -mfloat128 default on PowerPC
13310 (TARGET_FLOAT128_TYPE): Likewise.
13311 (-mfloat128-type): Likewise.
13312 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13313 Delete the -mfloat128-type option and make -mfloat128 default on
13314 PowerPC Linux systems. Always use __ieee128 to be the keyword for
13315 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
13316 128-bit floating point is enabled. Change tests from using
13317 -mfloat128-type to -mfloat128.
13318 (rs6000_mangle_type): Use the correct mangling for the __float128
13319 type even if normal long double is restricted to 64-bits.
13320 (floatn_mode): Enable the _Float128 type by default on VSX Linux
13322 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
13323 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
13324 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
13326 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
13327 documentation for -mfloat128.
13329 2017-09-06 Olivier Hainque <hainque@adacore.com>
13331 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
13333 2017-09-06 Wish Wu <wishwu007@gmail.com>
13334 Jakub Jelinek <jakub@redhat.com>
13336 * asan.c (initialize_sanitizer_builtins): Add
13337 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
13338 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
13339 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
13340 BT_FN_VOID_UINT64_PTR variables.
13341 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
13342 (BT_FN_VOID_UINT16_UINT16): Likewise.
13343 (BT_FN_VOID_UINT32_UINT32): Likewise.
13344 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
13345 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
13346 (BT_FN_VOID_UINT64_PTR): Likewise.
13347 * common.opt (flag_sanitize_coverage): New variable.
13348 (fsanitize-coverage=trace-pc): Remove.
13349 (fsanitize-coverage=): Add.
13350 * flag-types.h (enum sanitize_coverage_code): New enum.
13351 * fold-const.c (fold_range_test): Disable non-short-circuit
13352 optimization if flag_sanitize_coverage.
13353 (fold_truth_andor): Likewise.
13354 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
13355 * opts.c (COVERAGE_SANITIZER_OPT): Define.
13356 (coverage_sanitizer_opts): New array.
13357 (get_closest_sanitizer_option): Add OPTS argument, handle also
13358 OPT_fsanitize_coverage_.
13359 (parse_sanitizer_options): Adjusted to also handle
13360 OPT_fsanitize_coverage_.
13361 (common_handle_option): Add OPT_fsanitize_coverage_.
13362 * sancov.c (instrument_comparison, instrument_switch): New function.
13363 (sancov_pass): Add trace-cmp support.
13364 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
13365 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
13366 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
13367 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
13368 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
13369 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
13370 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
13371 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
13372 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
13373 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
13375 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13377 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13378 error. Only quit immediately if parsing is complete.
13379 (BEGIN): Initialize fatal_err and parse_done.
13380 (begin fpu, end fpu): Check number of arguments.
13381 (begin arch, end arch): Likewise.
13382 (begin cpu, end cpu): Likewise.
13383 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13384 (optalias): Likewise.
13386 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13388 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13389 * config/arm/arm-isa.h: Delete. Move definitions to ...
13390 * arm-cpus.in: ... here. Use new feature and fgroup values.
13391 * config/arm/arm.c (arm_option_override): Use lower case for feature
13393 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13394 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13395 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13397 (print_isa_bits_for): New function.
13398 (gen_isa): New function.
13399 (gen_comm_data): Use print_isa_bits_for.
13400 (define feature): New keyword.
13401 (define fgroup): New keyword.
13402 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
13403 (arm-isa.h): Add rule to generate file.
13404 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13405 case for feature bit names.
13407 2017-09-06 Richard Biener <rguenther@suse.de>
13409 * tree-ssa-pre.c (NECESSARY): Remove.
13410 (create_expression_by_pieces): Do not touch pass-local flags.
13411 (insert_into_preds_of_block): Likewise.
13412 (do_pre_regular_insertion): Likewise.
13413 (eliminate_insert): Likewise.
13414 (eliminate_dom_walker::before_dom_children): Likewise.
13415 (fini_eliminate): Do not look at inserted_exprs.
13416 (mark_operand_necessary): Remove.
13417 (remove_dead_inserted_code): Replace with simple work-list
13418 algorithm based on inserted_exprs and SSA uses.
13419 (pass_pre::execute): Re-order fini_eliminate and
13420 remove_dead_inserted_code.
13422 2017-09-06 Olivier Hainque <hainque@adacore.com>
13424 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
13425 for VxWorks 7. Adjust surrounding comments.
13427 2017-09-06 Richard Biener <rguenther@suse.de>
13429 * gimple-ssa-strength-reduction.c
13430 (find_candidates_dom_walker::before_dom_children): Also allow
13433 2017-09-06 Richard Biener <rguenther@suse.de>
13435 PR tree-optimization/82108
13436 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
13437 for gap in the non-permutation SLP case.
13439 2017-09-06 Martin Jambor <mjambor@suse.cz>
13441 PR tree-optimization/82078
13442 * tree-sra.c (sort_and_splice_var_accesses): Move call to
13443 add_access_to_work_queue...
13444 (build_accesses_from_assign): ...here.
13445 (propagate_all_subaccesses): Make sure racc is the group
13446 representative, if there is one.
13448 2017-09-06 Jakub Jelinek <jakub@redhat.com>
13450 PR middle-end/82095
13451 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
13454 2017-09-06 Richard Biener <rguenther@suse.de>
13456 * gimple-ssa-strength-reduction.c
13457 (find_candidates_dom_walker::before_doom_children): Use a
13458 type and not a mode check.
13460 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
13463 * config/arm/predicates.md (arm_general_adddi_operand): Create new
13465 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
13467 2017-09-05 Jeff Law <law@redhat.com>
13469 PR tree-optimization/64910
13470 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
13471 cases where we have 3 or more operands.
13473 2017-09-05 Jakub Jelinek <jakub@redhat.com>
13475 PR middle-end/81768
13476 * omp-low.c (lower_omp_for): Recompute tree invariant if
13477 gimple_omp_for_initial/final is ADDR_EXPR.
13479 PR middle-end/81768
13480 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
13481 into gimple val before gimplification fo the COND_EXPR.
13483 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
13485 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
13486 REGION_COPY argument.
13487 (thread_through_all_blocks): Remove unused argument to
13488 duplicate_thread_path.
13490 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13491 Alan Hayward <alan.hayward@arm.com>
13492 David Sherwood <david.sherwood@arm.com>
13494 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
13495 Take a scalar_mode rather than a machine_mode.
13496 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13497 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
13498 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13499 (aarch64_gen_adjusted_ldpstp): Likewise.
13500 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
13502 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13503 Alan Hayward <alan.hayward@arm.com>
13504 David Sherwood <david.sherwood@arm.com>
13506 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
13507 Take a scalar_int_mode instead of a machine_mode.
13508 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13509 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13510 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13511 (aarch64_simd_attr_length_rglist): Delete.
13512 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
13513 a scalar_int_mode instead of a machine_mode.
13514 (aarch64_add_offset): Likewise.
13515 (aarch64_internal_mov_immediate): Likewise
13516 (aarch64_add_constant_internal): Likewise.
13517 (aarch64_add_constant): Likewise.
13518 (aarch64_movw_imm): Likewise.
13519 (aarch64_rtx_arith_op_extract_p): Likewise.
13520 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13521 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13522 Remove assert that the mode isn't a vector.
13523 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13524 (aarch64_expand_mov_immediate): Update calls after above changes.
13525 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
13526 (aarch64_and_bitmask_imm): Check for scalar integer modes.
13527 (aarch64_move_imm): Likewise.
13528 (aarch64_can_const_movi_rtx_p): Likewise.
13529 (aarch64_strip_extend): Likewise.
13530 (aarch64_extr_rtx_p): Likewise.
13531 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
13532 a CONST_INT when the mode parameter is VOIDmode.
13533 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
13535 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13537 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
13538 * stor-layout.c (bitwise_mode_for_mode): Likewise.
13539 (bitwise_type_for_mode): Update accordingly.
13541 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13543 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
13544 * stor-layout.c (mode_for_size_tree): Likewise.
13545 (mode_for_array): Update accordingly.
13546 (layout_decl): Likewise.
13547 (compute_record_mode): Likewise. Only set the mode once.
13549 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13551 * target.def (get_mask_mode): Change return type to opt_mode.
13553 * doc/tm.texi: Regenerate.
13554 * targhooks.h (default_get_mask_mode): Return an opt_mode.
13555 * targhooks.c (default_get_mask_mode): Likewise.
13556 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
13557 * optabs-query.c (can_vec_mask_load_store_p): Update use of
13558 targetm.get_mask_mode.
13559 * tree.c (build_truth_vector_type): Likewise.
13561 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13563 * machmode.h (mode_for_vector): Return an opt_mode.
13564 * stor-layout.c (mode_for_vector): Likewise.
13565 (mode_for_int_vector): Update accordingly.
13566 (layout_type): Likewise.
13567 * config/i386/i386.c (emit_memmov): Likewise.
13568 (ix86_expand_set_or_movmem): Likewise.
13569 (ix86_expand_vector_init): Likewise.
13570 (ix86_get_mask_mode): Likewise.
13571 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
13573 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
13574 * expmed.c (extract_bit_field_1): Likewise.
13575 * expr.c (expand_expr_real_2): Likewise.
13576 * optabs-query.c (can_vec_perm_p): Likewise.
13577 (can_vec_mask_load_store_p): Likewise.
13578 * optabs.c (expand_vec_perm): Likewise.
13579 * targhooks.c (default_get_mask_mode): Likewise.
13580 * tree-vect-stmts.c (vectorizable_store): Likewise.
13581 (vectorizable_load): Likewise.
13582 (get_vectype_for_scalar_type_and_size): Likewise.
13584 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13586 * machmode.h (mode_for_int_vector): New function.
13587 * stor-layout.c (mode_for_int_vector): Likewise.
13588 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
13589 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
13590 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
13591 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
13592 (s390_expand_vcond): Likewise.
13594 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13596 * machmode.h (opt_machine_mode): New type.
13597 (opt_mode<T>): Allow construction from anything that can be
13599 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
13600 (mode_for_size): Return an opt_machine_mode.
13601 * stor-layout.c (mode_for_size): Likewise.
13602 (mode_for_size_tree): Update call accordingly.
13603 (bitwise_mode_for_mode): Likewise.
13604 (make_fract_type): Likewise.
13605 (make_accum_type): Likewise.
13606 * caller-save.c (replace_reg_with_saved_mem): Update call
13608 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13609 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13610 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13611 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13612 * expmed.c (extract_bit_field_1): Likewise.
13613 * reload.c (get_secondary_mem): Likewise.
13614 * varasm.c (assemble_integer): Likewise.
13615 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
13618 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13620 * machmode.h (decimal_float_mode_for_size): New function.
13621 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
13622 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
13623 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
13624 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
13625 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
13626 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
13628 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13630 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
13631 (get_builtin_sync_mode): Likewise.
13632 (expand_ifn_atomic_compare_exchange): Likewise.
13633 (expand_builtin_atomic_clear): Likewise.
13634 (expand_builtin_atomic_test_and_set): Likewise.
13635 (fold_builtin_atomic_always_lock_free): Likewise.
13636 * calls.c (compute_argument_addresses): Likewise.
13637 (emit_library_call_value_1): Likewise.
13638 (store_one_arg): Likewise.
13639 * combine.c (combine_instructions): Likewise.
13640 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
13641 * config/arm/arm.c (arm_function_value): Likewise.
13642 (aapcs_allocate_return_reg): Likewise.
13643 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
13644 * config/i386/i386.c (construct_container): Likewise.
13645 (ix86_gimplify_va_arg): Likewise.
13646 (ix86_expand_sse_cmp): Likewise.
13647 (emit_memmov): Likewise.
13648 (emit_memset): Likewise.
13649 (expand_small_movmem_or_setmem): Likewise.
13650 (ix86_expand_pextr): Likewise.
13651 (ix86_expand_pinsr): Likewise.
13652 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
13653 * config/microblaze/microblaze.c (microblaze_block_move_straight):
13655 * config/mips/mips.c (mips_function_value_1) Likewise.
13656 (mips_block_move_straight): Likewise.
13657 (mips_expand_ins_as_unaligned_store): Likewise.
13658 * config/powerpcspe/powerpcspe.c
13659 (rs6000_darwin64_record_arg_advance_flush): Likewise.
13660 (rs6000_darwin64_record_arg_flush): Likewise.
13661 * config/rs6000/rs6000.c
13662 (rs6000_darwin64_record_arg_advance_flush): Likewise.
13663 (rs6000_darwin64_record_arg_flush): Likewise.
13664 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
13665 (sparc_function_value_1): Likewise.
13666 * config/spu/spu.c (adjust_operand): Likewise.
13667 (spu_emit_branch_or_set): Likewise.
13668 (arith_immediate_p): Likewise.
13669 * emit-rtl.c (gen_lowpart_common): Likewise.
13670 * expr.c (expand_expr_real_1): Likewise.
13671 * function.c (assign_parm_setup_block): Likewise.
13672 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
13673 * reload1.c (alter_reg): Likewise.
13674 * stor-layout.c (mode_for_vector): Likewise.
13675 (layout_type): Likewise.
13677 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13679 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
13680 (spu_convert_move): Likewise.
13681 * lower-subreg.c (resolve_simple_move): Likewise.
13683 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13686 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
13687 define_insn to a define_expand.
13688 (altivec_vsum2sws_direct): New define_insn.
13689 (altivec_vsumsws): Convert from a define_insn to a define_expand.
13691 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
13693 * config/arm/arm.c (arm_option_params_internal): Improve setting of
13696 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
13701 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
13702 realign stack if stack alignment needed is less than incoming
13705 2017-09-05 Marek Polacek <polacek@redhat.com>
13708 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
13711 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
13713 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
13715 2017-09-05 Richard Biener <rguenther@suse.de>
13717 PR tree-optimization/82084
13718 * fold-const.c (can_native_encode_string_p): Handle wide characters.
13720 2017-09-05 Richard Biener <rguenther@suse.de>
13722 PR tree-optimization/82102
13723 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
13725 2017-09-05 Martin Liska <mliska@suse.cz>
13727 PR tree-optimization/82032
13728 * tree-cfg.c (generate_range_test): New function.
13729 * tree-cfg.h (generate_range_test): Declared here.
13730 * tree-cfgcleanup.c (convert_single_case_switch): New function.
13731 (cleanup_control_expr_graph): Use it.
13732 * tree-switch-conversion.c (try_switch_expansion): Remove
13734 (emit_case_nodes): Use generate_range_test.
13736 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
13739 * config/i386/i386.md (*<btsc><mode>_mask): Add
13740 TARGET_USE_BT to insn constraint.
13741 (*btr<mode>_mask): Ditto.
13743 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
13745 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
13746 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
13748 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
13751 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
13752 TARGET_NEON and TARGET_IWMMXT.
13753 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
13754 TARGET_NEON and TARGET_IWMMXT.
13755 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
13757 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
13759 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
13760 (ix86_rewrite_tls_address): Ditto.
13761 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
13762 (ix86_rewrite_tls_address_1): Ditto.
13763 (ix86_rewrite_tls_address): Ditto.
13764 * config/i386/predicates.md (tls_address_pattern): New predicate.
13765 * config/i386/i386.md (TLS address splitter): New splitter.
13767 2017-09-04 Richard Biener <rguenther@suse.de>
13769 PR tree-optimization/82084
13770 * fold-const.h (can_native_encode_string_p): Declare.
13771 * fold-const.c (can_native_encode_string_p): Factor out from ...
13772 (native_encode_string): ... here.
13773 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
13774 vectorizing stores from constants we later cannot handle.
13776 2017-09-04 Marek Polacek <polacek@redhat.com>
13779 * doc/invoke.texi: Update -Wtautological-compare documentation.
13781 2017-09-04 Jeff Law <law@redhat.com>
13783 PR tree-optimization/64910
13784 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
13785 swap the first and last operand if the last is a constant.
13787 2017-09-04 Marek Polacek <polacek@redhat.com>
13790 * convert.c (do_narrow): When sanitizing signed integer overflows,
13791 bail out for signed types.
13792 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
13794 2017-09-04 Richard Biener <rguenther@suse.de>
13796 PR tree-optimization/82060
13797 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13798 Move devirtualization after stmt folding and before EH/AB/noreturn
13799 cleanup to get the stmt refs canonicalized. Use a bool instead
13800 of gimple_modified_p since that doesn't work for NOPs. Schedule
13801 NOPs generated by folding for removal.
13803 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13804 Alan Hayward <alan.hayward@arm.com>
13805 David Sherwood <david.sherwood@arm.com>
13807 * coretypes.h (pad_direction): New enum.
13808 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
13809 (FUNCTION_ARG_PADDING): Likewise.
13810 * target.def (function_arg_padding): New hook.
13811 * targhooks.h (default_function_arg_padding): Declare.
13812 * targhooks.c (default_function_arg_padding): New function.
13813 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
13814 (TARGET_FUNCTION_ARG_PADDING): ...this.
13815 * doc/tm.texi: Regenerate.
13816 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
13817 instead of direction.
13818 (compute_argument_addresses): Likewise.
13819 (load_register_parameters): Likewise.
13820 (emit_library_call_value_1): Likewise.
13821 (store_one_arg): Use targetm.calls.function_arg_padding instead
13822 of FUNCTION_ARG_PADDING.
13823 (must_pass_in_stack_var_size_or_pad): Likewise.
13824 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
13825 (emit_group_store): Likewise.
13826 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
13827 instead of FUNCTION_ARG_PADDING.
13828 (emit_push_insn): Likewise, and propagate enum change throughout
13830 * function.h (direction): Delete.
13831 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
13833 * function.c (assign_parm_find_stack_rtl): Likewise.
13834 (assign_parm_setup_block_p): Likewise.
13835 (assign_parm_setup_block): Likewise.
13836 (gimplify_parameters): Likewise.
13837 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
13838 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
13840 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
13841 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13842 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
13843 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
13844 (aarch64_function_arg_padding): ...this new function.
13845 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
13846 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13847 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
13848 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13849 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
13850 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13851 (arm_pad_arg_upward): Replace with...
13852 (arm_function_arg_padding): ...this new function.
13853 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
13855 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
13856 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
13857 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13858 (ia64_hpux_function_arg_padding): Replace with...
13859 (ia64_function_arg_padding): ...this new function. Use pad_direction
13860 instead of direction. Check for TARGET_HPUX.
13861 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
13862 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13863 (iq2000_function_arg_padding): New function.
13864 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
13865 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
13866 (mips_function_arg_padding): ...this new function.
13867 (mips_pad_reg_upward): Update accordingly.
13868 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13869 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
13870 targetm.calls.function_arg_padding.
13871 (FUNCTION_ARG_PADDING): Delete.
13872 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13873 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
13874 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13875 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
13876 (nios2_block_reg_padding): Return pad_direction instead of direction.
13877 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
13878 instead of direction.
13879 (nios2_function_arg_padding): Likewise. Make static.
13880 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13881 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
13882 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
13883 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
13884 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13885 (pa_function_arg_padding): Make static. Return pad_direction instead
13887 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
13888 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13889 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
13890 instead of direction. Use targetm.calls.function_arg_padding.
13891 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
13892 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13893 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
13894 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
13895 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
13897 (function_arg_padding): Rename to...
13898 (rs6000_function_arg_padding): ...this. Make static. Return
13899 pad_direction instead of direction.
13900 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13901 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
13902 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13903 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
13904 instead of direction. Use targetm.calls.function_arg_padding.
13905 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
13906 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13907 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
13908 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
13909 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
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/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
13915 * config/s390/s390.c (s390_function_arg_padding): New function.
13916 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13917 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
13918 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
13919 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13920 (function_arg_padding): Rename to...
13921 (sparc_function_arg_padding): ...this. Make static. Return
13922 pad_direction instead of direction.
13923 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
13924 * config/spu/spu.c (spu_function_arg_padding): New function.
13925 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13926 * system.h (FUNCTION_ARG_PADDING): Poison.
13928 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13929 Alan Hayward <alan.hayward@arm.com>
13930 David Sherwood <david.sherwood@arm.com>
13932 * target.def (modes_tieable_p): New hook.
13933 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
13934 (TARGET_MODES_TIEABLE_P): ...this.
13935 * doc/tm.texi.in: Regenerate.
13936 * hooks.h (hook_bool_mode_mode_true): Declare.
13937 * hooks.c (hook_bool_mode_mode_true): New function.
13938 * combine.c (subst): Use targetm.modes_tieable_p instead of
13940 * dse.c (find_shift_sequence): Likewise.
13941 * expmed.c (extract_low_bits): Likewise.
13942 * lower-subreg.c: Include target.h.
13943 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
13945 * rtlanal.c (rtx_cost): Likewise.
13946 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
13947 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
13948 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
13949 (TARGET_MODES_TIEABLE_P): Redefine.
13950 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
13951 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
13952 (TARGET_MODES_TIEABLE_P): Redefine.
13953 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
13954 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
13955 (arc_modes_tieable_p): New function.
13956 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
13957 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
13958 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
13959 (arm_modes_tieable_p): Make static.
13960 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
13961 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
13962 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
13963 (TARGET_MODES_TIEABLE_P): Redefine.
13964 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
13965 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
13966 (TARGET_MODES_TIEABLE_P): Redefine.
13967 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
13968 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
13969 (cr16_modes_tieable_p): New function.
13970 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
13971 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
13972 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
13973 (TRULY_NOOP_TRUNCATION): Update comment.
13974 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
13975 (TRULY_NOOP_TRUNCATION): Update comment.
13976 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
13977 (frv_modes_tieable_p): New function.
13978 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
13979 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
13980 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
13981 (TARGET_MODES_TIEABLE_P): Redefine.
13982 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
13983 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
13984 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
13985 (TARGET_MODES_TIEABLE_P): Redefine.
13986 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
13987 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
13988 (ia64_modes_tieable_p): New function.
13989 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
13990 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
13991 (iq2000_modes_tieable_p): New function.
13992 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
13993 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
13994 (lm32_modes_tieable_p): New function.
13995 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
13996 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
13997 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
13998 (TARGET_MODES_TIEABLE_P): Redefine.
13999 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
14000 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
14001 (m32r_modes_tieable_p): New function.
14002 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
14003 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
14004 (m68k_modes_tieable_p): New function.
14005 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
14006 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
14007 (mcore_modes_tieable_p): New function.
14008 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
14009 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
14011 (TARGET_MODES_TIEABLE_P): Redefine.
14012 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
14013 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
14014 * config/mips/mips.c (mips_modes_tieable_p): Make static.
14015 (TARGET_MODES_TIEABLE_P): Redefine.
14016 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
14017 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
14018 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
14019 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
14020 (mn10300_modes_tieable_p): ...this and make static.
14021 (TARGET_MODES_TIEABLE_P): Redefine.
14022 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
14023 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
14024 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
14025 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
14026 (msp430_modes_tieable_p): Make static.
14027 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
14028 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
14029 (TARGET_MODES_TIEABLE_P): Redefine.
14030 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
14031 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
14032 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
14033 (TARGET_MODES_TIEABLE_P): Redefine.
14034 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
14035 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
14036 * config/pa/pa.c (pa_modes_tieable_p): Make static.
14037 (TARGET_MODES_TIEABLE_P): Redefine.
14038 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
14039 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
14040 (pdp11_modes_tieable_p): New function.
14041 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
14042 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
14043 (rs6000_modes_tieable_p): New function.
14044 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
14045 * config/powerpcspe/powerpcspe.md: Update comment.
14046 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
14047 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
14048 (TARGET_MODES_TIEABLE_P): Redefine.
14049 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
14050 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
14051 (rl78_modes_tieable_p): New function.
14052 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
14053 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
14054 (rs6000_modes_tieable_p): New function.
14055 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
14056 * config/rs6000/rs6000.md: Update comment.
14057 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
14058 * config/rx/rx.c (rx_modes_tieable_p): New function.
14059 (TARGET_MODES_TIEABLE_P): Redefine.
14060 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
14061 * config/s390/s390.c (s390_modes_tieable_p): New function.
14062 (TARGET_MODES_TIEABLE_P): Redefine.
14063 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
14064 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
14065 (sh_modes_tieable_p): New function.
14066 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
14067 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
14068 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
14069 (sparc_modes_tieable_p): Make static.
14070 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
14071 * config/spu/spu.c (spu_modes_tieable_p): New function.
14072 (TARGET_MODES_TIEABLE_P): Redefine.
14073 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
14074 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
14075 (TARGET_MODES_TIEABLE_P): Redefine.
14076 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
14077 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
14078 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
14079 * config/v850/v850.c (v850_modes_tieable_p): New function.
14080 (TARGET_MODES_TIEABLE_P): Redefine.
14081 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
14082 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
14083 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
14084 (visium_modes_tieable_p): New function.
14085 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
14086 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
14087 (xtensa_modes_tieable_p): New function.
14088 * system.h (MODES_TIEABLE_P): Poison.
14090 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14091 Alan Hayward <alan.hayward@arm.com>
14092 David Sherwood <david.sherwood@arm.com>
14094 * target.def (hard_regno_mode_ok): New hook.
14095 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
14096 (TARGET_HARD_REGNO_MODE_OK): ...this.
14097 * doc/tm.texi.in: Regenerate.
14098 * hooks.h (hook_bool_uint_mode_true): Declare.
14099 * hooks.c (hook_bool_uint_mode_true): New function.
14100 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
14101 HARD_REGNO_MODE_OK.
14102 * genpreds.c (write_insn_preds_c): Add an include of target.h.
14103 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
14104 instead of HARD_REGNO_MODE_OK.
14105 * caller-save.c: Include target.h.
14106 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
14107 HARD_REGNO_MODE_OK.
14108 * combine.c (can_combine_p): Likewise.
14109 (combinable_i3pat): Likewise.
14110 (can_change_dest_mode): Likewise.
14111 * expr.c (init_expr_target): Likewise.
14112 (convert_move): Likewise.
14113 (convert_modes): Likewise.
14114 * ira.c (setup_prohibited_class_mode_regs): Likewise.
14115 (setup_prohibited_mode_move_regs): Likewise.
14116 * ira.h (target_ira): Likewise.
14117 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14118 * lra-constraints.c (process_alt_operands): Likewise.
14119 (split_reg): Likewise.
14120 * recog.c (peep2_find_free_register): Likewise.
14121 * ree.c (combine_reaching_defs): Likewise.
14122 * regcprop.c (maybe_mode_change): Likewise.
14123 * reginfo.c (init_reg_sets_1): Likewise.
14124 (choose_hard_reg_mode): Likewise.
14125 (simplifiable_subregs): Likewise.
14126 * regrename.c (check_new_reg_p): Likewise.
14127 * reload.c (find_valid_class): Likewise.
14128 (find_valid_class_1): Likewise.
14129 (reload_inner_reg_of_subreg): Likewise.
14130 (push_reload): Likewise.
14131 (combine_reloads): Likewise.
14132 (find_dummy_reload): Likewise.
14133 (find_reloads): Likewise.
14134 * reload1.c (find_reg): Likewise.
14135 (set_reload_reg): Likewise.
14136 (allocate_reload_reg): Likewise.
14137 (choose_reload_regs): Likewise.
14138 (reload_adjust_reg_for_temp): Likewise.
14139 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
14140 (simplify_subreg_regno): Likewise.
14141 * sel-sched.c (init_regs_for_mode): Likewise.
14142 * varasm.c (make_decl_rtl): Likewise.
14143 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
14144 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
14145 HARD_REGNO_MODE_OK.
14146 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
14147 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
14148 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14149 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
14150 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
14151 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14152 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
14153 (arc_mode_class): Delete.
14154 (HARD_REGNO_MODE_OK): Delete.
14155 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14156 (arc_hard_regno_mode_ok): Rename old array to...
14157 (arc_hard_regno_mode_ok_modes): ...this.
14158 (arc_conditional_register_usage): Update accordingly.
14159 (arc_mode_class): Make static.
14160 (arc_hard_regno_mode_ok): New function.
14161 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
14162 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
14163 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14164 (arm_hard_regno_mode_ok): Make static.
14165 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
14166 HARD_REGNO_MODE_OK.
14167 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
14168 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
14169 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
14171 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14172 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
14173 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
14174 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
14175 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
14176 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14177 * config/bfin/predicates.md (valid_reg_operand): Use
14178 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
14179 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
14180 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
14181 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14182 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
14183 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
14184 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14185 (cr16_hard_regno_mode_ok): Make static and return a bool.
14186 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
14187 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14188 (cris_hard_regno_mode_ok): New function.
14189 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
14190 (epiphany_mode_class): Delete.
14191 (HARD_REGNO_MODE_OK): Delete.
14192 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
14193 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14194 (hard_regno_mode_ok): Rename to...
14195 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
14196 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
14197 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
14198 HARD_REGNO_MODE_OK.
14199 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
14200 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
14201 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14202 (frv_hard_regno_mode_ok): Make static and return a bool.
14203 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
14204 HARD_REGNO_MODE_OK.
14205 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
14206 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
14207 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
14208 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
14210 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14211 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
14212 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
14213 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
14215 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14216 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
14217 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14218 (ia64_hard_regno_mode_ok): New function.
14219 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
14220 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14221 (iq2000_hard_regno_mode_ok): New function.
14222 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
14223 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14224 (lm32_hard_regno_mode_ok): New function.
14225 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
14226 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
14227 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
14228 instead of HARD_REGNO_MODE_OK.
14229 (m32c_hard_regno_ok): Rename to...
14230 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
14231 (m32c_cannot_change_mode_class): Update accordingly.
14232 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14233 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
14234 (m32r_mode_class): Delete.
14235 (HARD_REGNO_MODE_OK): Delete.
14236 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14237 (m32r_hard_regno_mode_ok): Rename to...
14238 (m32r_hard_regno_modes): ...this.
14239 (m32r_mode_class): Make static.
14240 (m32r_hard_regno_mode_ok): New function.
14241 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
14242 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
14243 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14244 (m68k_hard_regno_mode_ok): Make static.
14245 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
14246 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14247 (mcore_hard_regno_mode_ok): New function.
14248 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
14249 (HARD_REGNO_MODE_OK): Delete.
14250 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
14252 (microblaze_hard_regno_mode_ok_p): ...this and make static.
14253 (microblaze_hard_regno_mode_ok): New function.
14254 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14255 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
14256 (mips_hard_regno_mode_ok): Delete.
14257 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
14258 (mips_hard_regno_mode_ok_p): ...this and make static.
14259 (mips_hard_regno_mode_ok_p): Rename to...
14260 (mips_hard_regno_mode_ok_uncached): ...this.
14261 (mips_hard_regno_mode_ok): New function.
14262 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
14263 of HARD_REGNO_MODE_OK.
14264 (mips_option_override): Update after above name changes.
14265 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14266 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
14267 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
14268 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
14269 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
14270 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14271 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
14272 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
14273 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
14274 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14275 (msp430_hard_regno_mode_ok): Make static and return a bool.
14276 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
14277 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
14278 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
14280 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14281 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
14282 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
14283 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
14284 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
14285 (PA_HARD_REGNO_MODE_OK): ...this
14286 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
14287 (PA_HARD_REGNO_MODE_OK): ...this.
14288 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14289 (pa_hard_regno_mode_ok): New function.
14290 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
14291 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14292 (pdp11_hard_regno_mode_ok): New function.
14293 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
14294 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
14296 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
14298 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14299 (rs6000_hard_regno_mode_ok): Rename to...
14300 (rs6000_hard_regno_mode_ok_uncached): ...this.
14301 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14302 (rs6000_hard_regno_mode_ok): New function.
14303 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
14304 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
14305 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
14306 (riscv_hard_regno_mode_ok): ...this and make static.
14307 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14308 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
14309 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
14310 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14311 (rl78_hard_regno_mode_ok): Make static and return bool.
14312 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
14313 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
14315 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
14316 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14317 (rs6000_hard_regno_mode_ok): Rename to...
14318 (rs6000_hard_regno_mode_ok_uncached): ...this.
14319 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14320 (rs6000_hard_regno_mode_ok): New function.
14321 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
14322 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
14323 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14324 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
14325 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
14326 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
14327 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14328 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
14329 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
14330 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14331 (sh_hard_regno_mode_ok): Make static.
14332 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
14333 instead of HARD_REGNO_MODE_OK.
14334 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
14335 (sparc_mode_class): Delete.
14336 (HARD_REGNO_MODE_OK): Delete.
14337 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14338 (hard_regno_mode_classes): Make static.
14339 (sparc_mode_class): Likewise.
14340 (sparc_hard_regno_mode_ok): New function.
14341 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
14342 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
14343 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
14345 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14346 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
14347 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
14348 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
14349 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
14350 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14351 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
14352 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
14353 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14354 (visium_hard_regno_mode_ok): New function.
14355 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
14356 instead of HARD_REGNO_MODE_OK.
14357 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
14358 (HARD_REGNO_MODE_OK): Delete.
14359 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
14360 (xtensa_hard_regno_mode_ok_p): ...this and make static.
14361 (xtensa_option_override): Update accordingly.
14362 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14363 (xtensa_hard_regno_mode_ok): New function.
14364 * system.h (HARD_REGNO_MODE_OK): Poison.
14366 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14367 Alan Hayward <alan.hayward@arm.com>
14368 David Sherwood <david.sherwood@arm.com>
14370 * target.def (hard_regno_call_part_clobbered): New hook.
14371 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
14372 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
14373 * doc/tm.texi: Regenerate.
14374 * hooks.h (hook_bool_uint_mode_false): Declare.
14375 * hooks.c (hook_bool_uint_mode_false): New function.
14376 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14377 * cselib.c (cselib_process_insn): Use
14378 targetm.hard_regno_call_part_clobbered instead of
14379 HARD_REGNO_CALL_PART_CLOBBERED.
14380 * ira-conflicts.c (ira_build_conflicts): Likewise.
14381 * ira-costs.c (ira_tune_allocno_costs): Likewise.
14382 * lra-constraints.c (need_for_call_save_p): Likewise.
14383 * lra-lives.c: Include target.h.
14384 (check_pseudos_live_through_calls): Use
14385 targetm.hard_regno_call_part_clobbered instead of
14386 HARD_REGNO_CALL_PART_CLOBBERED.
14387 * regcprop.c: Include target.h.
14388 (copyprop_hardreg_forward_1): Use
14389 targetm.hard_regno_call_part_clobbered instead of
14390 HARD_REGNO_CALL_PART_CLOBBERED.
14391 * reginfo.c (choose_hard_reg_mode): Likewise.
14392 * regrename.c (check_new_reg_p): Likewise.
14393 * reload.c (find_equiv_reg): Likewise.
14394 * reload1.c (emit_reload_insns): Likewise.
14395 * sched-deps.c (deps_analyze_insn): Likewise.
14396 * sel-sched.c (init_regs_for_mode): Likewise.
14397 (mark_unavailable_hard_regs): Likewise.
14398 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
14399 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14400 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
14402 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14403 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14404 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
14406 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
14408 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14409 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14410 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
14412 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14413 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14414 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
14416 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14417 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
14419 * config/powerpcspe/powerpcspe.c
14420 (rs6000_hard_regno_call_part_clobbered): New function.
14421 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14422 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14423 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
14425 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14426 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14427 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
14429 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14430 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14431 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
14433 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14434 Alan Hayward <alan.hayward@arm.com>
14435 David Sherwood <david.sherwood@arm.com>
14437 * rtl.h (subreg_memory_offset): Declare.
14438 * emit-rtl.c (subreg_memory_offset): New function.
14439 * expmed.c (store_bit_field_1): Use it.
14440 * expr.c (undefined_operand_subword_p): Likewise.
14441 * simplify-rtx.c (simplify_subreg): Likewise.
14443 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
14445 PR rtl-optimization/57448
14448 * optabs.c (expand_atomic_load): Place compiler memory barriers if
14449 using atomic_load pattern.
14450 (expand_atomic_store): Likewise.
14452 2017-09-04 Jakub Jelinek <jakub@redhat.com>
14455 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
14456 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
14457 handling. Use replace_call_with_value with NULL instead of
14458 gsi_replace, unlink_stmt_vdef and release_defs.
14460 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
14463 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
14465 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14468 * rtl.h (emit_library_call_value_1): Declare.
14469 (emit_library_call): Replace declaration with a series of overloads.
14470 Remove the parameter count argument.
14471 (emit_library_call_value): Likewise.
14472 * calls.c (emit_library_call_value_1): Make global. Replace varargs
14473 with an "rtx_mode_t *".
14474 (emit_library_call_value): Delete.
14475 (emit_library_call): Likewise.
14476 * asan.c (asan_emit_stack_protection): Update calls accordingly.
14477 (asan_emit_allocas_unpoison): Likewise.
14478 * builtins.c (expand_builtin_powi): Likewise.
14479 (expand_asan_emit_allocas_unpoison): Likewise.
14480 * cfgexpand.c (expand_main_function): Likewise.
14481 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
14482 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
14483 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
14484 * config/arm/arm.c (arm_trampoline_init): Likewise.
14485 (arm_call_tls_get_addr): Likewise.
14486 (arm_expand_divmod_libfunc): Likewise.
14487 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
14488 (smulsi3_highpart): Likewise.
14489 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
14490 (c6x_expand_compare): Likewise.
14491 (c6x_expand_movmem): Likewise.
14492 * config/frv/frv.c (frv_trampoline_init): Likewise.
14493 * config/i386/i386.c (ix86_trampoline_init): Likewise.
14494 (ix86_expand_divmod_libfunc): Likewise.
14495 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
14496 (ia64_expand_compare): Likewise.
14497 (ia64_profile_hook): Likewise.
14498 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
14499 (nonlocal_goto): Likewise.
14500 (restore_stack_nonlocal): Likewise.
14501 * config/m32r/m32r.c (block_move_call): Likewise.
14502 (m32r_trampoline_init): Likewise.
14503 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
14504 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
14505 (m68k_call_m68k_read_tp): Likewise.
14506 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
14507 (microblaze_expand_divide): Likewise.
14508 * config/mips/mips.h (mips_args): Likewise.
14509 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
14510 (MIPS_ICACHE_SYNC): Likewise.
14511 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
14512 (nios2_trampoline_init): Likewise.
14513 * config/pa/pa.c (hppa_tls_call): Likewise.
14514 (pa_trampoline_init): Likewise.
14515 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
14516 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
14517 (expand_strn_compare): Likewise.
14518 (rs6000_generate_compare): Likewise.
14519 (rs6000_expand_float128_convert): Likewise.
14520 (output_profile_hook): Likewise.
14521 (rs6000_trampoline_init): Likewise.
14522 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
14523 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
14524 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
14525 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
14526 (rs6000_generate_compare): Likewise.
14527 (rs6000_expand_float128_convert): Likewise.
14528 (output_profile_hook): Likewise.
14529 (rs6000_trampoline_init): Likewise.
14530 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
14531 * config/sh/sh.c (sh_trampoline_init): Likewise.
14532 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
14533 (sparc_emit_float_lib_cmp): Likewise.
14534 (sparc32_initialize_trampoline): Likewise.
14535 (sparc64_initialize_trampoline): Likewise.
14536 (sparc_profile_hook): Likewise.
14537 * config/spu/spu.c (ea_load_store): Likewise.
14538 * config/spu/spu.md (floatunssidf2): Likewise.
14539 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
14540 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
14541 * config/visium/visium.c (expand_block_move_4): Likewise.
14542 (expand_block_move_2): Likewise.
14543 (expand_block_move_1): Likewise.
14544 (expand_block_set_4): Likewise.
14545 (expand_block_set_2): Likewise.
14546 (expand_block_set_1): Likewise.
14547 (visium_trampoline_init): Likewise.
14548 (visium_profile_hook): Likewise.
14549 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
14550 (xtensa_setup_frame_addresses): Likewise.
14551 (xtensa_trampoline_init): Likewise.
14552 * except.c (sjlj_emit_function_enter): Likewise.
14553 (sjlj_emit_function_exit): Likewise.
14554 * explow.c (allocate_dynamic_stack_space): Likewise.
14555 (probe_stack_range): Likewise.
14556 * expr.c (convert_mode_scalar): Likewise.
14557 * optabs.c (expand_binop): Likewise.
14558 (expand_twoval_binop_libfunc): Likewise.
14559 (expand_unop): Likewise.
14560 (prepare_cmp_insn): Likewise.
14561 (prepare_float_lib_cmp): Likewise.
14562 (expand_float): Likewise.
14563 (expand_fix): Likewise.
14564 (expand_fixed_convert): Likewise.
14565 (maybe_emit_sync_lock_test_and_set): Likewise.
14566 (expand_atomic_compare_and_swap): Likewise.
14567 (expand_mem_thread_fence): Likewise.
14568 (expand_atomic_fetch_op): Likewise.
14570 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
14572 * doc/generic.texi (OpenACC): Adjust URL.
14573 * doc/invoke.texi (C Dialect Options): Ditto.
14575 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
14577 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
14578 predicate for operand 1. Add (m,<S>) constraint.
14579 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
14580 Prevent memory operand 1 with register operand 2.
14582 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
14584 PR rtl-optimization/82024
14585 * combine.c (try_combine): If the combination result is a PARALLEL,
14586 and we only need to retain the SET in there that would be placed
14587 at I2, check that we can place that at I3 instead, before doing so.
14589 2017-09-01 Jakub Jelinek <jakub@redhat.com>
14592 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
14594 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
14595 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
14598 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
14599 Jeff Law <law@redhat.com>
14601 * varasm.c (bss_initializer_p): Do not put constants into .bss
14602 (categorize_decl_for_section): Handle bss_initializer_p returning
14603 false when DECL_INITIAL is NULL.
14605 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14608 * config/s390/s390.c (s390_can_inline_p): New function.
14610 2017-09-01 Jeff Law <law@redhat.com>
14612 PR tree-optimization/82052
14613 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
14614 Always initialize the returned slot after a hash table miss
14615 when INSERT is true.
14617 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
14619 * config/s390/s390.md (mem_signal_fence): Remove.
14620 * doc/md.texi (mem_signal_fence): Remove.
14621 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
14623 * target-insns.def (mem_signal_fence): Remove.
14625 2017-09-01 Jakub Jelinek <jakub@redhat.com>
14628 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
14631 * asan.c (create_odr_indicator): Strip name encoding from assembler
14632 name before appending it after __odr_asan_.
14634 2017-09-01 Martin Liska <mliska@suse.cz>
14636 PR tree-optimization/82059
14637 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
14638 frequency only when an edge is redirected.
14640 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14642 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
14643 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
14644 (arc_conditional_register_usage): Remove ARC600 lp_count
14646 (arc_file_start): Emit Tag_ARC_CPU_variation.
14647 (arc_can_use_doloop_p): New conditions to use ZOLs.
14648 (hwloop_fail): New function.
14649 (hwloop_optimize): Likewise.
14650 (hwloop_pattern_reg): Likewise.
14651 (arc_doloop_hooks): New struct, to be used with reorg_loops.
14652 (arc_reorg_loops): New function, calls reorg_loops.
14653 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
14654 (arc600_corereg_hazard): Remove ZOL checking, case handled by
14656 (arc_loop_hazard): Remove function, functionality moved into
14658 (arc_hazard): Remove arc_loop_hazard call.
14659 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
14660 into hwloop_optimize.
14661 (arc_label_align): Remove ZOL handling.
14662 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
14663 * config/arc/arc.md (doloop_begin): Remove pattern.
14664 (doloop_begin_i): Likewise.
14665 (doloop_end_i): Likewise.
14666 (doloop_fallback): Likewise.
14667 (doloop_fallback_m): Likewise.
14668 (doloop_end): Reimplement expand.
14669 (arc_lp): New pattern for LP instruction.
14670 (loop_end): New pattern.
14671 (loop_fail): Likewise.
14672 (decrement_and_branch_until_zero): Likewise.
14673 * config/arc/arc.opt (mlpc-width): New option.
14674 * doc/invoke.texi (mlpc-width): Document option.
14676 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14678 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
14679 (arc_ccfsm_advance): Fix checking for delay slots.
14680 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
14682 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14684 * config/arc/arc.md (movqi_insn): Add stores to save constant long
14686 (movhi_insn): Update store instruction constraint which are saving
14687 6-bit short immediates.
14688 (movsi_insn): Consider also short scaled load operations.
14689 (zero_extendhisi2_i): Use Usd constraint instead of T.
14690 (extendhisi2_i): Add q constraint.
14691 (arc_clzsi2): Add type and length attributes.
14692 (arc_ctzsi2): Likewise.
14693 * config/arc/constraints.md (Usc): Update constraint, the
14694 assembler can parse two relocations for a single instruction.
14696 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14698 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
14699 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
14701 2017-08-31 Olivier Hainque <hainque@adacore.com>
14703 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
14704 match as powerpc-wrs-vxworks*.
14706 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
14708 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
14709 register constraint for by-element operand.
14710 (aarch64_mls_elt_merge<mode>): Likewise.
14712 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14714 * config/arc/arc.c (arc_can_follow_jump): Check for short
14717 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14719 * config.gcc: Use g.opt for arc.
14720 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
14721 functionality moved to ...
14722 (legitimate_scaled_address_p): New function, ...here.
14723 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
14724 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
14725 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
14727 (arc_override_options): Handle G option.
14728 (arc_output_pic_addr_const): Correct function definition.
14729 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
14730 (arc_decl_anon_ns_mem_p): Delete.
14731 (arc_in_small_data_p): Overhaul this function to take into
14732 consideration the value given via G option.
14733 (arc_rewrite_small_data_1): Renamed and corrected old
14734 arc_rewrite_small_data function.
14735 (arc_rewrite_small_data): New function.
14736 (small_data_pattern): Don't use pic_offset_table_rtx.
14737 * config/arc/arc.h (CC1_SPEC): Recognize G option.
14738 * config/arc/simdext.md (movmisalignv2hi): Use
14739 prepare_move_operands function.
14741 (movmisalign*): Likewise.
14742 * doc/invoke.texi (ARC options): Document -G option.
14744 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14746 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
14748 * config/arc/arc.c (arc_print_operand): Output scalled address for
14749 sdata whenever is possible.
14750 (arc_in_small_data_p): Allow sdata for 64bit datum when double
14751 load/stores are available.
14752 (compact_sda_memory_operand): Check for the alignment required by
14753 code density instructions.
14754 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
14756 * config/arc/constraints.md (Usd): Update constraint.
14757 (Us0): New constraint.
14758 (Usc): Update constraint.
14760 2017-08-31 Richard Biener <rguenther@suse.de>
14762 PR middle-end/82054
14763 * dwarf2out.c (dwarf2out_early_global_decl): Process each
14764 function only once.
14766 2017-08-31 Tamar Christina <tamar.christina@arm.com>
14768 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
14769 Resize type_signature.
14771 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14772 Alan Hayward <alan.hayward@arm.com>
14773 David Sherwood <david.sherwood@arm.com>
14775 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
14776 subregs whose inner modes can be stored in GPRs.
14777 (aarch64_classify_index): Likewise.
14779 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14780 Alan Hayward <alan.hayward@arm.com>
14781 David Sherwood <david.sherwood@arm.com>
14783 * config/aarch64/iterators.md (V_cmp_result): Rename to...
14784 (V_INT_EQUIV): ...this.
14785 (v_cmp_result): Rename to...
14786 (v_int_equiv): ...this.
14787 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
14788 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
14789 (copysign<mode>3): Likewise.
14790 (aarch64_simd_bsl<mode>_internal): Likewise.
14791 (aarch64_simd_bsl<mode>): Likewise.
14792 (vec_cmp<mode><mode>): Likewise.
14793 (vcond<mode><mode>): Likewise.
14794 (vcond<v_cmp_mixed><mode>): Likewise.
14795 (vcondu<mode><v_cmp_mixed>): Likewise.
14796 (aarch64_cm<optab><mode>): Likewise.
14797 (aarch64_cmtst<mode>): Likewise.
14798 (aarch64_fac<optab><mode>): Likewise.
14799 (vec_perm_const<mode>): Likewise.
14800 (vcond_mask_<mode><v_cmp_result>): Rename to...
14801 (vcond_mask_<mode><v_int_equiv>): ...this.
14802 (vec_cmp<mode><v_cmp_result>): Rename to...
14803 (vec_cmp<mode><v_int_equiv>): ...this.
14805 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14806 Alan Hayward <alan.hayward@arm.com>
14807 David Sherwood <david.sherwood@arm.com>
14809 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
14811 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
14812 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
14813 (UNSPEC_LD4_DREG): New unspecs.
14814 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
14815 (aarch64_ld2<mode>_dreg_be): Replace with...
14816 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
14818 (aarch64_ld3<mode>_dreg_le)
14819 (aarch64_ld3<mode>_dreg_be): Replace with...
14820 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
14822 (aarch64_ld4<mode>_dreg_le)
14823 (aarch64_ld4<mode>_dreg_be): Replace with...
14824 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
14827 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14829 PR tree-optimization/81987
14830 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
14831 insert an initializer in a location not dominated by the stride
14834 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
14836 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
14837 on the entire header of the finally block in the fallthru case.
14839 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
14841 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
14843 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
14845 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
14846 rs6000_emit_move_from_cr and call renamed function.
14847 (rs6000_emit_prologue): Call renamed functions.
14848 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
14849 movesi_from_cr, remove volatile CRs.
14851 2017-08-30 Jon Beniston <jon@beniston.com>
14852 Richard Biener <rguenther@suse.de>
14854 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
14855 of VECTOR_MODE_P check.
14856 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
14857 element vector types.
14859 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14861 * df.h (df_read_modify_subreg_p): Remove in favor of...
14862 * rtl.h (read_modify_subreg_p): ...this new function. Take a
14863 const_rtx instead of an rtx.
14864 * cprop.c (local_cprop_find_used_regs): Update accordingly.
14865 * df-problems.c (df_word_lr_mark_ref): Likewise.
14866 * ira-lives.c (mark_pseudo_reg_live): Likewise.
14867 (mark_pseudo_reg_dead): Likewise.
14868 (mark_ref_dead): Likewise.
14869 * reginfo.c (init_subregs_of_mode): Likewise.
14870 * sched-deps.c (sched_analyze_1): Likewise.
14871 * df-scan.c (df_def_record_1): Likewise.
14872 (df_uses_record): Likewise.
14873 (df_read_modify_subreg_p): Remove in favor of...
14874 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
14875 const_rtx instead of an rtx.
14877 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14878 Alan Hayward <alan.hayward@arm.com>
14879 David Sherwood <david.sherwood@arm.com>
14881 * rtl.h (partial_subreg_p): New function.
14882 * caller-save.c (save_call_clobbered_regs): Use it.
14883 * calls.c (expand_call): Likewise.
14884 * combine.c (combinable_i3pat): Likewise.
14885 (simplify_set): Likewise.
14886 (make_extraction): Likewise.
14887 (make_compound_operation_int): Likewise.
14888 (gen_lowpart_or_truncate): Likewise.
14889 (force_to_mode): Likewise.
14890 (make_field_assignment): Likewise.
14891 (reg_truncated_to_mode): Likewise.
14892 (record_truncated_value): Likewise.
14893 (move_deaths): Likewise.
14894 * cse.c (record_jump_cond): Likewise.
14895 (cse_insn): Likewise.
14896 * cselib.c (cselib_lookup_1): Likewise.
14897 * expmed.c (extract_bit_field_using_extv): Likewise.
14898 * function.c (assign_parm_setup_reg): Likewise.
14899 * ifcvt.c (noce_convert_multiple_sets): Likewise.
14900 * ira-build.c (create_insn_allocnos): Likewise.
14901 * lra-coalesce.c (merge_pseudos): Likewise.
14902 * lra-constraints.c (match_reload): Likewise.
14903 (simplify_operand_subreg): Likewise.
14904 (curr_insn_transform): Likewise.
14905 * lra-lives.c (process_bb_lives): Likewise.
14906 * lra.c (new_insn_reg): Likewise.
14907 (lra_substitute_pseudo): Likewise.
14908 * regcprop.c (mode_change_ok): Likewise.
14909 (maybe_mode_change): Likewise.
14910 (copyprop_hardreg_forward_1): Likewise.
14911 * reload.c (push_reload): Likewise.
14912 (find_reloads): Likewise.
14913 (find_reloads_subreg_address): Likewise.
14914 * reload1.c (alter_reg): Likewise.
14915 (eliminate_regs_1): Likewise.
14916 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14918 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
14920 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
14921 back to if statements, including unpack.
14923 2017-08-30 Martin Liska <mliska@suse.cz>
14925 PR inline-asm/82001
14926 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
14928 (func_checker::compare_asm_inputs_outputs): ... this function.
14929 (func_checker::compare_gimple_asm): Use the function to compare
14930 also ASM constrains.
14931 * ipa-icf-gimple.h: Rename the function.
14933 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14934 Alan Hayward <alan.hayward@arm.com>
14935 David Sherwood <david.sherwood@arm.com>
14937 * coretypes.h (complex_mode): New type.
14938 * gdbhooks.py (build_pretty_printer): Handle it.
14939 * machmode.h (complex_mode): New class.
14940 (complex_mode::includes_p): New function.
14941 (is_complex_int_mode): Likewise.
14942 (is_complex_float_mode): Likewise.
14943 * genmodes.c (get_mode_class): Handle complex mode classes.
14944 * function.c (expand_function_end): Use is_complex_int_mode.
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 (scalar_mode_pod): New typedef.
14951 * gdbhooks.py (build_pretty_printer): Handle it.
14952 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
14953 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
14954 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
14955 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
14956 as_a <scalar_mode>.
14958 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14959 Alan Hayward <alan.hayward@arm.com>
14960 David Sherwood <david.sherwood@arm.com>
14962 * machmode.h (mode_for_vector): Take a scalar_mode instead
14964 * stor-layout.c (mode_for_vector): Likewise.
14965 * explow.c (promote_mode): Use as_a <scalar_mode>.
14966 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
14968 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14969 Alan Hayward <alan.hayward@arm.com>
14970 David Sherwood <david.sherwood@arm.com>
14972 * target.def (preferred_simd_mode): Take a scalar_mode
14973 instead of a machine_mode.
14974 * targhooks.h (default_preferred_simd_mode): Likewise.
14975 * targhooks.c (default_preferred_simd_mode): Likewise.
14976 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
14977 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
14978 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
14979 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
14980 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
14981 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
14982 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
14983 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
14985 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
14986 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
14987 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
14988 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
14989 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
14990 * doc/tm.texi: Regenerate.
14991 * optabs-query.c (can_vec_mask_load_store_p): Return false for
14994 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14995 Alan Hayward <alan.hayward@arm.com>
14996 David Sherwood <david.sherwood@arm.com>
14998 * target.def (scalar_mode_supported_p): Take a scalar_mode
14999 instead of a machine_mode.
15000 * targhooks.h (default_scalar_mode_supported_p): Likewise.
15001 * targhooks.c (default_scalar_mode_supported_p): Likewise.
15002 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
15003 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
15004 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
15005 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
15006 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
15007 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
15008 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
15009 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
15010 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
15011 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
15012 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
15013 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
15015 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
15016 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
15017 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
15018 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
15019 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
15021 * doc/tm.texi: Regenerate.
15023 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15024 Alan Hayward <alan.hayward@arm.com>
15025 David Sherwood <david.sherwood@arm.com>
15027 * coretypes.h (opt_scalar_mode): New typedef.
15028 * gdbhooks.py (build_pretty_printers): Handle it.
15029 * machmode.h (mode_iterator::get_2xwider): Add overload for
15031 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
15033 * expr.c (convert_mode_scalar): Likewise.
15034 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15035 * optabs.c (expand_float): Likewise.
15036 (expand_fix): Likewise.
15037 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15039 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15040 Alan Hayward <alan.hayward@arm.com>
15041 David Sherwood <david.sherwood@arm.com>
15043 * optabs.c (expand_float): Explicitly check for scalars before
15044 using a branching expansion.
15045 (expand_fix): Likewise.
15047 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15048 Alan Hayward <alan.hayward@arm.com>
15049 David Sherwood <david.sherwood@arm.com>
15051 * expr.c (convert_mode): Split scalar handling out into...
15052 (convert_mode_scalar): ...this new function. Treat the modes
15055 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15056 Alan Hayward <alan.hayward@arm.com>
15057 David Sherwood <david.sherwood@arm.com>
15059 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
15061 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
15063 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15064 Alan Hayward <alan.hayward@arm.com>
15065 David Sherwood <david.sherwood@arm.com>
15067 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
15068 rather than a machine_mode.
15069 (fixed_from_string): Likewise.
15070 (fixed_convert): Likewise.
15071 (fixed_convert_from_int): Likewise.
15072 (fixed_convert_from_real): Likewise.
15073 (real_convert_from_fixed): Likewise.
15074 * fixed-value.c (fixed_from_double_int): Likewise.
15075 (fixed_from_string): Likewise.
15076 (fixed_convert): Likewise.
15077 (fixed_convert_from_int): Likewise.
15078 (fixed_convert_from_real): Likewise.
15079 (real_convert_from_fixed): Likewise.
15080 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
15082 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15083 Alan Hayward <alan.hayward@arm.com>
15084 David Sherwood <david.sherwood@arm.com>
15086 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
15087 of separate mode class checks. Do not allow vector modes here.
15088 (immed_wide_int_const): Use as_a <scalar_mode>.
15089 * explow.c (trunc_int_for_mode): Likewise.
15090 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
15091 (wi::shwi): Likewise.
15092 (wi::min_value): Likewise.
15093 (wi::max_value): Likewise.
15094 * dwarf2out.c (loc_descriptor): Likewise.
15095 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
15096 for CONST_WIDE_INT.
15098 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15099 Alan Hayward <alan.hayward@arm.com>
15100 David Sherwood <david.sherwood@arm.com>
15102 * tree.h (SCALAR_TYPE_MODE): New macro.
15103 * expr.c (expand_expr_addr_expr_1): Use it.
15104 (expand_expr_real_2): Likewise.
15105 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
15106 (fold_convert_const_fixed_from_int): Likewise.
15107 (fold_convert_const_fixed_from_real): Likewise.
15108 (native_encode_fixed): Likewise
15109 (native_encode_complex): Likewise
15110 (native_encode_vector): Likewise.
15111 (native_interpret_fixed): Likewise.
15112 (native_interpret_real): Likewise.
15113 (native_interpret_complex): Likewise.
15114 (native_interpret_vector): Likewise.
15115 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
15116 (simd_clone_adjust_argument_types): Likewise.
15117 (simd_clone_init_simd_arrays): Likewise.
15118 (simd_clone_adjust): Likewise.
15119 * stor-layout.c (layout_type): Likewise.
15120 * tree.c (build_minus_one_cst): Likewise.
15121 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15122 * tree-inline.c (estimate_move_cost): Likewise.
15123 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
15124 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
15125 (vectorizable_reduction): Likewise.
15126 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
15127 (vect_recog_mixed_size_cond_pattern): Likewise.
15128 (check_bool_pattern): Likewise.
15129 (adjust_bool_pattern): Likewise.
15130 (search_type_for_mask_1): Likewise.
15131 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
15132 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15133 (vectorizable_load): Likewise.
15134 (vectorizable_store): Likewise.
15135 * ubsan.c (ubsan_encode_value): Likewise.
15136 * varasm.c (output_constant): Likewise.
15138 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15139 Alan Hayward <alan.hayward@arm.com>
15140 David Sherwood <david.sherwood@arm.com>
15142 * coretypes.h (scalar_mode): New class.
15143 * machmode.h (scalar_mode): Likewise.
15144 (scalar_mode::includes_p): New function.
15145 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
15146 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
15147 * genmodes.c (get_mode_class): Handle remaining scalar modes.
15148 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
15149 * expmed.c (store_bit_field_1): Likewise.
15150 (extract_bit_field_1): Likewise.
15151 * expr.c (write_complex_part): Likewise.
15152 (read_complex_part): Likewise.
15153 (emit_move_complex_push): Likewise.
15154 (expand_expr_real_2): Likewise.
15155 * function.c (assign_parm_setup_reg): Likewise.
15156 (assign_parms_unsplit_complex): Likewise.
15157 * optabs.c (expand_binop): Likewise.
15158 * rtlanal.c (subreg_get_info): Likewise.
15159 * simplify-rtx.c (simplify_immed_subreg): Likewise.
15160 * varasm.c (output_constant_pool_2): Likewise.
15162 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15163 Alan Hayward <alan.hayward@arm.com>
15164 David Sherwood <david.sherwood@arm.com>
15166 * expmed.c (extract_high_half): Use scalar_int_mode and remove
15168 (expmed_mult_highpart_optab): Likewise.
15169 (expmed_mult_highpart): Likewise.
15171 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15172 Alan Hayward <alan.hayward@arm.com>
15173 David Sherwood <david.sherwood@arm.com>
15175 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
15176 instead of a machine_mode.
15177 (builtin_memset_read_str): Likewise.
15178 * builtins.c (c_readstr): Likewise.
15179 (builtin_memcpy_read_str): Likewise.
15180 (builtin_strncpy_read_str): Likewise.
15181 (builtin_memset_read_str): Likewise.
15182 (builtin_memset_gen_str): Likewise.
15183 (expand_builtin_signbit): Use scalar_int_mode for local variables.
15184 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
15185 instead of a machine_mode.
15186 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
15188 (make_extraction): Likewise.
15189 (try_widen_shift_mode): Take and return scalar_int_modes instead
15191 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
15192 a scalar_int_mode instead of a machine_mode.
15193 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
15194 (avr_addr_space_pointer_mode): Likewise.
15195 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
15196 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
15197 (msp430_unwind_word_mode): Likewise.
15198 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
15199 (spu_addr_space_pointer_mode): Likewise.
15200 (spu_addr_space_address_mode): Likewise.
15201 (spu_libgcc_cmp_return_mode): Likewise.
15202 (spu_libgcc_shift_count_mode): Likewise.
15203 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
15204 (rl78_addr_space_pointer_mode): Likewise.
15205 (fl78_unwind_word_mode): Likewise.
15206 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
15208 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
15209 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
15210 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
15211 (mips_valid_pointer_mode): Likewise.
15212 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
15213 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
15214 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
15216 (ft32_addr_space_address_mode): Likewise.
15217 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
15218 scalar_int_mode instead of a machine_mode.
15219 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
15221 (m32c_addr_space_address_mode): Likewise.
15222 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
15223 (rs6000_eh_return_filter_mode): Likewise.
15224 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
15225 (rs6000_eh_return_filter_mode): Likewise.
15226 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
15227 (s390_libgcc_shift_count_mode): Likewise.
15228 (s390_unwind_word_mode): Likewise.
15229 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
15231 * target.def (mode_rep_extended): Likewise.
15232 (valid_pointer_mode): Likewise.
15233 (addr_space.valid_pointer_mode): Likewise.
15234 (eh_return_filter_mode): Return a scalar_int_mode rather than
15236 (libgcc_cmp_return_mode): Likewise.
15237 (libgcc_shift_count_mode): Likewise.
15238 (unwind_word_mode): Likewise.
15239 (addr_space.pointer_mode): Likewise.
15240 (addr_space.address_mode): Likewise.
15241 * doc/tm.texi: Regenerate.
15242 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
15244 (do_jump): Use scalar_int_mode for local variables.
15245 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
15246 rather than a machine_mode.
15247 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
15248 (scompare_loc_descriptor_wide): Likewise.
15249 (scompare_loc_descriptor_narrow): Likewise.
15250 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
15252 * except.c (sjlj_emit_dispatch_table): Likewise.
15253 (expand_builtin_eh_copy_values): Likewise.
15254 * explow.c (convert_memory_address_addr_space_1): Likewise.
15255 Take a scalar_int_mode rather than a machine_mode.
15256 (convert_memory_address_addr_space): Take a scalar_int_mode rather
15257 than a machine_mode.
15258 (memory_address_addr_space): Use scalar_int_mode for local variables.
15259 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
15260 rather than a machine_mode.
15261 * expmed.c (mask_rtx): Likewise.
15262 (init_expmed_one_conv): Likewise.
15263 (expand_mult_highpart_adjust): Likewise.
15264 (extract_high_half): Likewise.
15265 (expmed_mult_highpart_optab): Likewise.
15266 (expmed_mult_highpart): Likewise.
15267 (expand_smod_pow2): Likewise.
15268 (expand_sdiv_pow2): Likewise.
15269 (emit_store_flag_int): Likewise.
15270 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
15272 (extract_low_bits): Likewise.
15273 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
15275 * expr.c (pieces_addr::adjust): Likewise.
15276 (can_store_by_pieces): Likewise.
15277 (store_by_pieces): Likewise.
15278 (clear_by_pieces_1): Likewise.
15279 (expand_expr_addr_expr_1): Likewise.
15280 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
15281 (expand_expr_real_1): Likewise.
15282 (try_casesi): Likewise.
15283 * final.c (shorten_branches): Likewise.
15284 * fold-const.c (fold_convert_const_int_from_fixed): Change the
15285 type of "mode" to machine_mode.
15286 * internal-fn.c (expand_arith_overflow_result_store): Take a
15287 scalar_int_mode rather than a machine_mode.
15288 (expand_mul_overflow): Use scalar_int_mode for local variables.
15289 * loop-doloop.c (doloop_modify): Likewise.
15290 (doloop_optimize): Likewise.
15291 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
15292 than a machine_mode.
15293 (expand_doubleword_shift_condmove): Likewise.
15294 (expand_doubleword_shift): Likewise.
15295 (expand_doubleword_clz): Likewise.
15296 (expand_doubleword_popcount): Likewise.
15297 (expand_doubleword_parity): Likewise.
15298 (expand_absneg_bit): Use scalar_int_mode for local variables.
15299 (prepare_float_lib_cmp): Likewise.
15300 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
15301 rather than a machine_mode.
15302 (convert_memory_address_addr_space): Likewise.
15303 (get_mode_bounds): Likewise.
15304 (get_address_mode): Return a scalar_int_mode rather than a
15306 * rtlanal.c (get_address_mode): Likewise.
15307 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
15308 than a machine_mode.
15309 * targhooks.c (default_mode_rep_extended): Likewise.
15310 (default_valid_pointer_mode): Likewise.
15311 (default_addr_space_valid_pointer_mode): Likewise.
15312 (default_eh_return_filter_mode): Return a scalar_int_mode rather
15313 than a machine_mode.
15314 (default_libgcc_cmp_return_mode): Likewise.
15315 (default_libgcc_shift_count_mode): Likewise.
15316 (default_unwind_word_mode): Likewise.
15317 (default_addr_space_pointer_mode): Likewise.
15318 (default_addr_space_address_mode): Likewise.
15319 * targhooks.h (default_eh_return_filter_mode): Likewise.
15320 (default_libgcc_cmp_return_mode): Likewise.
15321 (default_libgcc_shift_count_mode): Likewise.
15322 (default_unwind_word_mode): Likewise.
15323 (default_addr_space_pointer_mode): Likewise.
15324 (default_addr_space_address_mode): Likewise.
15325 (default_mode_rep_extended): Take a scalar_int_mode rather than
15327 (default_valid_pointer_mode): Likewise.
15328 (default_addr_space_valid_pointer_mode): Likewise.
15329 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
15331 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
15332 rather than a machine_mode.
15333 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
15334 for local variables.
15335 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15336 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
15337 than a machine_mode.
15339 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15340 Alan Hayward <alan.hayward@arm.com>
15341 David Sherwood <david.sherwood@arm.com>
15343 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
15344 the mode argument to scalar_int_mode.
15345 (do_jump_by_parts_zero_rtx): Likewise.
15346 (do_jump_by_parts_equality_rtx): Likewise.
15347 (do_jump_by_parts_greater): Take a mode argument.
15348 (do_jump_by_parts_equality): Likewise.
15349 (do_jump_1): Update calls accordingly.
15351 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15352 Alan Hayward <alan.hayward@arm.com>
15353 David Sherwood <david.sherwood@arm.com>
15355 * is-a.h (safe_dyn_cast): New function.
15356 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
15357 (jump_table_for_label): Likewise.
15358 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
15359 instead of an rtx_insn *.
15360 (shorten_branches): Use dyn_cast instead of LABEL_P and
15361 JUMP_TABLE_DATA_P. Use jump_table_for_label and
15362 rtx_jump_table_data::get_data_mode.
15363 (final_scan_insn): Likewise.
15365 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15366 Alan Hayward <alan.hayward@arm.com>
15367 David Sherwood <david.sherwood@arm.com>
15369 * combine.c (try_combine): Use is_a <scalar_int_mode> when
15370 trying to combine a full-register integer set with a subreg
15373 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15374 Alan Hayward <alan.hayward@arm.com>
15375 David Sherwood <david.sherwood@arm.com>
15377 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
15378 that is always either address_mode or pointer_mode.
15380 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15381 Alan Hayward <alan.hayward@arm.com>
15382 David Sherwood <david.sherwood@arm.com>
15384 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
15385 when the two are known to be equal.
15387 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15388 Alan Hayward <alan.hayward@arm.com>
15389 David Sherwood <david.sherwood@arm.com>
15391 * simplify-rtx.c (simplify_const_unary_operation): Use
15392 is_a <scalar_int_mode> instead of checking for a nonzero
15393 precision. Forcibly convert op_mode to a scalar_int_mode
15394 in that case. More clearly differentiate the operand and
15395 result modes and use the former when deciding what the value
15396 of a count-bits operation should be. Use is_int_mode instead
15397 of checking for a MODE_INT. Remove redundant check for whether
15398 this mode has a zero precision.
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 * optabs.c (widen_leading): Change the type of the mode argument
15405 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
15406 (widen_bswap): Likewise.
15407 (expand_parity): Likewise.
15408 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
15409 (expand_ffs): Likewise.
15410 (epand_unop): Check for scalar integer modes before calling the
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 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
15419 (expand_expr_real_1): Update call accordingly.
15421 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15422 Alan Hayward <alan.hayward@arm.com>
15423 David Sherwood <david.sherwood@arm.com>
15425 * expmed.c (store_bit_field_using_insv): Add op0_mode and
15426 value_mode arguments. Use scalar_int_mode internally.
15427 (store_bit_field_1): Rename the new integer mode from imode
15428 to op0_mode and use it instead of GET_MODE (op0). Update calls
15429 to store_split_bit_field, store_bit_field_using_insv and
15430 store_fixed_bit_field.
15431 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
15432 Use scalar_int_mode internally. Use a bit count rather than a mode
15433 when calculating the largest bit size for get_best_mode.
15434 Update calls to store_split_bit_field and store_fixed_bit_field_1.
15435 (store_fixed_bit_field_1): Add mode and value_mode arguments.
15436 Remove assertion that OP0 has a scalar integer mode.
15437 (store_split_bit_field): Add op0_mode and value_mode arguments.
15438 Update calls to extract_fixed_bit_field.
15439 (extract_bit_field_using_extv): Add an op0_mode argument.
15440 Use scalar_int_mode internally.
15441 (extract_bit_field_1): Rename the new integer mode from imode to
15442 op0_mode and use it instead of GET_MODE (op0). Update calls to
15443 extract_split_bit_field, extract_bit_field_using_extv and
15444 extract_fixed_bit_field.
15445 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
15446 to extract_split_bit_field and extract_fixed_bit_field_1.
15447 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
15448 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
15449 on the target mode.
15450 (extract_split_bit_field): Add an op0_mode argument. Update call
15451 to extract_fixed_bit_field.
15453 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15454 Alan Hayward <alan.hayward@arm.com>
15455 David Sherwood <david.sherwood@arm.com>
15457 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
15458 * explow.c (hard_function_value): Likewise.
15459 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
15460 convert_to_mode call outside the loop.
15461 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
15462 for the mode iterator. Require the mode specified by max_pieces
15464 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
15466 (copy_blkmode_to_reg): Likewise.
15467 (set_storage_via_setmem): Likewise.
15468 * optabs.c (prepare_cmp_insn): Likewise.
15469 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15470 * stor-layout.c (finish_bitfield_representative): Likewise.
15472 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15473 Alan Hayward <alan.hayward@arm.com>
15474 David Sherwood <david.sherwood@arm.com>
15476 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
15477 * expr.c (convert_move): Use them.
15478 (convert_modes): Likewise.
15479 (store_expr_with_bounds): Likewise.
15481 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15482 Alan Hayward <alan.hayward@arm.com>
15483 David Sherwood <david.sherwood@arm.com>
15485 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
15486 parameter for the mode of "x". Remove the "known_x", "known_mode"
15487 and "known_ret" arguments. Change the type of the mode argument
15488 to scalar_int_mode.
15489 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
15490 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
15491 (reg_num_sign_bit_copies_for_combine): Likewise.
15492 * rtlanal.c (nonzero_bits1): Likewise.
15493 (num_sign_bit_copies1): Likewise.
15494 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
15495 (reg_num_sign_bit_copies_general): Likewise.
15496 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
15497 (reg_nonzero_bits_general): Likewise.
15499 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15500 Alan Hayward <alan.hayward@arm.com>
15501 David Sherwood <david.sherwood@arm.com>
15503 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
15504 than in subroutines. Return 1 for non-integer modes.
15505 (cached_num_sign_bit_copies): Change the type of the mode parameter
15506 to scalar_int_mode.
15507 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
15508 classes. Handle CONST_INT_P first and then check whether X also
15509 has a scalar integer mode. Check the same thing for inner registers
15510 of a SUBREG and for values that are being extended or truncated.
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 (nonzero_bits): Handle VOIDmode here rather than
15517 in subroutines. Return the mode mask for non-integer modes.
15518 (cached_nonzero_bits): Change the type of the mode parameter
15519 to scalar_int_mode.
15520 (nonzero_bits1): Likewise. Remove early exit for other mode
15521 classes. Handle CONST_INT_P first and then check whether X
15522 also has a scalar integer mode.
15524 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15525 Alan Hayward <alan.hayward@arm.com>
15526 David Sherwood <david.sherwood@arm.com>
15528 * expr.c (widest_int_mode_for_size): Make the comment match the code.
15529 Return a scalar_int_mode and assert that the size is greater than
15531 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
15532 (op_by_pieces_d::op_by_pieces_d): Likewise.
15533 (op_by_pieces_d::run): Likewise.
15534 (can_store_by_pieces): Likewise.
15536 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15537 Alan Hayward <alan.hayward@arm.com>
15538 David Sherwood <david.sherwood@arm.com>
15540 * combine.c (extract_left_shift): Add a mode argument and update
15542 (make_compound_operation_int): Change the type of the mode parameter
15543 to scalar_int_mode and update the call to extract_left_shift.
15545 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15546 Alan Hayward <alan.hayward@arm.com>
15547 David Sherwood <david.sherwood@arm.com>
15549 * combine.c (simplify_and_const_int): Change the type of the mode
15550 parameter to scalar_int_mode.
15551 (simplify_and_const_int_1): Likewise. Update recursive call.
15553 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15554 Alan Hayward <alan.hayward@arm.com>
15555 David Sherwood <david.sherwood@arm.com>
15557 * combine.c (simplify_compare_const): Check that the mode is a
15558 scalar_int_mode (rather than VOIDmode) before testing its
15560 (simplify_comparison): Move COMPARISON_P handling out of the
15561 loop and restrict the latter part of the loop to scalar_int_modes.
15562 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
15563 and when considering SUBREG_REGs. Use is_int_mode instead of
15564 checking GET_MODE_CLASS against MODE_INT.
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 (try_widen_shift_mode): Move check for equal modes to...
15571 (simplify_shift_const_1): ...here. Use scalar_int_mode for
15572 shift_unit_mode and for modes involved in scalar shifts.
15574 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15575 Alan Hayward <alan.hayward@arm.com>
15576 David Sherwood <david.sherwood@arm.com>
15578 * combine.c (force_int_to_mode): New function, split out from...
15579 (force_to_mode): ...here. Keep xmode up-to-date and use it
15580 instead of GET_MODE (x).
15582 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15583 Alan Hayward <alan.hayward@arm.com>
15584 David Sherwood <david.sherwood@arm.com>
15586 * optabs-query.h (extraction_insn::struct_mode): Change type to
15587 opt_scalar_int_mode and update comment.
15588 (extraction_insn::field_mode): Change type to scalar_int_mode.
15589 (extraction_insn::pos_mode): Likewise.
15590 * combine.c (make_extraction): Update accordingly.
15591 * optabs-query.c (get_traditional_extraction_insn): Likewise.
15592 (get_optab_extraction_insn): Likewise.
15593 * recog.c (simplify_while_replacing): Likewise.
15594 * expmed.c (narrow_bit_field_mem): Change the type of the mode
15595 parameter to opt_scalar_int_mode.
15597 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15598 Alan Hayward <alan.hayward@arm.com>
15599 David Sherwood <david.sherwood@arm.com>
15601 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
15602 to a scalar_int_mode instead of a machine_mode.
15603 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
15604 (get_best_mode): Return a boolean and use a pointer argument to store
15605 the selected mode. Replace the limit mode parameter with a bit limit.
15606 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
15607 for the values returned by bit_field_mode_iterator::next_mode.
15608 (store_bit_field): Update call to get_best_mode.
15609 (store_fixed_bit_field): Likewise.
15610 (extract_fixed_bit_field): Likewise.
15611 * expr.c (optimize_bitfield_assignment_op): Likewise.
15612 * fold-const.c (optimize_bit_field_compare): Likewise.
15613 (fold_truth_andor_1): Likewise.
15614 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
15615 Update for new type of m_mode.
15616 (get_best_mode): As above.
15618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15619 Alan Hayward <alan.hayward@arm.com>
15620 David Sherwood <david.sherwood@arm.com>
15622 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
15623 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
15624 (store_bit_field): Check is_a <scalar_int_mode> before calling
15625 strict_volatile_bitfield_p.
15626 (extract_bit_field): Likewise.
15628 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15629 Alan Hayward <alan.hayward@arm.com>
15630 David Sherwood <david.sherwood@arm.com>
15632 * target.def (cstore_mode): Return a scalar_int_mode.
15633 * doc/tm.texi: Regenerate.
15634 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
15635 * targhooks.h (default_cstore_mode): Likewise.
15636 * targhooks.c (default_cstore_mode): Likewise, using a forced
15638 * expmed.c (emit_cstore): Expect the target of the cstore to be
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 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
15647 (niter_desc): Likewise mode.
15648 (iv_analyze): Add a mode parameter.
15650 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
15651 and change its type to scalar_int_mode.
15652 * loop-iv.c: Update commentary at head of file.
15653 (iv_constant): Pass the mode paraeter before the rtx it describes
15654 and change its type to scalar_int_mode. Remove VOIDmode handling.
15655 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
15656 (iv_extend): Likewise.
15657 (shorten_into_mode): Likewise.
15658 (iv_add): Use scalar_int_mode.
15659 (iv_mult): Likewise.
15660 (iv_shift): Likewise.
15661 (canonicalize_iv_subregs): Likewise.
15662 (get_biv_step_1): Pass the outer_mode parameter before the rtx
15663 it describes and change its mode to scalar_int_mode. Also change
15664 the type of the returned inner_mode to scalar_int_mode.
15665 (get_biv_step): Likewise, turning outer_mode from a pointer
15666 into a direct parameter. Update call to get_biv_step_1.
15667 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
15668 iv_constant and get_biv_step.
15669 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
15670 and change its type to scalar_int_mode. Don't initialise iv->mode
15671 to VOIDmode and remove later checks for its still being VOIDmode.
15672 Update calls to iv_analyze_op and iv_analyze_expr. Check
15673 is_a <scalar_int_mode> when changing the mode under consideration.
15674 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
15675 Update call to iv_analyze_expr.
15676 (iv_analyze_op): Add a mode parameter. Reject subregs whose
15677 inner register is not also a scalar_int_mode. Update call to
15679 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
15680 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
15681 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
15682 separate mode class checks. Update calls to iv_analyze. Remove
15683 fix-up of VOIDmodes after iv_analyze_biv.
15684 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
15685 don't have a scalar_int_mode. Update call to biv_p.
15687 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15688 Alan Hayward <alan.hayward@arm.com>
15689 David Sherwood <david.sherwood@arm.com>
15691 * cfgexpand.c (convert_debug_memory_address): Use
15692 as_a <scalar_int_mode>.
15693 * combine.c (expand_compound_operation): Likewise.
15694 (make_extraction): Likewise.
15695 (change_zero_ext): Likewise.
15696 (simplify_comparison): Likewise.
15697 * cse.c (cse_insn): Likewise.
15698 * dwarf2out.c (minmax_loc_descriptor): Likewise.
15699 (mem_loc_descriptor): Likewise.
15700 (loc_descriptor): Likewise.
15701 * expmed.c (init_expmed_one_mode): Likewise.
15702 (synth_mult): Likewise.
15703 (emit_store_flag_1): Likewise.
15704 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
15705 of a comparison with size.
15706 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
15707 (reduce_to_bit_field_precision): Likewise.
15708 * function.c (expand_function_end): Likewise.
15709 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
15710 * loop-doloop.c (doloop_modify): Likewise.
15711 * optabs.c (expand_binop): Likewise.
15712 (expand_unop): Likewise.
15713 (expand_copysign_absneg): Likewise.
15714 (prepare_cmp_insn): Likewise.
15715 (maybe_legitimize_operand): Likewise.
15716 * recog.c (const_scalar_int_operand): Likewise.
15717 * rtlanal.c (get_address_mode): Likewise.
15718 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15719 (simplify_cond_clz_ctz): Likewise.
15720 * tree-nested.c (get_nl_goto_field): Likewise.
15721 * tree.c (build_vector_type_for_mode): Likewise.
15722 * var-tracking.c (use_narrower_mode): Likewise.
15724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15725 Alan Hayward <alan.hayward@arm.com>
15726 David Sherwood <david.sherwood@arm.com>
15728 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
15729 * builtins.c (expand_builtin_signbit): Use it.
15730 * cfgexpand.c (expand_debug_expr): Likewise.
15731 * dojump.c (do_jump): Likewise.
15732 (do_compare_and_jump): Likewise.
15733 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
15734 * expmed.c (make_tree): Likewise.
15735 * expr.c (expand_expr_real_2): Likewise.
15736 (expand_expr_real_1): Likewise.
15737 (try_casesi): Likewise.
15738 * fold-const-call.c (fold_const_call_ss): Likewise.
15739 * fold-const.c (unextend): Likewise.
15740 (extract_muldiv_1): Likewise.
15741 (fold_single_bit_test): Likewise.
15742 (native_encode_int): Likewise.
15743 (native_encode_string): Likewise.
15744 (native_interpret_int): Likewise.
15745 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
15746 * internal-fn.c (expand_addsub_overflow): Likewise.
15747 (expand_neg_overflow): Likewise.
15748 (expand_mul_overflow): Likewise.
15749 (expand_arith_overflow): Likewise.
15750 * match.pd: Likewise.
15751 * stor-layout.c (layout_type): Likewise.
15752 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15753 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
15754 * tree-ssanames.c (get_range_info): Likewise.
15755 * tree-switch-conversion.c (array_value_type) Likewise.
15756 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
15757 (vect_recog_divmod_pattern): Likewise.
15758 (vect_recog_mixed_size_cond_pattern): Likewise.
15759 * tree-vrp.c (extract_range_basic): Likewise.
15760 (simplify_float_conversion_using_ranges): Likewise.
15761 * tree.c (int_fits_type_p): Likewise.
15762 * ubsan.c (instrument_bool_enum_load): Likewise.
15763 * varasm.c (mergeable_string_section): Likewise.
15764 (narrowing_initializer_constant_valid_p): Likewise.
15765 (output_constant): Likewise.
15767 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15768 Alan Hayward <alan.hayward@arm.com>
15769 David Sherwood <david.sherwood@arm.com>
15771 * machmode.h (NARROWEST_INT_MODE): New macro.
15772 * expr.c (alignment_for_piecewise_move): Use it instead of
15773 GET_CLASS_NARROWEST_MODE (MODE_INT).
15774 (push_block): Likewise.
15775 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
15777 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15779 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15780 Alan Hayward <alan.hayward@arm.com>
15781 David Sherwood <david.sherwood@arm.com>
15783 * postreload.c (move2add_valid_value_p): Change the type of the
15784 mode parameter to scalar_int_mode.
15785 (move2add_use_add2_insn): Add a mode parameter and use it instead
15787 (move2add_use_add3_insn): Likewise.
15788 (reload_cse_move2add): Update accordingly.
15790 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15791 Alan Hayward <alan.hayward@arm.com>
15792 David Sherwood <david.sherwood@arm.com>
15794 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
15796 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
15797 * optabs.c (expand_unop): Likewise.
15799 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15800 Alan Hayward <alan.hayward@arm.com>
15801 David Sherwood <david.sherwood@arm.com>
15803 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
15804 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
15805 (popcount_loc_descriptor): Likewise.
15806 (bswap_loc_descriptor): Likewise.
15807 (rotate_loc_descriptor): Likewise.
15808 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
15809 calling the functions above.
15811 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15812 Alan Hayward <alan.hayward@arm.com>
15813 David Sherwood <david.sherwood@arm.com>
15815 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
15817 (try_combine): Likewise.
15818 (simplify_if_then_else): Likewise.
15819 * cse.c (cse_insn): Likewise.
15820 * dwarf2out.c (mem_loc_descriptor): Likewise.
15821 * emit-rtl.c (gen_lowpart_common): Likewise.
15822 * simplify-rtx.c (simplify_truncation): Likewise.
15823 (simplify_binary_operation_1): Likewise.
15824 (simplify_const_relational_operation): Likewise.
15825 (simplify_ternary_operation): Likewise.
15826 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
15828 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15829 Alan Hayward <alan.hayward@arm.com>
15830 David Sherwood <david.sherwood@arm.com>
15832 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
15833 * reload.c (push_reload): Likewise.
15834 (find_reloads): Likewise.
15836 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15837 Alan Hayward <alan.hayward@arm.com>
15838 David Sherwood <david.sherwood@arm.com>
15840 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
15841 (make_compound_operation_int): Likewise.
15842 (change_zero_ext): Likewise.
15843 * expr.c (convert_move): Likewise.
15844 (convert_modes): Likewise.
15845 * fwprop.c (forward_propagate_subreg): Likewise.
15846 * loop-iv.c (get_biv_step_1): Likewise.
15847 * optabs.c (widen_operand): Likewise.
15848 * postreload.c (move2add_valid_value_p): Likewise.
15849 * recog.c (simplify_while_replacing): Likewise.
15850 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15851 (simplify_binary_operation_1): Likewise. Remove redundant
15852 mode equality check.
15854 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15855 Alan Hayward <alan.hayward@arm.com>
15856 David Sherwood <david.sherwood@arm.com>
15858 * combine.c (combine_simplify_rtx): Add checks for
15859 is_a <scalar_int_mode>.
15860 (simplify_if_then_else): Likewise.
15861 (make_field_assignment): Likewise.
15862 (simplify_comparison): Likewise.
15863 * ifcvt.c (noce_try_bitop): Likewise.
15864 * loop-invariant.c (canonicalize_address_mult): Likewise.
15865 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
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 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
15872 is_a <scalar_int_mode> instead of != BLKmode.
15874 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15875 Alan Hayward <alan.hayward@arm.com>
15876 David Sherwood <david.sherwood@arm.com>
15878 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
15879 instead of != VOIDmode.
15880 * combine.c (if_then_else_cond): Likewise.
15881 (change_zero_ext): Likewise.
15882 * dwarf2out.c (mem_loc_descriptor): Likewise.
15883 (loc_descriptor): Likewise.
15884 * rtlanal.c (canonicalize_condition): Likewise.
15885 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
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 * simplify-rtx.c (simplify_binary_operation_1): Use
15892 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
15894 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15895 Alan Hayward <alan.hayward@arm.com>
15896 David Sherwood <david.sherwood@arm.com>
15898 * wide-int.h (int_traits<unsigned char>) New class.
15899 (int_traits<unsigned short>) Likewise.
15900 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
15901 Use GET_MODE_UNIT_PRECISION and remove redundant test for
15903 * combine.c (set_nonzero_bits_and_sign_copies): Use
15904 is_a <scalar_int_mode>.
15905 (find_split_point): Likewise.
15906 (combine_simplify_rtx): Likewise.
15907 (simplify_logical): Likewise.
15908 (expand_compound_operation): Likewise.
15909 (expand_field_assignment): Likewise.
15910 (make_compound_operation): Likewise.
15911 (extended_count): Likewise.
15912 (change_zero_ext): Likewise.
15913 (simplify_comparison): Likewise.
15914 * dwarf2out.c (scompare_loc_descriptor): Likewise.
15915 (ucompare_loc_descriptor): Likewise.
15916 (minmax_loc_descriptor): Likewise.
15917 (mem_loc_descriptor): Likewise.
15918 (loc_descriptor): Likewise.
15919 * expmed.c (init_expmed_one_mode): Likewise.
15920 * lra-constraints.c (lra_constraint_offset): Likewise.
15921 * optabs.c (prepare_libcall_arg): Likewise.
15922 * postreload.c (move2add_note_store): Likewise.
15923 * reload.c (operands_match_p): Likewise.
15924 * rtl.h (load_extend_op): Likewise.
15925 * rtlhooks.c (gen_lowpart_general): Likewise.
15926 * simplify-rtx.c (simplify_truncation): Likewise.
15927 (simplify_unary_operation_1): Likewise.
15928 (simplify_binary_operation_1): Likewise.
15929 (simplify_const_binary_operation): Likewise.
15930 (simplify_const_relational_operation): Likewise.
15931 (simplify_subreg): Likewise.
15932 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15933 * var-tracking.c (adjust_mems): Likewise.
15934 (prepare_call_arguments): Likewise.
15936 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15937 Alan Hayward <alan.hayward@arm.com>
15938 David Sherwood <david.sherwood@arm.com>
15940 * machmode.h (is_int_mode): New fuction.
15941 * combine.c (find_split_point): Use it.
15942 (combine_simplify_rtx): Likewise.
15943 (simplify_if_then_else): Likewise.
15944 (simplify_set): Likewise.
15945 (simplify_shift_const_1): Likewise.
15946 (simplify_comparison): Likewise.
15947 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
15948 * cse.c (notreg_cost): Likewise.
15949 (cse_insn): Likewise.
15950 * cselib.c (cselib_lookup_1): Likewise.
15951 * dojump.c (do_jump_1): Likewise.
15952 (do_compare_rtx_and_jump): Likewise.
15953 * dse.c (get_call_args): Likewise.
15954 * dwarf2out.c (rtl_for_decl_init): Likewise.
15955 (native_encode_initializer): Likewise.
15956 * expmed.c (emit_store_flag_1): Likewise.
15957 (emit_store_flag): Likewise.
15958 * expr.c (convert_modes): Likewise.
15959 (store_field): Likewise.
15960 (expand_expr_real_1): Likewise.
15961 * fold-const.c (fold_read_from_constant_string): Likewise.
15962 * gimple-ssa-sprintf.c (get_format_string): Likewise.
15963 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
15964 * optabs.c (expand_binop): Likewise.
15965 (expand_unop): Likewise.
15966 (expand_abs_nojump): Likewise.
15967 (expand_one_cmpl_abs_nojump): Likewise.
15968 * simplify-rtx.c (mode_signbit_p): Likewise.
15969 (val_signbit_p): Likewise.
15970 (val_signbit_known_set_p): Likewise.
15971 (val_signbit_known_clear_p): Likewise.
15972 (simplify_relational_operation_1): Likewise.
15973 * tree.c (vector_type_mode): Likewise.
15975 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15976 Alan Hayward <alan.hayward@arm.com>
15977 David Sherwood <david.sherwood@arm.com>
15979 * machmode.h (smallest_mode_for_size): Fix formatting.
15980 (smallest_int_mode_for_size): New function.
15981 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
15982 instead of smallest_mode_for_size.
15983 * combine.c (make_extraction): Likewise.
15984 * config/arc/arc.c (arc_expand_movmem): Likewise.
15985 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
15986 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15987 * config/s390/s390.c (s390_expand_insv): Likewise.
15988 * config/sparc/sparc.c (assign_int_registers): Likewise.
15989 * config/spu/spu.c (spu_function_value): Likewise.
15990 (spu_function_arg): Likewise.
15991 * coverage.c (get_gcov_type): Likewise.
15992 (get_gcov_unsigned_t): Likewise.
15993 * dse.c (find_shift_sequence): Likewise.
15994 * expmed.c (store_bit_field_1): Likewise.
15995 * expr.c (convert_move): Likewise.
15996 (store_field): Likewise.
15997 * internal-fn.c (expand_arith_overflow): Likewise.
15998 * optabs-query.c (get_best_extraction_insn): Likewise.
15999 * optabs.c (expand_twoval_binop_libfunc): Likewise.
16000 * stor-layout.c (layout_type): Likewise.
16001 (initialize_sizetypes): Likewise.
16002 * targhooks.c (default_get_mask_mode): Likewise.
16003 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
16005 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16006 Alan Hayward <alan.hayward@arm.com>
16007 David Sherwood <david.sherwood@arm.com>
16009 * machmode.h (opt_mode::else_blk): New function.
16010 (int_mode_for_mode): Declare.
16011 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
16012 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
16014 * cfgexpand.c (expand_debug_expr): Likewise.
16015 * combine.c (gen_lowpart_or_truncate): Likewise.
16016 (gen_lowpart_for_combine): Likewise.
16017 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
16018 * config/avr/avr.c (avr_to_int_mode): Likewise.
16019 (avr_out_plus_1): Likewise.
16020 (avr_out_plus): Likewise.
16021 (avr_out_round): Likewise.
16022 * config/i386/i386.c (ix86_split_to_parts): Likewise.
16023 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
16024 (s390_expand_vcond): Likewise.
16025 * config/spu/spu.c (spu_split_immediate): Likewise.
16026 (spu_expand_mov): Likewise.
16027 * dse.c (get_stored_val): Likewise.
16028 * expmed.c (store_bit_field_1): Likewise.
16029 (convert_extracted_bit_field): Use int_mode_for_mode instead of
16031 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
16032 (extract_low_bits): Likewise.
16033 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
16034 handling rather than repeating the check.
16035 (emit_group_store): Likewise.
16036 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
16037 * optabs.c (expand_absneg_bit): Likewise.
16038 (expand_copysign_absneg): Likewise.
16039 (expand_copysign_bit): Likewise.
16040 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
16041 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
16042 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
16043 * var-tracking.c (prepare_call_arguments): Likewise.
16044 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
16045 int_mode_for_mode instead of mode_for_size.
16046 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
16048 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16049 Alan Hayward <alan.hayward@arm.com>
16050 David Sherwood <david.sherwood@arm.com>
16052 * machmode.h (int_mode_for_size): New function.
16053 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
16054 instead of mode_for_size.
16055 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
16057 * combine.c (expand_field_assignment): Use int_mode_for_size
16058 instead of mode_for_size.
16059 (make_extraction): Likewise.
16060 (simplify_shift_const_1): Likewise.
16061 (simplify_comparison): Likewise.
16062 * dojump.c (do_jump): Likewise.
16063 * dwarf2out.c (mem_loc_descriptor): Likewise.
16064 * emit-rtl.c (init_derived_machine_modes): Likewise.
16065 * expmed.c (flip_storage_order): Likewise.
16066 (convert_extracted_bit_field): Likewise.
16067 * expr.c (copy_blkmode_from_reg): Likewise.
16068 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
16069 * internal-fn.c (expand_mul_overflow): Likewise.
16070 * lower-subreg.c (simple_move): Likewise.
16071 * optabs-libfuncs.c (init_optabs): Likewise.
16072 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16073 * tree.c (vector_type_mode): Likewise.
16074 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
16075 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
16076 * tree-vect-generic.c (expand_vector_parallel): Likewise.
16077 * tree-vect-stmts.c (vectorizable_load): Likewise.
16078 (vectorizable_store): Likewise.
16080 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16081 Alan Hayward <alan.hayward@arm.com>
16082 David Sherwood <david.sherwood@arm.com>
16084 * coretypes.h (pod_mode): New type.
16085 (scalar_int_mode_pod): New typedef.
16086 * machmode.h (pod_mode): New class.
16087 (int_n_data_t::m): Change type to scalar_int_mode_pod.
16088 * genmodes.c (emit_mode_int_n): Update accordingly.
16089 * lower-subreg.h (target_lower_subreg): Change type to
16090 scalar_int_mode_pod.
16091 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
16092 scalar_int_mode_pod.
16094 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16095 Alan Hayward <alan.hayward@arm.com>
16096 David Sherwood <david.sherwood@arm.com>
16098 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
16099 machine_mode to scalar_int_mode.
16100 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
16101 (rs6000_option_override_internal): Remove cast to int.
16102 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
16103 machine_mode to scalar_int_mode.
16104 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
16105 (rs6000_option_override_internal): Remove cast to int.
16106 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
16107 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
16109 * config/s390/s390.c (s390_expand_builtin): Likewise.
16110 * coretypes.h (scalar_int_mode): New type.
16111 (opt_scalar_int_mode): New typedef.
16112 * machmode.h (scalar_int_mode): New class.
16113 (scalar_int_mode::includes_p): New function.
16114 (byte_mode): Change type to scalar_int_mode.
16115 (word_mode): Likewise.
16116 (ptr_mode): Likewise.
16117 * emit-rtl.c (byte_mode): Likewise.
16118 (word_mode): Likewise.
16119 (ptr_mode): Likewise.
16120 (init_derived_machine_modes): Update accordingly.
16121 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
16122 and MODE_PARTIAL_INT.
16123 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
16124 opt_scalar_int_mode.
16126 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16127 Alan Hayward <alan.hayward@arm.com>
16128 David Sherwood <david.sherwood@arm.com>
16130 * target.def (libgcc_floating_mode_supported_p): Take a
16132 * doc/tm.texi: Regenerate.
16133 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
16135 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16136 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
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 (default_floatn_mode): Return an opt_scalar_float_mode.
16144 * doc/tm.texi: Regenerate.
16145 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
16146 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
16147 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
16148 * targhooks.h (default_floatn_mode): Likewise.
16149 * targhooks.c (default_floatn_mode): Likewise.
16150 * tree.c (build_common_tree_nodes): Update accordingly.
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 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
16157 (mode_iterator::iterate_p): Likewise.
16158 (mode_iterator::get_wider): Likewise.
16159 * expr.c (init_expr_target): Use opt_scalar_float_mode.
16161 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16162 Alan Hayward <alan.hayward@arm.com>
16163 David Sherwood <david.sherwood@arm.com>
16165 * coretypes.h (opt_scalar_float_mode): New typedef.
16166 * machmode.h (float_mode_for_size): New function.
16167 * emit-rtl.c (double_mode): Delete.
16168 (init_emit_once): Use float_mode_for_size.
16169 * stor-layout.c (layout_type): Likewise.
16170 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
16172 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16173 Alan Hayward <alan.hayward@arm.com>
16174 David Sherwood <david.sherwood@arm.com>
16176 * output.h (assemble_real): Take a scalar_float_mode.
16177 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
16178 * config/arm/arm.md (consttable_4): Likewise.
16179 (consttable_8): Likewise.
16180 (consttable_16): Likewise.
16181 * config/mips/mips.md (consttable_float): Likewise.
16182 * config/s390/s390.c (s390_output_pool_entry): Likewise.
16183 * varasm.c (assemble_real): Take a scalar_float_mode.
16184 (output_constant_pool_2): Update accordingly.
16185 (output_constant): Likewise.
16187 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16188 Alan Hayward <alan.hayward@arm.com>
16189 David Sherwood <david.sherwood@arm.com>
16191 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
16192 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
16193 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
16194 (native_encode_real): Likewise.
16195 (native_interpret_real): Likewise.
16196 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
16197 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16199 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16200 Alan Hayward <alan.hayward@arm.com>
16201 David Sherwood <david.sherwood@arm.com>
16203 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
16204 <scalar_float_mode>. Simplify.
16205 (gen_extend_conv_libfunc): Likewise.
16207 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16208 Alan Hayward <alan.hayward@arm.com>
16209 David Sherwood <david.sherwood@arm.com>
16211 * coretypes.h (scalar_float_mode): New type.
16212 * machmode.h (mode_traits::from_int): Use machine_mode if
16213 USE_ENUM_MODES is defined.
16214 (is_a): New function.
16216 (dyn_cast): Likewise.
16217 (scalar_float_mode): New class.
16218 (scalar_float_mode::includes_p): New function.
16219 (is_float_mode): Likewise.
16220 * gdbhooks.py (MachineModePrinter): New class.
16221 (build_pretty_printer): Use it for scalar_float_mode.
16222 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
16223 (format_helper::format_helper): Turn into a template.
16224 * genmodes.c (get_mode_class): New function.
16225 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
16226 or machine_mode if none.
16227 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
16228 as_a <scalar_float_mode>.
16229 * dwarf2out.c (mem_loc_descriptor): Likewise.
16230 (insert_float): Likewise.
16231 (add_const_value_attribute): Likewise.
16232 * simplify-rtx.c (simplify_immed_subreg): Likewise.
16233 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
16234 (expand_unop): Update accordingly.
16235 (expand_abs_nojump): Likewise.
16236 (expand_copysign_absneg): Take a scalar_float_mode.
16237 (expand_copysign_bit): Likewise.
16238 (expand_copysign): Update accordingly.
16240 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16241 Alan Hayward <alan.hayward@arm.com>
16242 David Sherwood <david.sherwood@arm.com>
16244 * coretypes.h (opt_mode): New class.
16245 * machmode.h (opt_mode): Likewise.
16246 (opt_mode::else_void): New function.
16247 (opt_mode::require): Likewise.
16248 (opt_mode::exists): Likewise.
16249 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
16250 (GET_MODE_2XWIDER_MODE): Likewise.
16251 (mode_iterator::get_wider): Update accordingly.
16252 (mode_iterator::get_2xwider): Likewise.
16253 (mode_iterator::get_known_wider): Likewise, turning into a template.
16254 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
16255 forcing a wider mode to exist.
16256 * config/cr16/cr16.h (LONG_REG_P): Likewise.
16257 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16258 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
16259 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
16260 * lower-subreg.c (init_lower_subreg): Likewise.
16261 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
16262 on the final iteration.
16263 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
16264 a wider mode exists before asking for a move pattern.
16265 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
16266 forcing a wider mode to exist.
16267 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
16268 returning false if no such mode exists.
16269 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
16270 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
16271 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
16272 Avoid checking for a MODE_INT if we already know the mode is not a
16274 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
16275 forcing a wider mode to exist.
16276 (expmed_mult_highpart_optab): Likewise.
16277 (expmed_mult_highpart): Likewise.
16278 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
16280 * lto-streamer-in.c (lto_input_mode_table): Likewise.
16281 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
16282 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16283 * internal-fn.c (expand_mul_overflow): Update use of
16284 GET_MODE_2XWIDER_MODE.
16285 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16286 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
16287 GET_MODE_WIDER_MODE.
16288 (convert_plusminus_to_widen): Likewise.
16289 * tree-switch-conversion.c (array_value_type): Likewise.
16290 * var-tracking.c (emit_note_insn_var_location): Likewise.
16291 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16292 Return false inside rather than outside the loop if no wider mode
16294 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
16295 and GET_MODE_2XWIDER_MODE
16296 (can_compare_p): Use else_void.
16297 * gdbhooks.py (OptMachineModePrinter): New class.
16298 (build_pretty_printer): Use it for opt_mode.
16300 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16301 Alan Hayward <alan.hayward@arm.com>
16302 David Sherwood <david.sherwood@arm.com>
16304 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
16305 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
16307 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16308 Alan Hayward <alan.hayward@arm.com>
16309 David Sherwood <david.sherwood@arm.com>
16311 * machmode.h (mode_traits): New structure.
16312 (get_narrowest_mode): New function.
16313 (mode_iterator::start): Likewise.
16314 (mode_iterator::iterate_p): Likewise.
16315 (mode_iterator::get_wider): Likewise.
16316 (mode_iterator::get_known_wider): Likewise.
16317 (mode_iterator::get_2xwider): Likewise.
16318 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
16319 (FOR_EACH_MODE): Likewise.
16320 (FOR_EACH_MODE_FROM): Likewise.
16321 (FOR_EACH_MODE_UNTIL): Likewise.
16322 (FOR_EACH_WIDER_MODE): Likewise.
16323 (FOR_EACH_2XWIDER_MODE): Likewise.
16324 * builtins.c (expand_builtin_strlen): Use new mode iterators.
16325 * combine.c (simplify_comparison): Likewise
16326 * config/i386/i386.c (type_natural_mode): Likewise.
16327 * cse.c (cse_insn): Likewise.
16328 * dse.c (find_shift_sequence): Likewise.
16329 * emit-rtl.c (init_derived_machine_modes): Likewise.
16330 (init_emit_once): Likewise.
16331 * explow.c (hard_function_value): Likewise.
16332 * expmed.c (extract_fixed_bit_field_1): Likewise.
16333 (extract_bit_field_1): Likewise.
16334 (expand_divmod): Likewise.
16335 (emit_store_flag_1): Likewise.
16336 * expr.c (init_expr_target): Likewise.
16337 (convert_move): Likewise.
16338 (alignment_for_piecewise_move): Likewise.
16339 (widest_int_mode_for_size): Likewise.
16340 (emit_block_move_via_movmem): Likewise.
16341 (copy_blkmode_to_reg): Likewise.
16342 (set_storage_via_setmem): Likewise.
16343 (compress_float_constant): Likewise.
16344 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16345 * optabs-query.c (get_best_extraction_insn): Likewise.
16346 * optabs.c (expand_binop): Likewise.
16347 (expand_twoval_unop): Likewise.
16348 (expand_twoval_binop): Likewise.
16349 (widen_leading): Likewise.
16350 (widen_bswap): Likewise.
16351 (expand_parity): Likewise.
16352 (expand_unop): Likewise.
16353 (prepare_cmp_insn): Likewise.
16354 (prepare_float_lib_cmp): Likewise.
16355 (expand_float): Likewise.
16356 (expand_fix): Likewise.
16357 (expand_sfix_optab): Likewise.
16358 * postreload.c (move2add_use_add2_insn): Likewise.
16359 * reg-stack.c (reg_to_stack): Likewise.
16360 * reginfo.c (choose_hard_reg_mode): Likewise.
16361 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16362 * stor-layout.c (mode_for_size): Likewise.
16363 (smallest_mode_for_size): Likewise.
16364 (mode_for_vector): Likewise.
16365 (finish_bitfield_representative): Likewise.
16366 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
16367 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
16368 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16369 * var-tracking.c (prepare_call_arguments): Likewise.
16371 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16372 Alan Hayward <alan.hayward@arm.com>
16373 David Sherwood <david.sherwood@arm.com>
16375 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
16376 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
16377 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
16378 * machmode.h (mode_size): Move earlier in file.
16379 (mode_precision): Likewise.
16380 (mode_inner): Likewise.
16381 (mode_nunits): Likewise.
16382 (mode_unit_size): Likewise.
16383 (unit_unit_precision): Likewise.
16384 (mode_wider): Likewise.
16385 (mode_2xwider): Likewise.
16386 (machine_mode): New class.
16387 (mode_to_bytes): New function.
16388 (mode_to_bits): Likewise.
16389 (mode_to_precision): Likewise.
16390 (mode_to_inner): Likewise.
16391 (mode_to_unit_size): Likewise.
16392 (mode_to_unit_precision): Likewise.
16393 (mode_to_nunits): Likewise.
16394 (GET_MODE_SIZE): Use mode_to_bytes.
16395 (GET_MODE_BITSIZE): Use mode_to_bits.
16396 (GET_MODE_PRECISION): Use mode_to_precision.
16397 (GET_MODE_INNER): Use mode_to_inner.
16398 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
16399 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
16400 (GET_MODE_NUNITS): Use mode_to_nunits.
16401 * system.h (ALWAYS_INLINE): New macro.
16402 * config/powerpcspe/powerpcspe-c.c
16403 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
16404 int for arg1_mode and arg2_mode.
16406 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16407 Alan Hayward <alan.hayward@arm.com>
16408 David Sherwood <david.sherwood@arm.com>
16410 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
16411 Prefix mode names with E_ in case statements.
16412 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16413 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
16414 (aarch64_split_simd_move): Likewise.
16415 (aarch64_gen_storewb_pair): Likewise.
16416 (aarch64_gen_loadwb_pair): Likewise.
16417 (aarch64_gen_store_pair): Likewise.
16418 (aarch64_gen_load_pair): Likewise.
16419 (aarch64_get_condition_code_1): Likewise.
16420 (aarch64_constant_pool_reload_icode): Likewise.
16421 (get_rsqrte_type): Likewise.
16422 (get_rsqrts_type): Likewise.
16423 (get_recpe_type): Likewise.
16424 (get_recps_type): Likewise.
16425 (aarch64_gimplify_va_arg_expr): Likewise.
16426 (aarch64_simd_container_mode): Likewise.
16427 (aarch64_emit_load_exclusive): Likewise.
16428 (aarch64_emit_store_exclusive): Likewise.
16429 (aarch64_expand_compare_and_swap): Likewise.
16430 (aarch64_gen_atomic_cas): Likewise.
16431 (aarch64_emit_bic): Likewise.
16432 (aarch64_emit_atomic_swap): Likewise.
16433 (aarch64_emit_atomic_load_op): Likewise.
16434 (aarch64_evpc_trn): Likewise.
16435 (aarch64_evpc_uzp): Likewise.
16436 (aarch64_evpc_zip): Likewise.
16437 (aarch64_evpc_ext): Likewise.
16438 (aarch64_evpc_rev): Likewise.
16439 (aarch64_evpc_dup): Likewise.
16440 (aarch64_gen_ccmp_first): Likewise.
16441 (aarch64_gen_ccmp_next): Likewise.
16442 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
16443 (alpha_emit_xfloating_libcall): Likewise.
16444 (emit_insxl): Likewise.
16445 (alpha_arg_type): Likewise.
16446 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
16447 (arc_preferred_simd_mode): Likewise.
16448 (arc_secondary_reload): Likewise.
16449 (get_arc_condition_code): Likewise.
16450 (arc_print_operand): Likewise.
16451 (arc_legitimate_constant_p): Likewise.
16452 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16453 * config/arc/arc.md (casesi_load): Likewise.
16454 (casesi_compact_jump): Likewise.
16455 * config/arc/predicates.md (proper_comparison_operator): Likewise.
16456 (cc_use_register): Likewise.
16457 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16458 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
16459 (arm_init_iwmmxt_builtins): Likewise.
16460 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
16461 (neon_expand_vector_init): Likewise.
16462 (arm_attr_length_move_neon): Likewise.
16463 (maybe_get_arm_condition_code): Likewise.
16464 (arm_emit_vector_const): Likewise.
16465 (arm_preferred_simd_mode): Likewise.
16466 (arm_output_iwmmxt_tinsr): Likewise.
16467 (thumb1_output_casesi): Likewise.
16468 (thumb2_output_casesi): Likewise.
16469 (arm_emit_load_exclusive): Likewise.
16470 (arm_emit_store_exclusive): Likewise.
16471 (arm_expand_compare_and_swap): Likewise.
16472 (arm_evpc_neon_vuzp): Likewise.
16473 (arm_evpc_neon_vzip): Likewise.
16474 (arm_evpc_neon_vrev): Likewise.
16475 (arm_evpc_neon_vtrn): Likewise.
16476 (arm_evpc_neon_vext): Likewise.
16477 (arm_validize_comparison): Likewise.
16478 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
16479 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
16480 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
16481 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
16482 (c6x_preferred_simd_mode): Likewise.
16483 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
16484 (epiphany_rtx_costs): Likewise.
16485 * config/epiphany/predicates.md (proper_comparison_operator):
16487 * config/frv/frv.c (condexec_memory_operand): Likewise.
16488 (frv_emit_move): Likewise.
16489 (output_move_single): Likewise.
16490 (output_condmove_single): Likewise.
16491 (frv_hard_regno_mode_ok): Likewise.
16492 (frv_matching_accg_mode): Likewise.
16493 * config/h8300/h8300.c (split_adds_subs): Likewise.
16494 (h8300_rtx_costs): Likewise.
16495 (h8300_print_operand): Likewise.
16496 (compute_mov_length): Likewise.
16497 (output_logical_op): Likewise.
16498 (compute_logical_op_length): Likewise.
16499 (compute_logical_op_cc): Likewise.
16500 (h8300_shift_needs_scratch_p): Likewise.
16501 (output_a_shift): Likewise.
16502 (compute_a_shift_length): Likewise.
16503 (compute_a_shift_cc): Likewise.
16504 (expand_a_rotate): Likewise.
16505 (output_a_rotate): Likewise.
16506 * config/i386/i386.c (classify_argument): Likewise.
16507 (function_arg_advance_32): Likewise.
16508 (function_arg_32): Likewise.
16509 (function_arg_64): Likewise.
16510 (function_value_64): Likewise.
16511 (ix86_gimplify_va_arg): Likewise.
16512 (ix86_legitimate_constant_p): Likewise.
16513 (put_condition_code): Likewise.
16514 (split_double_mode): Likewise.
16515 (ix86_avx256_split_vector_move_misalign): Likewise.
16516 (ix86_expand_vector_logical_operator): Likewise.
16517 (ix86_split_idivmod): Likewise.
16518 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
16519 (ix86_build_const_vector): Likewise.
16520 (ix86_build_signbit_mask): Likewise.
16521 (ix86_match_ccmode): Likewise.
16522 (ix86_cc_modes_compatible): Likewise.
16523 (ix86_expand_branch): Likewise.
16524 (ix86_expand_sse_cmp): Likewise.
16525 (ix86_expand_sse_movcc): Likewise.
16526 (ix86_expand_int_sse_cmp): Likewise.
16527 (ix86_expand_vec_perm_vpermi2): Likewise.
16528 (ix86_expand_vec_perm): Likewise.
16529 (ix86_expand_sse_unpack): Likewise.
16530 (ix86_expand_int_addcc): Likewise.
16531 (ix86_split_to_parts): Likewise.
16532 (ix86_vectorize_builtin_gather): Likewise.
16533 (ix86_vectorize_builtin_scatter): Likewise.
16534 (avx_vpermilp_parallel): Likewise.
16535 (inline_memory_move_cost): Likewise.
16536 (ix86_tieable_integer_mode_p): Likewise.
16537 (x86_maybe_negate_const_int): Likewise.
16538 (ix86_expand_vector_init_duplicate): Likewise.
16539 (ix86_expand_vector_init_one_nonzero): Likewise.
16540 (ix86_expand_vector_init_one_var): Likewise.
16541 (ix86_expand_vector_init_concat): Likewise.
16542 (ix86_expand_vector_init_interleave): Likewise.
16543 (ix86_expand_vector_init_general): Likewise.
16544 (ix86_expand_vector_set): Likewise.
16545 (ix86_expand_vector_extract): Likewise.
16546 (emit_reduc_half): Likewise.
16547 (ix86_emit_i387_round): Likewise.
16548 (ix86_mangle_type): Likewise.
16549 (ix86_expand_round_sse4): Likewise.
16550 (expand_vec_perm_blend): Likewise.
16551 (canonicalize_vector_int_perm): Likewise.
16552 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
16553 (expand_vec_perm_1): Likewise.
16554 (expand_vec_perm_interleave3): Likewise.
16555 (expand_vec_perm_even_odd_pack): Likewise.
16556 (expand_vec_perm_even_odd_1): Likewise.
16557 (expand_vec_perm_broadcast_1): Likewise.
16558 (ix86_vectorize_vec_perm_const_ok): Likewise.
16559 (ix86_expand_vecop_qihi): Likewise.
16560 (ix86_expand_mul_widen_hilo): Likewise.
16561 (ix86_expand_sse2_abs): Likewise.
16562 (ix86_expand_pextr): Likewise.
16563 (ix86_expand_pinsr): Likewise.
16564 (ix86_preferred_simd_mode): Likewise.
16565 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
16566 * config/i386/sse.md (*andnot<mode>3): Likewise.
16567 (<mask_codefor><code><mode>3<mask_name>): Likewise.
16568 (*<code><mode>3): Likewise.
16569 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
16570 (ia64_expand_atomic_op): Likewise.
16571 (ia64_arg_type): Likewise.
16572 (ia64_mode_to_int): Likewise.
16573 (ia64_scalar_mode_supported_p): Likewise.
16574 (ia64_vector_mode_supported_p): Likewise.
16575 (expand_vec_perm_broadcast): Likewise.
16576 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
16577 (iq2000_function_arg_advance): Likewise.
16578 (iq2000_function_arg): Likewise.
16579 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
16580 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
16581 (m68k_libcall_value): Likewise.
16582 (m68k_function_value): Likewise.
16583 (sched_attr_op_type): Likewise.
16584 * config/mcore/mcore.c (mcore_output_move): Likewise.
16585 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
16587 (microblaze_function_arg): Likewise.
16588 * config/mips/mips.c (mips16_build_call_stub): Likewise.
16589 (mips_print_operand): Likewise.
16590 (mips_mode_ok_for_mov_fmt_p): Likewise.
16591 (mips_vector_mode_supported_p): Likewise.
16592 (mips_preferred_simd_mode): Likewise.
16593 (mips_expand_vpc_loongson_even_odd): Likewise.
16594 (mips_expand_vec_unpack): Likewise.
16595 (mips_expand_vi_broadcast): Likewise.
16596 (mips_expand_vector_init): Likewise.
16597 (mips_expand_vec_reduc): Likewise.
16598 (mips_expand_msa_cmp): Likewise.
16599 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
16600 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
16601 (cc_flags_for_mode): Likewise.
16602 * config/msp430/msp430.c (msp430_print_operand): Likewise.
16603 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
16604 (nds32_output_casesi_pc_relative): Likewise.
16605 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16606 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
16607 (nvptx_gen_unpack): Likewise.
16608 (nvptx_gen_pack): Likewise.
16609 (nvptx_gen_shuffle): Likewise.
16610 (nvptx_gen_wcast): Likewise.
16611 (nvptx_preferred_simd_mode): Likewise.
16612 * config/pa/pa.c (pa_secondary_reload): Likewise.
16613 * config/pa/predicates.md (base14_operand): Likewise.
16614 * config/powerpcspe/powerpcspe-c.c
16615 (altivec_resolve_overloaded_builtin): Likewise.
16616 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
16618 (rs6000_preferred_simd_mode): Likewise.
16619 (output_vec_const_move): Likewise.
16620 (rs6000_expand_vector_extract): Likewise.
16621 (rs6000_split_vec_extract_var): Likewise.
16622 (reg_offset_addressing_ok_p): Likewise.
16623 (rs6000_legitimate_offset_address_p): Likewise.
16624 (rs6000_legitimize_address): Likewise.
16625 (rs6000_emit_set_const): Likewise.
16626 (rs6000_const_vec): Likewise.
16627 (rs6000_emit_move): Likewise.
16628 (spe_build_register_parallel): Likewise.
16629 (rs6000_darwin64_record_arg_recurse): Likewise.
16630 (swap_selector_for_mode): Likewise.
16631 (spe_init_builtins): Likewise.
16632 (paired_init_builtins): Likewise.
16633 (altivec_init_builtins): Likewise.
16634 (do_load_for_compare): Likewise.
16635 (rs6000_generate_compare): Likewise.
16636 (rs6000_expand_float128_convert): Likewise.
16637 (emit_load_locked): Likewise.
16638 (emit_store_conditional): Likewise.
16639 (rs6000_output_function_epilogue): Likewise.
16640 (rs6000_handle_altivec_attribute): Likewise.
16641 (rs6000_function_value): Likewise.
16642 (emit_fusion_gpr_load): Likewise.
16643 (emit_fusion_p9_load): Likewise.
16644 (emit_fusion_p9_store): Likewise.
16645 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
16646 (fusion_gpr_mem_load): Likewise.
16647 (fusion_addis_mem_combo_load): Likewise.
16648 (fusion_addis_mem_combo_store): Likewise.
16649 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
16650 (fusion_gpr_mem_load): Likewise.
16651 (fusion_addis_mem_combo_load): Likewise.
16652 (fusion_addis_mem_combo_store): Likewise.
16653 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16655 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
16656 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16657 (rs6000_preferred_simd_mode): Likewise.
16658 (output_vec_const_move): Likewise.
16659 (rs6000_expand_vector_extract): Likewise.
16660 (rs6000_split_vec_extract_var): Likewise.
16661 (reg_offset_addressing_ok_p): Likewise.
16662 (rs6000_legitimate_offset_address_p): Likewise.
16663 (rs6000_legitimize_address): Likewise.
16664 (rs6000_emit_set_const): Likewise.
16665 (rs6000_const_vec): Likewise.
16666 (rs6000_emit_move): Likewise.
16667 (rs6000_darwin64_record_arg_recurse): Likewise.
16668 (swap_selector_for_mode): Likewise.
16669 (paired_init_builtins): Likewise.
16670 (altivec_init_builtins): Likewise.
16671 (rs6000_expand_float128_convert): Likewise.
16672 (emit_load_locked): Likewise.
16673 (emit_store_conditional): Likewise.
16674 (rs6000_output_function_epilogue): Likewise.
16675 (rs6000_handle_altivec_attribute): Likewise.
16676 (rs6000_function_value): Likewise.
16677 (emit_fusion_gpr_load): Likewise.
16678 (emit_fusion_p9_load): Likewise.
16679 (emit_fusion_p9_store): Likewise.
16680 * config/rx/rx.c (rx_gen_move_template): Likewise.
16681 (flags_from_mode): Likewise.
16682 * config/s390/predicates.md (s390_alc_comparison): Likewise.
16683 (s390_slb_comparison): Likewise.
16684 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
16685 (s390_vector_mode_supported_p): Likewise.
16686 (s390_cc_modes_compatible): Likewise.
16687 (s390_match_ccmode_set): Likewise.
16688 (s390_canonicalize_comparison): Likewise.
16689 (s390_emit_compare_and_swap): Likewise.
16690 (s390_branch_condition_mask): Likewise.
16691 (s390_rtx_costs): Likewise.
16692 (s390_secondary_reload): Likewise.
16693 (__SECONDARY_RELOAD_CASE): Likewise.
16694 (s390_expand_cs): Likewise.
16695 (s390_preferred_simd_mode): Likewise.
16696 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
16697 * config/sh/sh.c (sh_print_operand): Likewise.
16698 (dump_table): Likewise.
16699 (sh_secondary_reload): Likewise.
16700 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16701 * config/sh/sh.md (casesi_worker_1): Likewise.
16702 (casesi_worker_2): Likewise.
16703 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
16704 (fcc_comparison_operator): Likewise.
16705 * config/sparc/sparc.c (sparc_expand_move): Likewise.
16706 (emit_soft_tfmode_cvt): Likewise.
16707 (sparc_preferred_simd_mode): Likewise.
16708 (output_cbranch): Likewise.
16709 (sparc_print_operand): Likewise.
16710 (sparc_expand_vec_perm_bmask): Likewise.
16711 (vector_init_bshuffle): Likewise.
16712 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
16713 (spu_vector_mode_supported_p): Likewise.
16714 (spu_expand_insv): Likewise.
16715 (spu_emit_branch_or_set): Likewise.
16716 (spu_handle_vector_attribute): Likewise.
16717 (spu_builtin_splats): Likewise.
16718 (spu_builtin_extract): Likewise.
16719 (spu_builtin_promote): Likewise.
16720 (spu_expand_sign_extend): Likewise.
16721 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
16722 (tilegx_simd_int): Likewise.
16723 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
16724 (tilepro_simd_int): Likewise.
16725 * config/v850/v850.c (const_double_split): Likewise.
16726 (v850_print_operand): Likewise.
16727 (ep_memory_offset): Likewise.
16728 * config/vax/vax.c (vax_rtx_costs): Likewise.
16729 (vax_output_int_move): Likewise.
16730 (vax_output_int_add): Likewise.
16731 (vax_output_int_subtract): Likewise.
16732 * config/visium/predicates.md (visium_branch_operator): Likewise.
16733 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
16734 (visium_print_operand_address): Likewise.
16735 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16736 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
16737 (xtensa_expand_conditional_branch): Likewise.
16738 (xtensa_copy_incoming_a7): Likewise.
16739 (xtensa_output_literal): Likewise.
16740 * dfp.c (decimal_real_maxval): Likewise.
16741 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16743 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16744 Alan Hayward <alan.hayward@arm.com>
16745 David Sherwood <david.sherwood@arm.com>
16747 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
16748 (mode_nunits_inline): Likewise.
16749 (mode_inner_inline): Likewise.
16750 (mode_unit_size_inline): Likewise.
16751 (mode_unit_precision_inline): Likewise.
16752 (emit_insn_modes_h): Likewise. Also emit a #define of the
16754 (emit_mode_wider): Add an E_ prefix to mode names.
16755 (emit_mode_complex): Likewise.
16756 (emit_mode_inner): Likewise.
16757 (emit_mode_adjustments): Likewise.
16758 (emit_mode_int_n): Likewise.
16759 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
16760 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
16761 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
16762 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
16763 (CRC32_BUILTIN, ENTRY): Likewise.
16764 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
16765 (aarch64_pop_regs): Likewise.
16766 (aarch64_process_components): Likewise.
16767 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
16768 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
16769 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
16770 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
16771 * config/arm/arm.c (arm_init_libfuncs): Likewise.
16772 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
16774 * config/i386/i386-builtin.def (pcmpestr): Likewise.
16775 (pcmpistr): Likewise.
16776 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
16777 * config/mmix/mmix.c (mmix_output_condition): Likewise.
16778 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
16780 * config/rl78/rl78.c (mduc_regs): Likewise.
16781 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16782 (htm_expand_builtin): Likewise.
16783 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
16784 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
16785 * config/xtensa/xtensa.c (print_operand): Likewise.
16786 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
16787 (NUM_MODE_VECTOR_INT): Likewise.
16788 * genoutput.c (null_operand): Likewise.
16789 (output_operand_data): Likewise.
16790 * genrecog.c (print_parameter_value): Likewise.
16791 * lra.c (debug_operand_data): Likewise.
16793 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16795 * dwarf2out.c (scompare_loc_descriptor_wide)
16796 (scompare_loc_descriptor_narrow): New functions, split out from...
16797 (scompare_loc_descriptor): ...here.
16798 * expmed.c (emit_store_flag_int): New function, split out from...
16799 (emit_store_flag): ...here.
16801 2017-08-30 Richard Biener <rguenther@suse.de>
16803 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
16804 (dwarf2out_early_finish): Move setting of AT_pubnames from
16805 early debug output to early finish.
16807 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16809 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
16810 and -mdata-region to the assembler.
16812 2017-08-30 Richard Biener <rguenther@suse.de>
16814 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
16816 (gen_subprogram_die): Add DW_AT_object_pointer only early.
16817 (dwarf2out_early_global_decl): Only generate a DIE for the
16818 abstract origin if it doesn't already exist or is a declaration DIE.
16819 (resolve_addr): Do not add the linkage name twice when
16820 generating a stub DIE for the DW_TAG_GNU_call_site target.
16822 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16824 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16825 Use machine_mode rather than int for arg1_mode.
16827 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
16830 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
16831 that the second argument of the built-in functions to unpack
16832 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
16833 switch statement instead a lot of if statements.
16834 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
16835 Allow 64-bit values to be in Altivec registers as well as
16836 traditional floating point registers.
16837 (pack<mode>, FMOVE128_VSX iterator): Likewise.
16839 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
16841 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
16842 MAX_REGS_PER_ADDRESS == 1.
16844 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
16846 * config/i386/i386.opt (flag_fentry): Do not init to -1.
16847 * config/i386/i386.c (ix86_option_override_internal): Simplify
16848 setting of opts->x_flag_entry.
16850 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16851 Jakub Jelinek <jakub@redhat.com>
16852 Richard Biener <rguenther@suse.de>
16854 PR tree-optimization/81503
16855 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
16856 folded constant fits in the target type; reorder tests for clarity.
16858 2017-08-29 Martin Liska <mliska@suse.cz>
16860 * passes.def: Include pass_lower_switch.
16861 * stmt.c (dump_case_nodes): Remove and move to
16862 tree-switch-conversion.
16863 (case_values_threshold): Likewise.
16864 (expand_switch_as_decision_tree_p): Likewise.
16865 (emit_case_decision_tree): Likewise.
16866 (expand_case): Likewise.
16867 (balance_case_nodes): Likewise.
16868 (node_has_low_bound): Likewise.
16869 (node_has_high_bound): Likewise.
16870 (node_is_bounded): Likewise.
16871 (emit_case_nodes): Likewise.
16872 (struct simple_case_node): New struct.
16873 (add_case_node): Remove.
16874 (emit_case_dispatch_table): Use vector instead of case_list.
16875 (reset_out_edges_aux): Remove.
16876 (compute_cases_per_edge): Likewise.
16877 (expand_case): Build list of simple_case_node.
16878 (expand_sjlj_dispatch_table): Use it.
16879 * tree-switch-conversion.c (struct case_node): Moved from
16880 stmt.c and adjusted.
16881 (emit_case_nodes): Likewise.
16882 (node_has_low_bound): Likewise.
16883 (node_has_high_bound): Likewise.
16884 (node_is_bounded): Likewise.
16885 (case_values_threshold): Likewise.
16886 (reset_out_edges_aux): Likewise.
16887 (compute_cases_per_edge): Likewise.
16888 (add_case_node): Likewise.
16889 (dump_case_nodes): Likewise.
16890 (balance_case_nodes): Likewise.
16891 (expand_switch_as_decision_tree_p): Likewise.
16892 (emit_jump): Likewise.
16893 (emit_case_decision_tree): Likewise.
16894 (try_switch_expansion): Likewise.
16895 (do_jump_if_equal): Likewise.
16896 (emit_cmp_and_jump_insns): Likewise.
16897 (fix_phi_operands_for_edge): New function.
16898 (record_phi_operand_mapping): Likewise.
16899 (class pass_lower_switch): New pass.
16900 (pass_lower_switch::execute): New function.
16901 (make_pass_lower_switch): Likewise.
16902 (conditional_probability):
16903 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
16904 * tree-pass.h: Add make_pass_lower_switch.
16906 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16909 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
16912 2017-08-29 Richard Biener <rguenther@suse.de>
16914 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
16915 we do not add a DW_AT_inline attribute twice.
16916 (gen_subprogram_die): Remove code setting DW_AT_inline on
16917 DECL_ABSTRACT_P nodes.
16919 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
16921 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
16922 calls to internal functions.
16923 (gimplify_modify_expr): Likewise.
16924 * tree-call-cdce.c (use_internal_fn): Likewise.
16925 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
16926 (convert_to_divmod): Set the nothrow flag.
16927 * tree-if-conv.c (predicate_mem_writes): Likewise.
16928 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16929 (vectorizable_call): Likewise.
16930 (vectorizable_store): Likewise.
16931 (vectorizable_load): Likewise.
16932 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
16933 (vect_recog_mask_conversion_pattern): Likewise.
16935 2017-08-29 Martin Liska <mliska@suse.cz>
16938 * gcc.c (driver_handle_option): Add new argument.
16939 * opts-common.c (handle_option): Pass
16940 target_option_override_hook.
16941 * opts-global.c (lang_handle_option): Add new option.
16942 (set_default_handlers): Add new argument.
16943 (decode_options): Likewise.
16944 * opts.c (target_handle_option): Likewise.
16945 (common_handle_option): Call target_option_override_hook.
16946 * opts.h (struct cl_option_handler_func): Add hook for
16947 target option override.
16948 (struct cl_option_handlers): Likewise.
16949 (set_default_handlers): Add new argument.
16950 (decode_options): Likewise.
16951 (common_handle_option): Likewise.
16952 (target_handle_option): Likewise.
16953 * toplev.c (toplev::main): Pass targetm.target_option.override
16956 2017-08-29 Richard Biener <rguenther@suse.de>
16957 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
16959 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
16960 life to the active subtree.
16962 2017-08-28 Jeff Law <law@redhat.com>
16964 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
16965 derive_equivalences.
16966 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
16967 Code moved into....
16968 (edge_info::derive_equivalences): New private member function
16970 * tree-ssa-dom.c (class edge_info): Changed from a struct
16971 to a class. Add ctor/dtor, methods and data members.
16972 (edge_info::edge_info): Renamed from allocate_edge_info.
16973 Initialize additional members.
16974 (edge_info::~edge_info): New.
16975 (free_dom_edge_info): Delete the edge info.
16976 (record_edge_info): Use new class & associated member functions.
16977 Tighten forms for testing for edge equivalences.
16978 (record_temporary_equivalences): Iterate over the simple
16979 equivalences rather than assuming there's only one per edge.
16980 (cprop_into_successor_phis): Iterate over the simple
16981 equivalences rather than assuming there's only one per edge.
16982 (optimize_stmt): Use operand_equal_p rather than pointer
16983 equality for mini-DSE code.
16985 2017-08-28 Nathan Sidwell <nathan@acm.org>
16987 * gcc.c (execute): Fold SIGPIPE handling into switch
16988 statement. Adjust internal error message.
16990 2017-08-28 Richard Biener <rguenther@suse.de>
16993 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
16994 Do nothing for removed DIEs.
16996 2017-08-28 Richard Biener <rguenther@suse.de>
16998 PR tree-optimization/81977
16999 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
17002 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
17005 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
17006 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
17007 using targetm.gen_mem_thread_fence.
17009 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
17012 * config/i386/i386.md (*<btsc><mode>): Change operand 2
17013 predicate to register_operand. Reorder operands.
17014 (*btr<mode>): Ditto.
17015 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
17016 (*btr<mode>_mask): Ditto.
17018 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
17020 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
17021 * config/rs6000/xmmintrin.h: New file.
17022 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
17024 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17027 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
17028 parameter and_insn and return it.
17029 (recombine_lvx_pattern): Insert a copy to ensure availability of
17030 the base register of the copied masking operation at the point of
17031 the instruction replacement.
17032 (recombine_stvx_pattern): Likewise.
17034 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
17036 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
17037 undocumented switches.
17038 (-mpower9-dform-vector): Likewise.
17039 (-mpower9-dform): Likewise.
17040 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
17041 comments to delete references to -mpower9-dform* switches.
17042 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
17043 Delete reference to -mpower9-dform* switches, test for
17044 -mpower9-vector instead.
17045 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
17046 (OTHER_P9_VECTOR_MASKS): Likewise.
17047 (POWERPC_MASKS): Likewise.
17048 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
17049 tests against -mpower9-dform* to -mpower9-vector. Delete code
17050 that checked for -mpower9-dform* consistancy with other options.
17051 Add test for -mpower9-misc to enable other power9 switches.
17052 (rs6000_init_hard_regno_mode_ok): Likewise.
17053 (rs6000_option_override_internal): Likewise.
17054 (rs6000_emit_prologue): Likewise.
17055 (rs6000_emit_epilogue): Likewise.
17056 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
17057 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
17058 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
17060 (emit_fusion_p9_store): Likewise.
17061 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
17062 resetting these macros if the assembler does not support ISA 3.0
17064 (TARGET_P9_DFORM_VECTOR): Likewise.
17065 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
17066 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
17068 2017-08-25 Alan Modra <amodra@gmail.com>
17070 PR rtl-optimization/81747
17071 * cse.c (cse_extended_basic_block): Don't attempt to record
17072 equivalences for degenerate conditional branches that branch
17073 to their fall-through.
17075 2017-08-24 Martin Sebor <msebor@redhat.com>
17077 PR middle-end/81908
17078 * gimple-fold.c (size_must_be_zero_p): New function.
17079 (gimple_fold_builtin_memory_op): Call it.
17081 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
17083 * config/rs6000/mm_malloc.h: New file.
17085 2017-08-24 Bin Cheng <bin.cheng@arm.com>
17087 PR tree-optimization/81913
17088 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
17089 analysis when either IVs in condition can wrap.
17091 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
17093 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
17094 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
17096 2017-08-24 Richard Biener <rguenther@suse.de>
17099 * targhooks.c (default_target_can_inline_p): Properly
17100 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
17101 is present and always compare.
17102 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
17103 infer -mfpmath=sse from TARGET_SSE_P.
17104 (ix86_can_inline_p): Properly use target_option_default_node when
17105 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
17107 2017-08-24 Richard Biener <rguenther@suse.de>
17110 * dwarf2out.c (output_die): Handle flag_generate_offload like
17112 (output_comp_unit): Likewise.
17113 (gen_array_type_die): Likewise.
17114 (dwarf2out_early_finish): Likewise.
17115 (note_variable_value_in_expr): Likewise.
17116 (dwarf2out_finish): Likewise. Adjust assert.
17117 * cgraphunit.c (symbol_table::compile): Move setting of
17118 flag_generate_offload earlier ...
17119 (symbol_table::finalize_compilation_unit): ... here, before
17120 early debug finalization.
17122 2017-08-24 Richard Biener <rguenther@suse.de>
17124 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
17125 and ipa-fnsummary.h.
17126 (ix86_can_inline_p): When ix86_fpmath flags do not match
17127 check whether the callee uses FP math at all.
17129 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
17131 PR middle-end/81931
17132 * tree-ssanames.c (get_nonzero_bits): Use element_precision
17133 instead of TYPE_PRECISION.
17135 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
17136 Alan Hayward <alan.hayward@arm.com>
17137 David Sherwood <david.sherwood@arm.com>
17139 * combine.c (make_extraction): Use subreg_offset_from_lsb.
17141 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
17143 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
17145 (ix86_frame::stack_realign_allocate): New field.
17146 (struct machine_frame_state): Modify comments.
17147 (machine_frame_state::sp_realigned_fp_end): New field.
17148 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
17149 layout calculation.
17150 (sp_valid_at): Add assertion to assure no attempt to access invalid
17151 offset of a realigned stack.
17152 (fp_valid_at): Likewise.
17153 (choose_baseaddr): Modify comments.
17154 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
17155 ix86_expand_prologue.
17156 (ix86_expand_prologue): Modify stack realignment and allocation.
17157 (ix86_expand_epilogue): Modify comments.
17158 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
17159 avx2_runtime, avx512f, and avx512f_runtime.
17161 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
17163 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
17164 (mstackrealign): Do not init to -1.
17165 * config/i386/i386.c (ix86_option_override_internal):
17166 Check opts_set, not opts when setting default value of
17167 opts->x_ix86_force_align_arg_pointer.
17169 2017-08-23 Richard Biener <rguenther@suse.de>
17171 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
17172 lang_hooks.decl_printable_name.
17173 * print-rtl-function.c (print_rtx_function): Likewise.
17174 * tree-pretty-print.c (dump_function_header): Likewise.
17176 2017-08-23 Richard Biener <rguenther@suse.de>
17179 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
17180 -g0 at compile-time.
17182 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17184 PR middle-end/19706
17185 * doc/sourcebuild.texi (Other hardware attributes):
17188 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17190 PR middle-end/19706
17191 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
17192 Fix single-use check.
17194 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
17196 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
17198 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17200 * doc/install.texi: Modify to add more details on running selected
17203 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17205 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
17206 is combined with -mabi=ms.
17207 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
17210 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17212 PR tree-optimization/81488
17213 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
17214 and cached_basis fields.
17215 (MAX_SPREAD): New constant.
17216 (alloc_cand_and_find_basis): Initialize new fields.
17217 (clear_visited): New function.
17218 (create_phi_basis_1): Rename from create_phi_basis, set visited
17219 and cached_basis fields.
17220 (create_phi_basis): New wrapper function.
17221 (phi_add_costs_1): Rename from phi_add_costs, add spread
17222 parameter, set visited field, short-circuit when limits reached.
17223 (phi_add_costs): New wrapper function.
17224 (record_phi_increments_1): Rename from record_phi_increments, set
17226 (record_phi_increments): New wrapper function.
17227 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
17228 (phi_incr_cost): New wrapper function.
17229 (all_phi_incrs_profitable_1): Rename from
17230 all_phi_incrs_profitable, set visited field.
17231 (all_phi_incrs_profitable): New wrapper function.
17233 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17234 Alan Hayward <alan.hayward@arm.com>
17235 David Sherwood <david.sherwood@arm.com>
17237 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
17238 that takes the outer and inner modes.
17239 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
17240 comparison as the canonical test for a paradoxical subreg.
17241 * combine.c (simplify_set): Use paradoxical_subreg_p.
17242 (make_extraction): Likewise.
17243 (force_to_mode): Likewise.
17244 (rtx_equal_for_field_assignment_p): Likewise.
17245 (gen_lowpart_for_combine): Likewise.
17246 (simplify_comparison): Likewise.
17247 * cse.c (equiv_constant): Likewise.
17248 * expmed.c (store_bit_field_1): Likewise.
17249 * final.c (alter_subreg): Likewise.
17250 * fwprop.c (propagate_rtx): Likewise.
17251 (forward_propagate_subreg): Likewise.
17252 * ira-conflicts.c (ira_build_conflicts): Likewise.
17253 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
17254 * lra-constraints.c (curr_insn_transform): Likewise.
17255 (split_reg): Likewise.
17256 * lra-eliminations.c (move_plus_up): Likewise.
17257 (lra_eliminate_regs_1): Likewise.
17258 * recog.c (general_operand): Likewise.
17259 * ree.c (combine_reaching_defs): Likewise.
17260 * reload.c (push_reload): Likewise.
17261 (find_reloads): Likewise.
17262 * reload1.c (elimination_effects): Likewise.
17263 (compute_reload_subreg_offset): Likewise.
17264 (choose_reload_regs): Likewise.
17265 * rtlanal.c (subreg_lsb_1): Likewise.
17266 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17267 (simplify_subreg): Likewise.
17268 * var-tracking.c (track_loc_p): Likewise.
17269 * emit-rtl.c (byte_lowpart_offset): Likewise.
17270 (paradoxical_subreg_p): Delete out-of-line definition.
17272 2017-08-22 Jeff Law <law@redhat.com>
17274 PR tree-optimization/81741
17275 PR tree-optimization/71947
17276 * tree-ssa-dom.c: Include tree-inline.h.
17277 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
17278 equivalences if one is more expensive to compute than the other.
17279 * tree-ssa-scopedtables.h (class const_or_copies): Make
17280 record_const_or_copy_raw method private.
17281 (class avail_exprs_stack): New method simplify_binary_operation.
17282 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
17283 avail_exprs_stack::simplify_binary_operation as needed.
17284 (avail_exprs_stack::simplify_binary_operation): New function.
17286 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
17288 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
17289 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
17290 (DOT_SYMBOLS): Likewise.
17291 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
17292 (RELOCATABLE_NEEDS_FIXUP): Likewise.
17293 (RS6000_ABI_NAME): Likewise.
17294 (TARGET_CMODEL): Likewise.
17295 (TOC_SECTION_ASM_OP): Likewise.
17296 (SET_CMODEL): New macro.
17297 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
17299 2017-08-22 Richard Biener <rguenther@suse.de>
17301 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
17302 to allow for free-lang-data replacements similar to verify_type_variant.
17304 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17305 Alan Hayward <alan.hayward@arm.com>
17306 David Sherwood <david.sherwood@arm.com>
17308 * config/aarch64/aarch64.md (casesi): Use DImode rather than
17309 VOIDmode for the LABEL_REF.
17311 2017-08-22 Richard Biener <rguenther@suse.de>
17313 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
17315 2017-08-22 Richard Biener <rguenther@suse.de>
17317 * common.opt (feliminate-dwarf2-dups): Ignore.
17318 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
17319 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
17320 same_die_p_wrap, compute_section_prefix,
17321 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
17322 (comdat_symbol_id, comdat_symbol_number): Likewise.
17323 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
17325 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
17326 (output_die): Mark unreachable path unreachable.
17327 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
17328 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
17329 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
17330 (dwarf2out_early_finish): Likewise.
17332 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
17334 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
17336 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
17339 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
17340 not VAR_P. Filter attribute warnings with OPT_Wattributes.
17341 (avr_attribute_table) <io, io_low, address>: Initialize
17342 .decl_required with true.
17344 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
17346 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
17347 undocumented debugging options.
17348 (-mvsx-scalar-double): Likewise.
17349 (-mallow-df-permute): Likewise.
17350 (-mvectorize-builtins): Likewise.
17351 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17352 (rs6000_builtin_vectorized_function): Likewise.
17353 (rs6000_builtin_md_vectorized_function): Likewise.
17354 (rs6000_opt_vars): Likewise.
17356 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
17359 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
17360 (*btrq_imm): Rename from *btrq.
17361 (*btcq_imm): Rename from *btcq.
17362 (btsc): New code attribute.
17363 (*<btsc><mode>): New insn pattern.
17364 (*btr<mode>): Ditto.
17365 (*<btsc><mode>_mask): New insn_and_split pattern.
17366 (*btr<mode>_mask): Ditto.
17368 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17369 Alan Hayward <alan.hayward@arm.com>
17370 David Sherwood <david.sherwood@arm.com>
17372 * function.c (pad_below): Simplify padding calculation.
17374 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17375 Alan Hayward <alan.hayward@arm.com>
17376 David Sherwood <david.sherwood@arm.com>
17378 * target.def (function_prologue): Remove frame size argument.
17379 (function_epilogue): Likewise.
17380 * doc/tm.texi: Regenerate.
17381 * final.c (final_start_function): Update call to function_prologue.
17382 (final_end_function): Update call to function_epilogue.
17383 (default_function_pro_epilogue): Remove frame size argument.
17384 * output.h (default_function_pro_epilogue): Likewise.
17385 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
17386 (arm_output_function_prologue): Likewise.
17387 * config/frv/frv.c (frv_function_prologue): Likewise.
17388 (frv_function_epilogue): Likewise.
17389 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
17390 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
17391 (ia64_output_function_epilogue): Likewise.
17392 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
17393 (m32r_output_function_epilogue): Likewise.
17394 * config/microblaze/microblaze.c (microblaze_function_prologue)
17395 (microblaze_function_epilogue): Likewise.
17396 * config/mips/mips.c (mips_output_function_prologue): Likewise.
17397 (mips_output_function_epilogue): Likewise.
17398 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
17399 (mmix_target_asm_function_epilogue): Likewise.
17400 * config/msp430/msp430.c (msp430_start_function): Likewise.
17401 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
17402 (nds32_asm_function_epilogue): Likewise.
17403 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
17404 * config/pa/pa.c (pa_output_function_prologue): Likewise.
17405 (pa_output_function_epilogue): Likewise.
17406 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
17407 (rs6000_output_function_epilogue): Likewise.
17408 * config/rl78/rl78.c (rl78_start_function): Likewise.
17409 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
17410 (rs6000_output_function_epilogue): Likewise.
17411 * config/rx/rx.c (rx_output_function_prologue): Likewise.
17412 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
17413 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
17414 (sparc_asm_function_epilogue): Likewise.
17416 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17418 * tree.h (type_has_mode_precision_p): New function.
17419 * convert.c (convert_to_integer_1): Use it.
17420 * expr.c (expand_expr_real_2): Likewise.
17421 (expand_expr_real_1): Likewise.
17422 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
17423 * match.pd: Likewise.
17424 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
17425 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
17426 * tree-tailcall.c (process_assignment): Likewise.
17427 * tree-vect-loop.c (vectorizable_reduction): Likewise.
17428 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
17429 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
17430 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17431 (vectorizable_assignment): Likewise.
17432 (vectorizable_shift): Likewise.
17433 (vectorizable_operation): Likewise.
17434 * tree-vrp.c (register_edge_assert_for_2): Likewise.
17436 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
17438 * match.pd: Add pow (C, x) simplification.
17440 2017-08-21 Richard Biener <rguenther@suse.de>
17442 PR tree-optimization/81900
17443 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
17444 for blocks with abnormal predecessors.
17445 (compute_antic): Do not set visited flag prematurely.
17447 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
17450 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
17452 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17454 * stor-layout.h (vector_type_mode): Move to...
17455 * tree.h (vector_type_mode): ...here.
17456 * stor-layout.c (vector_type_mode): Move to...
17457 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
17459 2017-08-21 Richard Biener <rguenther@suse.de>
17461 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
17462 register_external_die hooks.
17463 (debug_false_tree_charstarstar_uhwistar): Declare.
17464 (debug_nothing_tree_charstar_uhwi): Likewise.
17465 * debug.c (do_nothing_debug_hooks): Adjust.
17466 (debug_false_tree_charstarstar_uhwistar): New do nothing.
17467 (debug_nothing_tree_charstar_uhwi): Likewise.
17468 * dbxout.c (dbx_debug_hooks): Adjust.
17469 (xcoff_debug_hooks): Likewise.
17470 * sdbout.c (sdb_debug_hooks): Likewise.
17471 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
17472 * dwarf2out.c (macinfo_label_base): New global.
17473 (dwarf2out_register_external_die): New function for the
17474 register_external_die hook.
17475 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
17476 (dwarf2_debug_hooks): Use them.
17477 (dwarf2_lineno_debug_hooks): Adjust.
17478 (struct die_struct): Add with_offset flag.
17479 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
17480 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
17481 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
17482 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
17483 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
17484 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
17485 defining section names for the early LTO debug variants.
17486 (reset_indirect_string): New helper.
17487 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
17488 (print_dw_val): Add support for offsetted symbol references.
17489 (get_ultimate_context): Split out from is_cxx.
17490 (is_cxx): Use get_ultimate_context.
17491 (is_fortran): Add decl overload.
17492 (compute_comp_unit_symbol): Split out worker from
17493 compute_section_prefix.
17494 (compute_section_prefix): Call compute_comp_unit_symbol and
17495 set comdat_type_p here.
17496 (output_die): Skip DIE symbol output for the LTO added one.
17497 Handle DIE symbol references with offset.
17498 (output_comp_unit): Guard section name mangling properly.
17499 For LTO debug sections emit a symbol at the section beginning
17500 which we use to refer to its DIEs.
17501 (add_abstract_origin_attribute): For DIEs registered via
17502 dwarf2out_register_external_die directly refer to the early
17503 DIE rather than indirectly through the shadow one we created.
17504 Remove obsolete call to dwarf2out_abstract_function for
17505 non-function/block origins.
17506 (gen_array_type_die): When generating early LTO debug do
17507 not emit DW_AT_string_length.
17508 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
17509 late when in LTO. As suggested place a gcc_unreachable for
17510 the DECL_ABSTRACT_P case.
17511 (gen_subprogram_die): Avoid another specification DIE
17512 for early built declarations/definitions for the late LTO case.
17513 (gen_variable_die): Add type references for late duplicated VLA dies
17515 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
17516 we have the abstract instance already.
17517 (process_scope_var): Adjust decl DIE contexts in LTO which
17518 first puts them in limbo.
17519 (gen_decl_die): Do not generate type DIEs late apart from
17520 types for VLAs or for decls we do not yet have a DIE. Do not
17521 call dwarf2out_abstract_function late.
17522 (dwarf2out_early_global_decl): Make sure to create DIEs
17523 for abstract instances of a decl first.
17524 (dwarf2out_late_global_decl): Adjust comment.
17525 (output_macinfo_op): With multiple macro sections use
17526 macinfo_label_base to distinguish labels.
17527 (output_macinfo): Likewise. Update macinfo_label_base.
17528 Pass in the line info label.
17529 (note_variable_value_in_expr): When generating LTO resolve
17530 all variable values here by generating DIEs as needed.
17531 (init_sections_and_labels): Add early LTO debug flag parameter
17532 and generate different sections and names if set. Add generation
17533 counter for the labels so we can have multiple of them.
17534 (reset_dies): Helper to allow DIEs to be output multiple times.
17535 (dwarf2out_finish): When outputting DIEs to the fat part of an
17536 LTO object first reset DIEs.
17537 (dwarf2out_early_finish): Output early DIEs when generating LTO.
17538 (modified_type_die): Check for decl_ultimate_origin being self
17540 (gen_type_die_with_usage): Likewise.
17541 (gen_typedef_die): Allow decl_ultimate_origin being self.
17542 (set_decl_abstract_flags): Remove.
17543 (set_block_abstract_flags): Likewise.
17544 (dwarf2out_abstract_function): Treat the early generated DIEs
17545 as the abstract copy and only add DW_AT_inline and
17546 DW_AT_artificial here and call set_decl_origin_self.
17547 If the DIE has an abstract origin don't do anything.
17548 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
17549 if we have none yet (Go fails to build one, PR78628).
17550 (variably_modified_type_p): Prevent endless recursion for Ada
17551 cyclic pointer types.
17552 * lto-streamer-in.c: Include debug.h.
17553 (dref_queue): New global.
17554 (lto_read_tree_1): Stream in DIE references.
17555 (lto_input_tree): Register DIE references.
17556 (input_function): Stream DECL_DEBUG_ARGS.
17557 * lto-streamer-out.c: Include debug.h.
17558 (lto_write_tree_1): Output DIE references.
17559 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
17560 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
17561 (output_function): Stream DECL_DEBUG_ARGS.
17562 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
17563 Stream DECL_ABSTRACT_ORIGIN.
17564 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
17565 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
17566 DECL_CONTEXT for file-scope decls.
17567 * lto-streamer.h (struct dref_entry): Declare.
17568 (dref_queue): Likewise.
17569 * cfgexpand.c (pass_expand::execute): Do not call the
17570 outlining_inline_function hook here.
17571 * lto-wrapper.c (debug_obj): New global.
17572 (tool_cleanup): Unlink it if required.
17573 (debug_objcopy): New function.
17574 (run_gcc): Handle early debug sections in the IL files by
17575 extracting them to separate files, partially linkin them and
17576 feeding the result back as result to the linker.
17577 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
17578 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
17579 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
17580 sections into a separate segment.
17581 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
17583 (darwin_asm_dwarf_section): Likewise.
17584 (darwin_asm_output_dwarf_offset): Likewise.
17585 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
17587 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17588 Alan Hayward <alan.hayward@arm.com>
17589 David Sherwood <david.sherwood@arm.com>
17591 * read-md.h (md_reader::record_potential_iterator_use): Replace
17592 pointer argument with an rtx and an index.
17593 * read-rtl.c (iterator_group::apply_iterator): Likewise.
17594 (apply_mode_iterator): Likewise.
17595 (apply_code_iterator): Likewise.
17596 (apply_int_iterator): Likewise.
17597 (apply_subst_iterator): Likewise.
17598 (record_iterator_use): Likewise.
17599 (record_attribute_use): Likewise.
17600 (md_reader::record_potential_iterator_use): Likewise. Update calls
17601 to record_iterator_use and apply_iterator.
17602 (iterator_use): Replace ptr with x and index.
17603 (attribute_use): Likewise.
17604 (apply_attribute_uses): Update calls to apply_iterator.
17605 (apply_iterators): Likewise. Update initialization of iterator_use.
17606 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
17607 and record_potential_iterator_use.
17608 (rtx_reader::read_rtx_operand): Likewise.
17610 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17611 Alan Hayward <alan.hayward@arm.com>
17612 David Sherwood <david.sherwood@arm.com>
17614 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
17617 2017-08-21 Richard Biener <rguenther@suse.de>
17619 PR middle-end/81884
17620 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
17621 at struct end conservatively when comparing common bases.
17623 2017-08-21 Richard Biener <rguenther@suse.de>
17625 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
17626 (mem_ref_in_stmt): Remove.
17627 (determine_max_movement): Use ref index to get at the reference.
17628 (invariantness_dom_walker::before_dom_children): Deal with
17629 lim data already initialized.
17630 (gather_mem_refs_stmt): Initialize lim data and record ref index.
17632 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
17634 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
17635 (TARGET_ISA_ROUND): Ditto.
17636 (TARGET_ROUND): Ditto.
17637 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
17638 * config/i386/i386.md: Ditto.
17639 * config/i386/sse.md: Ditto.
17640 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
17641 with OPTION_MASK_ISA_SSE4_1.
17643 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
17646 * doc/extend.texi (x86 Built-in Functions): Correct the name of
17647 __builtin_ia32_lzcnt_u16.
17649 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
17652 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
17653 (rs6000_set_current_function): Rewrite function to use it.
17655 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
17658 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
17659 and TYPE_WARN_IF_NOT_ALIGN.
17660 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
17661 (handle_warn_if_not_align): New.
17662 (place_union_field): Call handle_warn_if_not_align.
17663 (place_field): Call handle_warn_if_not_align.
17664 Copy TYPE_WARN_IF_NOT_ALIGN.
17665 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
17666 (layout_type): Likewise.
17667 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
17669 (tree_decl_common): Add warn_if_not_align.
17670 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
17671 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
17672 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
17673 (DECL_WARN_IF_NOT_ALIGN): Likewise.
17674 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
17675 * doc/extend.texi: Document warn_if_not_aligned attribute.
17676 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
17678 2017-08-17 Martin Liska <mliska@suse.cz>
17681 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
17682 (get_data_dependence): Use it as pointer type.
17683 (distribute_loop): Likewise.
17685 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17687 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
17688 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
17689 (p8_vmrgow_<mode>_direct): New define_insn.
17690 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
17691 handle endianness for vmrgew and vmrgow permute patterns.
17693 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
17695 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
17696 * config/rs6000/rs6000-cpus.def: Remove comment.
17697 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
17698 (POWERPC_MASKS): Likewise.
17699 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
17700 use of TARGET_VSX_TIMODE.
17701 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
17702 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
17703 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
17704 (rs6000_option_override_internal): Remove dead code.
17705 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
17706 (rs6000_legitimize_reload_address): Likewise.
17707 (rs6000_legitimate_address_p): Likewise.
17708 (rs6000_opt_masks): Delete "vsx-timode".
17709 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
17710 from function comment.
17711 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
17712 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
17713 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
17715 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
17716 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
17717 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
17718 (VSX_TI): Likewise.
17720 (define_peephole2): Likewise.
17722 2017-08-17 Martin Sebor <msebor@redhat.com>
17725 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
17726 past the end of an array.
17727 (test_pp_format): Add test cases.
17729 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
17731 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
17732 missing ECF_NOTHROW flags.
17734 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
17737 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
17738 operands residing in integer registers.
17739 (*vsx_le_perm_load_<mode>): Likewise.
17740 (*vsx_le_perm_store_<mode>): Likewise.
17741 (define_peephole2): Add peepholes to optimize the above.
17743 2017-08-17 Marek Polacek <polacek@redhat.com>
17745 PR middle-end/81814
17746 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
17747 to mimic what shorten_compare did. Change the return type to bool.
17748 (fold_cond_expr_with_comparison): Update call to
17749 operand_equal_for_comparison_p.
17750 (fold_ternary_loc): Likewise.
17752 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
17754 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
17756 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
17758 2017-08-17 Richard Biener <rguenther@suse.de>
17760 * tree-ssa-structalias.c (solve_graph): When propagating
17761 to successors update the graphs succ edges and avoid duplicate work.
17763 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
17766 * config/i386/i386.c (ix86_option_override_internal): Save target
17767 specific options after ix86_stack_protector_guard_reg was changed.
17769 2017-08-17 Richard Biener <rguenther@suse.de>
17771 PR tree-optimization/81827
17772 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
17773 (new_var_info): Initialize it conservatively.
17774 (get_call_vi): Mark register vars.
17775 (new_scalar_tmp_constraint_exp): Likewise.
17776 (handle_rhs_call): Likewise.
17777 (handle_const_call): Likewise.
17778 (create_function_info_for): Likewise.
17779 (solve_constraints): Sort varinfos to separate register from
17780 non-register vars to pack points-to solution bitmaps during
17783 2017-08-17 Marek Polacek <polacek@redhat.com>
17785 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
17787 2017-08-17 Richard Biener <rguenther@suse.de>
17789 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
17790 to true when overflow is undefined and we saturated the result.
17792 2017-08-17 Alan Modra <amodra@gmail.com>
17795 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
17796 Don't use store multiple if only one reg needs saving.
17797 (interesting_frame_related_regno): New function.
17798 (rs6000_frame_related): Don't emit frame info for regs that
17800 (rs6000_emit_epilogue): Likewise.
17802 2017-08-16 Nathan Sidwell <nathan@acm.org>
17804 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
17805 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
17806 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
17807 * tree.c (free_lang_data_in_type): Use else-if chain. Always
17808 clear TYPE_LANG_1. Remove obsolete member-function stripping.
17809 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
17810 (verify_type): Adjust for TYPE_BINFO move.
17811 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
17812 process TYPE_BINFO directly.
17813 (hash_tree): Likewise.
17814 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17816 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17819 2017-08-16 David Malcolm <dmalcolm@redhat.com>
17821 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
17823 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
17826 * config/i386/i386.md (*anddi_1_btr): Change predicates of
17827 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
17828 Add ix86_binary_operator_ok to insn constraint.
17829 (*iordi_1_bts): Ditto.
17830 (*xordi_1_btc): Ditto.
17831 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
17832 Update corresponding peephole2 pattern.
17836 2017-08-16 Bin Cheng <bin.cheng@arm.com>
17838 PR tree-optimization/81832
17839 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
17840 copy loop header which has IFN_LOOP_DIST_ALIAS call.
17842 2017-08-16 Marek Polacek <polacek@redhat.com>
17845 * fold-const.c (fold_indirect_ref_1): Restore original behavior
17846 regarding size_zero_node.
17848 2017-08-16 Martin Liska <mliska@suse.cz>
17851 * config.gcc: Respect previously set extra_objs in case
17854 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
17856 PR tree-optimization/81835
17857 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
17858 the phi SSA_NAME. Check that the condition in a COND_EXPR does
17859 not depend on the phi.
17861 2017-08-16 Alan Modra <amodra@gmail.com>
17863 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
17866 2017-08-16 Alan Modra <amodra@gmail.com>
17868 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
17869 (save_reg_p): ..into this. Update all callers.
17870 (first_reg_to_save): Simplify.
17872 2017-08-16 Alan Modra <amodra@gmail.com>
17874 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
17877 2017-08-15 Joseph Myers <joseph@codesourcery.com>
17881 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
17882 constant count if that count is less than 32.
17884 2017-08-15 Nathan Sidwell <nathan@acm.org>
17886 * gcc.c (execute): Emit friendlier message if inferior is killed
17887 by an external cause.
17889 2017-08-15 Richard Biener <rguenther@suse.de>
17891 PR tree-optimization/81790
17892 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
17893 CONSTRUCTORs from simplifying and VN.
17895 2017-08-14 Martin Sebor <msebor@redhat.com>
17897 * builtin-attrs.def: Add comments.
17899 2017-08-14 Martin Sebor <msebor@redhat.com>
17902 * doc/extend.texi (attribute nonstring): Document new attribute.
17904 2017-08-14 Martin Sebor <msebor@redhat.com>
17907 * tree-diagnostic.c (default_tree_printer): Handle %G.
17908 * gimple-pretty-print.h (percent_G_format): Declare new function.
17909 * gimple-pretty-print.c (percent_G_format): Define.
17910 * tree-pretty-print.c (percent_K_format): Add argument.
17912 2017-08-14 Martin Sebor <msebor@redhat.com>
17914 PR translation/79998
17915 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
17916 Remove a stray space.
17918 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
17921 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
17922 (*iordi_1_bts): Ditto.
17923 (*xordi_1_btc): Ditto.
17925 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17928 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
17929 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17931 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
17932 quoted strings, and make more translator-friendly.
17933 (darwin_rs6000_override_options): Likewise.
17934 (rs6000_option_override_internal): Likewise.
17935 (rs6000_return_in_memory): Fix overlong line.
17936 (init_cmulative_args): Use quoted strings, and make more
17937 translator-friendly.
17938 (rs6000_pass_by_reference): Fix overlong line.
17939 (def_builtin): Use quoted strings.
17940 (altivec_expand_predicate_builtin): Use quoted strings, and make
17941 more translator-friendly.
17942 (htm_expand_builtin): Use quoted strings.
17943 (cpu_expand_builtin): Use quoted strings, and make more
17944 translator-friendly.
17945 (altivec_expand_builtin): Likewise.
17946 (paired_expand_predicate_builtin): Likewise.
17947 (rs6000_invalid_builtin): Likewise.
17948 (builtin_function_type): Use quoted strings.
17949 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
17950 more translator-friendly.
17951 (rs6000_trampoline_init): Likewise.
17952 (rs6000_handle_altivec_attribute): Likewise.
17953 (rs6000_inner_target_options): Use quoted strings.
17954 (rs6000_disable_incompatible_switches): Likewise.
17955 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
17956 strings, and make more translator-friendly.
17957 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
17959 2017-08-14 Bin Cheng <bin.cheng@arm.com>
17961 PR tree-optimization/81799
17962 * tree-loop-distribution.c (version_loop_by_alias_check): Force
17963 cond_expr to simple gimple operand.
17965 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
17967 PR middle-end/46932
17968 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
17970 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
17974 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
17975 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
17976 TARGET_GASISR_PROLOGUES.
17977 * config/avr/avr.c (avr_option_override): Same.
17978 (avr_pass_pre_proep::execute): Same.
17980 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
17983 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
17984 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
17986 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
17988 * config/i386/i386.md (*load_tp_<mode>): Redefine as
17989 define_insn_and_split. Split to a memory load from 0 in
17990 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
17991 using PTR mode iterator.
17992 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
17993 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
17994 (*add_tp_<mode>): Redefine as define_insn_and_split.
17995 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
17996 address space. Merge with *add_tp_x32 using PTR mode iterator.
17997 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
17998 Split to an add with a memory load from 0 in
17999 DEFAULT_TLS_SEG_REG address space.
18001 2017-08-12 Andrew Pinski <apinski@cavium.com>
18003 * config/aarch64/aarch64-option-extensions.def (rdma):
18004 Fix feature string to what Linux prints out in /proc/cpuinfo.
18006 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
18009 * dwarf2out.c (modified_type_die): For C typedef types that have
18010 an ultimate origin, process the ultimate origin instead of the
18012 (gen_typedef_die): Assert that input DECLs have no ultimate
18014 (gen_type_die_with_usage): For typedef variants that have an
18015 ultimate origin, just call gen_decl_die on the original DECL.
18016 (process_scope_var): Avoid creating DIEs for local typedefs and
18017 concrete static variables.
18019 2017-08-12 Alan Modra <amodra@gmail.com>
18023 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
18024 match gnu-user.h startfile.
18025 (ENDFILE_LINUX_SPEC): Similarly.
18027 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
18030 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
18032 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
18034 2017-08-11 Tamar Christina <tamar.christina@arm.com>
18035 * config/aarch64/aarch64.md (mov<mode>): Change.
18036 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
18037 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
18038 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
18040 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
18042 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
18043 for storage order barriers.
18045 2017-08-11 Martin Liska <mliska@suse.cz>
18048 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
18049 variables of void type.
18051 2017-08-11 Martin Liska <mliska@suse.cz>
18053 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
18054 TARGET_SUPPORTS_ALIASES.
18055 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
18056 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
18057 (optimize_weakref): Likewise.
18058 * symtab.c (symtab_node::noninterposable_alias): Likewise.
18059 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
18060 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
18062 2017-08-11 Martin Liska <mliska@suse.cz>
18065 * config/i386/i386.c (make_resolver_func): Do complete
18066 refactoring of the function.
18068 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
18071 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
18072 * config/i386/i386.c (ix86_stack_protect_guard): Use
18073 ix86_stack_protect_guard_symbol_str to generate varible declaration.
18074 * doc/invoke.texi (x86 Options): Document
18075 -mstack-protector-guard-symbol= option.
18077 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
18079 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
18080 * config/i386/i386.c (ix86_split_stack_guard): New function.
18081 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
18082 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
18083 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
18084 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
18085 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
18086 (split_stack_space_check): Call ix86_split_stack_guard.
18088 2017-08-10 Martin Sebor <msebor@redhat.com>
18090 * print-tree.c (print_node): Print location using the established
18092 Replace spaces with colons.
18093 (debug_raw, debug): Ditto.
18095 2017-08-10 Martin Sebor <msebor@redhat.com>
18098 * pretty-print.c (pp_format): Correct the handling of %s precision.
18100 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
18103 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18105 (ix86_finalize_stack_frame_flags): This. Also clear
18106 frame_pointer_needed if -fno-omit-frame-pointer is used without
18108 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18109 with ix86_finalize_stack_frame_flags.
18110 (ix86_expand_epilogue): Likewise.
18111 (ix86_expand_split_stack_prologue): Likewise.
18112 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
18114 2017-08-10 Martin Liska <mliska@suse.cz>
18117 * c-attribs.c (handle_target_attribute):
18118 Report warning for an empty string argument of target attribute.
18120 2017-08-09 Jakub Jelinek <jakub@redhat.com>
18123 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
18125 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
18126 or DECL_NONLOCAL labels.
18127 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
18128 or DECL_NONLOCAL labels here.
18130 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
18132 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
18133 to indicate when early gimple folding has been disabled.
18134 (rs6000_gimple_fold_builtin): Add debug content.
18135 (rs6000_invalid_builtin): Fix whitespace.
18136 (rs6000_expand_builtin): Fix whitespace.
18137 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
18139 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
18142 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
18143 SAVE_MULTIPLE if not all the registers that saves, should be saved.
18145 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
18147 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
18148 (qdf24xx): Likewise.
18149 * config/aarch64/aarch64.md: Include falkor.md.
18150 * config/aarch64/falkor.md: New.
18152 2017-08-09 Marek Polacek <polacek@redhat.com>
18155 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
18156 * diagnostic.c (emit_diagnostic): Add a comment.
18157 (emit_diagnostic_valist): New function.
18159 2017-08-09 Marek Polacek <polacek@redhat.com>
18162 * input.c (make_location): New overload.
18163 * input.h (make_location): Declare.
18165 2017-08-08 Alan Modra <amodra@gmail.com>
18166 H.J. Lu <hongjiu.lu@intel.com>
18169 * gcc.c (NO_PIE_SPEC): Delete.
18170 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
18172 (LINK_PIE_SPEC): ..to here.
18173 (LINK_COMMAND_SPEC): Support -no-pie.
18174 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
18175 chain of crtbegin*.o selection, update for PIE_SPEC changes and
18177 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
18178 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
18179 (ENDFILE_CRTEND_SPEC): Similarly.
18181 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
18184 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
18185 (mstack-protector-guard-offset=): Ditto.
18186 * config/i386/i386.c (ix86_option_override): Handle
18187 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
18189 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
18190 ix86_stack_protect_guard_offset variables.
18191 (TARGET_STACK_PROTECT_GUARD): Always define.
18192 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
18193 and -mstack-protector-guard-offset= options.
18195 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18197 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
18198 boundary case for the last candidate.
18200 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18202 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
18203 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
18205 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18207 PR middle-end/19706
18208 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
18209 * config/aarch64/aarch64-builtins.c
18210 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
18211 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
18212 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
18214 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18215 Andrew Pinski <pinskia@gmail.com>
18217 PR middle-end/19706
18218 * internal-fn.def (XORSIGN): New.
18219 * optabs.def (xorsign_optab): New.
18220 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
18221 (convert_expand_mult_copysign): New.
18222 (pass_optimize_widening_mul::execute): Call
18223 convert_expand_mult_copysign.
18225 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18227 PR tree-optimization/81354
18228 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
18229 Insert on edges rather than explicitly creating landing pads.
18230 (analyze_candidates_and_replace): Commit edge inserts.
18232 2017-08-08 Richard Biener <rguenther@suse.de>
18234 PR middle-end/81719
18235 * tree-ssa-loop-niter.c: Include tree-dfa.h.
18236 (expand_simple_operations): Also look through ADDR_EXPRs with
18237 MEM_REF bases treating them as POINTER_PLUS_EXPR.
18239 2017-08-08 Richard Biener <rguenther@suse.de>
18241 PR tree-optimization/81723
18242 * tree-vect-slp.c (struct bst_traits): New hash traits.
18243 (bst_fail): New global.
18244 (vect_build_slp_tree_2): New worker, split out from ...
18245 (vect_build_slp_tree): ... this now wrapping it with using
18246 bst_fail set to cache SLP tree build fails. Properly handle
18248 (vect_analyze_slp_instance): Allocate and free bst_fail.
18250 2017-08-08 Martin Liska <mliska@suse.cz>
18253 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
18254 LABEL_DECLs that can be from a different function.
18256 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18258 PR tree-optimization/81744
18259 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
18260 loop's number of iterations.
18262 2017-08-08 Martin Liska <mliska@suse.cz>
18264 * asan.c: Include header files.
18265 * attribs.c (build_decl_attribute_variant): New function moved
18267 (build_type_attribute_qual_variant): Likewise.
18268 (cmp_attrib_identifiers): Likewise.
18269 (simple_cst_list_equal): Likewise.
18270 (omp_declare_simd_clauses_equal): Likewise.
18271 (attribute_value_equal): Likewise.
18272 (comp_type_attributes): Likewise.
18273 (build_type_attribute_variant): Likewise.
18274 (lookup_ident_attribute): Likewise.
18275 (remove_attribute): Likewise.
18276 (merge_attributes): Likewise.
18277 (merge_type_attributes): Likewise.
18278 (merge_decl_attributes): Likewise.
18279 (merge_dllimport_decl_attributes): Likewise.
18280 (handle_dll_attribute): Likewise.
18281 (attribute_list_equal): Likewise.
18282 (attribute_list_contained): Likewise.
18283 * attribs.h (lookup_attribute): New function moved from tree.[ch].
18284 (lookup_attribute_by_prefix): Likewise.
18285 * bb-reorder.c: Include header files.
18286 * builtins.c: Likewise.
18287 * calls.c: Likewise.
18288 * cfgexpand.c: Likewise.
18289 * cgraph.c: Likewise.
18290 * cgraphunit.c: Likewise.
18291 * convert.c: Likewise.
18292 * dwarf2out.c: Likewise.
18293 * final.c: Likewise.
18294 * fold-const.c: Likewise.
18295 * function.c: Likewise.
18296 * gimple-expr.c: Likewise.
18297 * gimple-fold.c: Likewise.
18298 * gimple-pretty-print.c: Likewise.
18299 * gimple.c: Likewise.
18300 * gimplify.c: Likewise.
18301 * hsa-common.c: Likewise.
18302 * hsa-gen.c: Likewise.
18303 * internal-fn.c: Likewise.
18304 * ipa-chkp.c: Likewise.
18305 * ipa-cp.c: Likewise.
18306 * ipa-devirt.c: Likewise.
18307 * ipa-fnsummary.c: Likewise.
18308 * ipa-inline.c: Likewise.
18309 * ipa-visibility.c: Likewise.
18311 * lto-cgraph.c: Likewise.
18312 * omp-expand.c: Likewise.
18313 * omp-general.c: Likewise.
18314 * omp-low.c: Likewise.
18315 * omp-offload.c: Likewise.
18316 * omp-simd-clone.c: Likewise.
18317 * opts-global.c: Likewise.
18318 * passes.c: Likewise.
18319 * predict.c: Likewise.
18320 * sancov.c: Likewise.
18321 * sanopt.c: Likewise.
18322 * symtab.c: Likewise.
18323 * toplev.c: Likewise.
18324 * trans-mem.c: Likewise.
18325 * tree-chkp.c: Likewise.
18326 * tree-eh.c: Likewise.
18327 * tree-into-ssa.c: Likewise.
18328 * tree-object-size.c: Likewise.
18329 * tree-parloops.c: Likewise.
18330 * tree-profile.c: Likewise.
18331 * tree-ssa-ccp.c: Likewise.
18332 * tree-ssa-live.c: Likewise.
18333 * tree-ssa-loop.c: Likewise.
18334 * tree-ssa-sccvn.c: Likewise.
18335 * tree-ssa-structalias.c: Likewise.
18336 * tree-ssa.c: Likewise.
18337 * tree-streamer-in.c: Likewise.
18338 * tree-vectorizer.c: Likewise.
18339 * tree-vrp.c: Likewise.
18340 * tsan.c: Likewise.
18341 * ubsan.c: Likewise.
18342 * varasm.c: Likewise.
18343 * varpool.c: Likewise.
18344 * tree.c: Remove functions moved to attribs.[ch].
18345 * tree.h: Likewise.
18346 * config/aarch64/aarch64.c: Add attrs.h header file.
18347 * config/alpha/alpha.c: Likewise.
18348 * config/arc/arc.c: Likewise.
18349 * config/arm/arm.c: Likewise.
18350 * config/avr/avr.c: Likewise.
18351 * config/bfin/bfin.c: Likewise.
18352 * config/c6x/c6x.c: Likewise.
18353 * config/cr16/cr16.c: Likewise.
18354 * config/cris/cris.c: Likewise.
18355 * config/darwin.c: Likewise.
18356 * config/epiphany/epiphany.c: Likewise.
18357 * config/fr30/fr30.c: Likewise.
18358 * config/frv/frv.c: Likewise.
18359 * config/ft32/ft32.c: Likewise.
18360 * config/h8300/h8300.c: Likewise.
18361 * config/i386/winnt.c: Likewise.
18362 * config/ia64/ia64.c: Likewise.
18363 * config/iq2000/iq2000.c: Likewise.
18364 * config/lm32/lm32.c: Likewise.
18365 * config/m32c/m32c.c: Likewise.
18366 * config/m32r/m32r.c: Likewise.
18367 * config/m68k/m68k.c: Likewise.
18368 * config/mcore/mcore.c: Likewise.
18369 * config/microblaze/microblaze.c: Likewise.
18370 * config/mips/mips.c: Likewise.
18371 * config/mmix/mmix.c: Likewise.
18372 * config/mn10300/mn10300.c: Likewise.
18373 * config/moxie/moxie.c: Likewise.
18374 * config/msp430/msp430.c: Likewise.
18375 * config/nds32/nds32-isr.c: Likewise.
18376 * config/nds32/nds32.c: Likewise.
18377 * config/nios2/nios2.c: Likewise.
18378 * config/nvptx/nvptx.c: Likewise.
18379 * config/pa/pa.c: Likewise.
18380 * config/pdp11/pdp11.c: Likewise.
18381 * config/powerpcspe/powerpcspe.c: Likewise.
18382 * config/riscv/riscv.c: Likewise.
18383 * config/rl78/rl78.c: Likewise.
18384 * config/rx/rx.c: Likewise.
18385 * config/s390/s390.c: Likewise.
18386 * config/sh/sh.c: Likewise.
18387 * config/sol2.c: Likewise.
18388 * config/sparc/sparc.c: Likewise.
18389 * config/spu/spu.c: Likewise.
18390 * config/stormy16/stormy16.c: Likewise.
18391 * config/tilegx/tilegx.c: Likewise.
18392 * config/tilepro/tilepro.c: Likewise.
18393 * config/v850/v850.c: Likewise.
18394 * config/vax/vax.c: Likewise.
18395 * config/visium/visium.c: Likewise.
18396 * config/xtensa/xtensa.c: Likewise.
18398 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18401 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
18402 constraints since the -mupper-regs-* switches have been
18404 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
18405 into a vector from a double word element that was extracted from
18406 another vector, and eliminate extra XXPERMDI instructions.
18407 (vsx_concat_<mode>_2): Likewise.
18408 (vsx_concat_<mode>_3): Likewise.
18409 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
18410 concat to allow optimizing inserts from previous extracts.
18412 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
18414 * config/i386/i386.c (ix86_stack_protect_guard): Generate
18415 memory reference to a SSP offset in TLS address space.
18416 (ix86_print_operand) <case '@'>: Remove.
18417 (ix86_print_operand_punct_valid_p): Remove '@' code.
18418 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
18419 UNSPEC_SP_TLS_TEST.
18420 (stack_tls_protect_set_<mode>): Remove.
18421 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
18422 (stack_tls_protect_test_<mode>): Remove.
18423 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
18425 2017-08-07 Olivier Hainque <hainque@adacore.com>
18428 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
18430 2017-08-07 Douglas Rupp <rupp@adacore.com>
18432 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
18433 variable was referenced as multidir in command.
18435 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18438 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
18441 2017-08-07 Martin Liska <mliska@suse.cz>
18443 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
18444 * config/rl78/rl78.c: Add include of attribs.h.
18445 * config/sh/sh.c: Likewise.
18446 * config/v850/v850.c: Likewise.
18448 2017-08-07 Tom de Vries <tom@codesourcery.com>
18450 PR middle-end/78266
18451 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
18453 2017-08-07 Martin Liska <mliska@suse.cz>
18455 * config/mips/mips.c: Include attribs.h.
18457 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
18460 * doc/invoke.texi: Document change in behvaior for -Ofast for
18463 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
18465 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
18467 (aarch64_pop_regs): Likewise.
18468 (aarch64_gen_load_pair): Likewise.
18469 (aarch64_save_callee_saves): Likewise.
18470 (aarch64_restore_callee_saves): Likewise.
18472 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18474 * config/i386/i386.c: Revert the last change.
18476 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18479 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18481 (ix86_finalize_stack_frame_flags): This. Also clear
18482 frame_pointer_needed if -fno-omit-frame-pointer is used without
18484 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18485 with ix86_finalize_stack_frame_flags.
18486 (ix86_expand_epilogue): Likewise.
18487 (ix86_expand_split_stack_prologue): Likewise.
18489 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18492 * config/i386/i386.c (get_builtin_code_for_version): Set priority
18493 to P_AES for Westmere.
18495 2017-08-07 Jonathan Yong <10walls@gmail.com>
18497 * config/i386/mingw.opt (fset-stack-executable): Removed.
18498 * config/i386/cygming.opt (fset-stack-executable): Moved
18500 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
18502 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
18504 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
18506 2017-08-07 Marek Polacek <polacek@redhat.com>
18508 PR middle-end/81737
18509 * fold-const.c (fold_indirect_ref_1): Check type_domain.
18511 2017-08-07 Martin Liska <mliska@suse.cz>
18513 * attribs.h (canonicalize_attr_name): New function.
18514 (cmp_attribs): Move from c-format.c and adjusted.
18515 (is_attribute_p): Moved from tree.h.
18516 * tree-inline.c: Add new includes.
18517 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
18518 (private_is_attribute_p): Remove.
18519 (private_lookup_attribute): Likewise.
18520 (private_lookup_attribute_by_prefix): Simplify.
18521 (remove_attribute): Use is_attribute_p.
18522 * tree.h: Remove removed declarations.
18524 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18526 PR middle-end/81698
18527 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
18528 instead of computing it in the function. Formatting fix.
18529 (expand_case): Don't rely on default_edge being the first edge,
18530 clear it if removing it, pass default_edge to
18531 emit_case_dispatch_table.
18532 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
18535 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
18537 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
18538 insn in the function, emit NOP after the insn.
18540 2017-08-06 Tom de Vries <tom@codesourcery.com>
18542 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
18545 2017-08-06 Tom de Vries <tom@codesourcery.com>
18547 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
18550 2017-08-04 Yury Gribov <tetra2005@gmail.com>
18552 PR tree-optimization/57371
18553 * match.pd: New pattern.
18555 2017-08-04 Marek Polacek <polacek@redhat.com>
18557 PR middle-end/81695
18558 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
18559 perform the computation in offset_int.
18561 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18563 PR tree-optimization/81136
18564 * tree-vectorizer.h: Include tree-hash-traits.h.
18565 (vec_base_alignments): New typedef.
18566 (vec_info): Add a base_alignments field.
18567 (vect_record_base_alignments): Declare.
18568 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
18570 (DR_IS_CONDITIONAL_IN_STMT): New macro.
18571 (create_data_ref): Add an is_conditional_in_stmt argument.
18572 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
18573 the is_conditional_in_stmt field.
18574 (data_ref_loc): Add an is_conditional_in_stmt field.
18575 (get_references_in_stmt): Set the is_conditional_in_stmt field.
18576 (find_data_references_in_stmt): Update call to create_data_ref.
18577 (graphite_find_data_references_in_stmt): Likewise.
18578 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
18579 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
18580 (vect_record_base_alignment): New function.
18581 (vect_record_base_alignments): Likewise.
18582 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
18583 for nested statements even if we fail to compute a misalignment.
18584 Use pooled base alignments for unconditional references.
18585 (vect_find_same_alignment_drs): Compare base addresses instead
18587 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
18588 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
18590 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18592 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
18593 Add an explicit name for the enum. Use auto_vec for slp_instances
18594 and grouped_stores.
18595 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
18597 (_bb_vec_info): Add a constructor and destructor.
18598 (vinfo_for_stmt): Return NULL for uids of -1 as well.
18599 (destroy_loop_vec_info): Delete.
18600 (vect_destroy_datarefs): Likewise.
18601 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
18602 (vec_info::vec_info): New function.
18603 (vec_info::~vec_info): Likewise.
18604 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
18605 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
18606 destroy_loop_vec_info.
18607 * tree-vect-loop.c (new_loop_vec_info): Replace with...
18608 (_loop_vec_info::_loop_vec_info): ...this.
18609 (destroy_loop_vec_info): Replace with...
18610 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
18611 the stmt_vec_infos. Leave handling of vec_info information to its
18612 destructor. Remove explicit vector releases.
18613 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
18614 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
18615 * tree-vect-slp.c (new_bb_vec_info): Replace with...
18616 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
18617 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
18618 (destroy_bb_vec_info): Replace with...
18619 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
18620 information to its destructor.
18621 (vect_slp_analyze_bb_1): Use new and delete instead of
18622 new_bb_vec_info and destroy_bb_vec_info.
18623 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
18626 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18628 * tree-data-ref.h (subscript): Add access_fn field.
18629 (data_dependence_relation): Add could_be_independent_p.
18630 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
18631 (same_access_functions): Move to tree-data-ref.c.
18632 * tree-data-ref.c (ref_contains_union_access_p): New function.
18633 (access_fn_component_p): Likewise.
18634 (access_fn_components_comparable_p): Likewise.
18635 (dr_analyze_indices): Add a reference to access_fn_component_p.
18636 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
18638 (constant_access_functions): Likewise.
18639 (add_other_self_distances): Likewise.
18640 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
18641 (initialize_data_dependence_relation): Use XCNEW and remove
18642 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
18643 of access functions that have the same type. Allow the
18644 subsequence to end with different bases in some circumstances.
18645 Record the chosen access functions in SUB_ACCESS_FN.
18646 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
18647 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
18648 (subscript_dependence_tester_1): Likewise dra and drb.
18649 (build_classic_dist_vector): Update calls accordingly.
18650 (subscript_dependence_tester): Likewise.
18651 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
18652 DDR_COULD_BE_INDEPENDENT_P.
18653 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
18654 comp_alias_ddrs instead of may_alias_ddrs.
18655 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
18657 (vect_analyze_data_ref_dependence): Use it if
18658 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
18659 distance vectors if that fails.
18660 (dependence_distance_ge_vf): New function.
18661 (vect_prune_runtime_alias_test_list): Use it. Don't clear
18662 LOOP_VINFO_MAY_ALIAS_DDRS.
18664 2017-08-04 Richard Biener <rguenther@suse.de>
18666 PR middle-end/81705
18667 * fold-const.c (fold_binary_loc): Properly restrict
18668 minus_var0 && minus_var1 case when associating undefined overflow
18671 2017-08-04 Tom de Vries <tom@codesourcery.com>
18673 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
18675 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18677 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18678 Don't start diagnostic messages with a capital letter.
18679 * config/rs6000/rs6000.c (rs6000_option_override_internal):
18681 (rs6000_invalid_builtin): Likewise.
18682 (rs6000_trampoline_init): Likewise.
18684 2017-08-03 Jakub Jelinek <jakub@redhat.com>
18687 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
18688 after setting changeable df flags.
18690 2017-08-03 Richard Biener <rguenther@suse.de>
18692 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
18693 up if the use is in USE - X.
18695 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
18697 * toplev.c (dumpfile.h): New include.
18698 (internal_error_reentered): New static function. Use it...
18699 (internal_error_function): ...here to handle reentered internal_error.
18701 2017-08-03 Richard Biener <rguenther@suse.de>
18703 PR middle-end/81148
18704 * fold-const.c (split_tree): Add minus_var and minus_con
18705 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
18706 here but always use minus_*.
18707 (associate_trees): Assert we never associate with MINUS_EXPR
18708 and NULL first operand. Do not recurse for PLUS_EXPR operands
18709 when associating as MINUS_EXPR either.
18710 (fold_binary_loc): Track minus_var and minus_con.
18712 2017-08-03 Tom de Vries <tom@codesourcery.com>
18715 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
18716 ACCEL_COMPILER, apply finish_options on
18717 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
18719 2017-08-03 Tom de Vries <tom@codesourcery.com>
18722 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
18723 function_entry_patch_area_size > 0.
18725 2017-08-03 Jakub Jelinek <jakub@redhat.com>
18728 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
18729 instead of 10??LU, perform unit multiplication in wide_int,
18730 don't change alloc_object_size_limit if the limit is larger
18733 PR tree-optimization/81655
18734 PR tree-optimization/81588
18735 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
18736 the case when ranges[i].low and high are 1 for unsigned type with
18739 PR middle-end/81052
18740 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
18741 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
18743 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18745 * tree-vrp.h: Add include guard.
18747 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
18750 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
18751 (ud2): New insn pattern.
18752 * config/i386/i386.c (ix86_expand_epilogue):
18753 For naked functions, generate ud2 instead of trap insn.
18755 2017-08-02 Marek Polacek <polacek@redhat.com>
18758 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
18760 2017-08-02 Tom de Vries <tom@codesourcery.com>
18761 Cesar Philippidis <cesar@codesourcery.com>
18763 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
18764 Add missing edge probabilities.
18766 2017-08-02 Tamar Christina <tamar.christina@arm.com>
18768 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
18769 Correct endianness.
18771 2017-08-02 Jakub Jelinek <jakub@redhat.com>
18773 PR middle-end/79499
18774 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
18775 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
18776 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
18777 prologue_seq sequences - if any.
18779 2017-08-02 Richard Biener <rguenther@suse.de>
18781 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
18782 via vectors if supported, integer extracts via punning if supported
18783 or otherwise vector extracts.
18785 2017-08-02 Richard Biener <rguenther@suse.de>
18787 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
18789 (bitmap_insert_into_set): ... this.
18791 2017-08-02 Richard Biener <rguenther@suse.de>
18793 PR tree-optimization/81633
18795 2015-08-17 Alan Hayward <alan.hayward@arm.com>
18797 PR tree-optimization/71752
18798 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
18800 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
18802 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
18803 (machine_function::call_ms2sysv_pad_out): Remove field.
18804 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
18805 (ix86_compute_frame_layout): Likewise.
18807 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
18810 * config/i386/i386.c (ix86_set_func_type): Disallow naked
18811 attribute with interrupt attribute.
18813 2017-08-01 Andrew Pinski <apinski@cavium.com>
18815 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
18816 BIT_INSERT_EXPR's operand 1
18817 to see if the types precision matches.
18819 2017-08-01 Martin Liska <mliska@suse.cz>
18821 PR middle-end/70140
18822 * builtins.c (expand_builtin_memcpy_args): Remove.
18823 (expand_builtin_memcpy): Call newly added function
18824 expand_builtin_memory_copy_args.
18825 (expand_builtin_memcpy_with_bounds): Likewise.
18826 (expand_builtin_mempcpy): Remove last argument.
18827 (expand_builtin_mempcpy_with_bounds): Likewise.
18828 (expand_builtin_memory_copy_args): New function created from
18829 expand_builtin_mempcpy_args with small modifications.
18830 (expand_builtin_mempcpy_args): Remove.
18831 (expand_builtin_stpcpy): Remove unused argument.
18832 (expand_builtin): Likewise.
18833 (expand_builtin_with_bounds): Likewise.
18835 2017-08-01 Martin Liska <mliska@suse.cz>
18838 Make mempcpy more optimal (PR middle-end/70140).
18840 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18843 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
18844 __builtin_vec_cmpne verify both arguments are compatible vectors
18845 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
18846 verify arg1_type is a pointer or array type. For __builtin_vec_st,
18847 move computation of aligned to after checking the argument types.
18851 * config/rs6000/vsx.md (vextract_fp_from_shorth,
18852 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
18855 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
18856 Doug Rupp <rupp@adacore.com>
18857 Olivier Hainque <hainque@adacore.com>
18859 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
18860 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
18862 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
18863 for TARGET_OS_CPP_BUILTIN.
18864 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
18865 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
18867 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
18868 passing required abi options to the assembler for EABI configurations.
18869 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
18870 of .text.hot and .text.unlikely sections for kernel modules when
18871 using ARM style exceptions.
18872 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
18873 options. Add EXTRA_CC1_SPEC.
18874 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
18876 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
18878 (ARM_TARGET2_DWARF_FORMAT): Define.
18879 * config/arm/t-vxworks: Adjust multilib control to removal of the
18880 Diab command line options.
18882 2017-08-01 Martin Liska <mliska@suse.cz>
18884 PR gcov-profile/81561
18885 * gcov.c (unblock): Make unblocking safe as we need to preserve
18886 index correspondence of blocks and block_lists.
18888 2017-08-01 Richard Biener <rguenther@suse.de>
18890 PR tree-optimization/81181
18891 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
18892 (compute_antic): ... end of iteration here.
18894 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
18896 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
18897 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
18898 (ftree-slp-vectorize): Likewise.
18899 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
18900 can no longer be set independent of flag_tree_loop_vectorize.
18901 * omp-general.c (emp_max_vf): Likewise.
18902 * opts.c (enable_fdo_optimizations): Remove references to
18903 flag_tree_vectorize, these are now implicit.
18904 (common_handle_option): Remove handling for OPT_ftree_vectorize,
18905 and leave it for the options machinery.
18907 2017-08-01 Martin Liska <mliska@suse.cz>
18909 PR middle-end/70140
18910 * builtins.c (expand_builtin_memcpy_args): Remove.
18911 (expand_builtin_memcpy): Call newly added function
18912 expand_builtin_memory_copy_args.
18913 (expand_builtin_memcpy_with_bounds): Likewise.
18914 (expand_builtin_mempcpy): Remove last argument.
18915 (expand_builtin_mempcpy_with_bounds): Likewise.
18916 (expand_builtin_memory_copy_args): New function created from
18917 expand_builtin_mempcpy_args with small modifications.
18918 (expand_builtin_mempcpy_args): Remove.
18919 (expand_builtin_stpcpy): Remove unused argument.
18920 (expand_builtin): Likewise.
18921 (expand_builtin_with_bounds): Likewise.
18923 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18926 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
18927 print "ds:" only for immediates in generic address space.
18929 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18932 * config/i386/i386.c (ix86_funciton_naked): New prototype.
18933 (ix86_function_ok_for_sibcall): Return false for naked functions.
18935 2017-08-01 Richard Biener <rguenther@suse.de>
18937 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
18938 (compute_antic): Seed worklist with exit block predecessors.
18939 * cfganal.c (dfs_find_deadend): For a cycle return the source
18940 of the edge closing it.
18942 2017-08-01 Tamar Christina <tamar.christina@arm.com>
18944 * config/aarch64/aarch64.c
18945 (aarch64_can_const_movi_rtx_p): Move 0 check.
18947 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18949 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
18950 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
18952 * match.pd: Ditto in address comparison pattern.
18954 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18956 PR tree-optimization/81627
18957 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
18958 closed ssa form for store-store chain.
18960 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18962 PR tree-optimization/81620
18963 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
18964 for store-store chain.
18966 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18968 PR tree-optimization/81588
18969 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
18970 ranges[i].in_p, invert comparison code ccode. For >/>=,
18971 swap rhs1 and rhs2 and comparison code unconditionally,
18972 for </<= don't do that. Don't swap rhs1/rhs2 again if
18973 ranges[i].in_p, instead invert comparison code ccode if
18974 opcode or oe->rank is BIT_IOR_EXPR.
18977 * optabs.def (vec_extract_optab, vec_init_optab): Change from
18978 a direct optab to conversion optab.
18979 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
18980 with GET_MODE_INNER as last argument instead of optab_handler.
18981 * expmed.c (extract_bit_field_1): Likewise. Use vector from
18982 vector extraction if possible and optab is available.
18983 * expr.c (store_constructor): Use convert_optab_handler instead
18984 of optab_handler. Use vector initialization from smaller
18985 vectors if possible and optab is available.
18986 * tree-vect-stmts.c (vectorizable_load): Likewise.
18987 * doc/md.texi (vec_extract, vec_init): Document that the optabs
18988 now have two modes.
18989 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
18990 of vec_init from half-sized vectors with the same element mode.
18991 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
18992 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
18993 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
18994 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
18995 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
18996 after mode in gen_vec_extract* calls.
18997 (vec_extract<mode>): Renamed to ...
18998 (vec_extract<mode><ssescalarmodelower>): ... this.
18999 (vec_extract<mode><ssehalfvecmodelower>): New expander.
19000 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
19001 element mode after mode in gen_vec_init* calls.
19002 (VEC_INIT_HALF_MODE): New mode iterator.
19003 (vec_init<mode>): Renamed to ...
19004 (vec_init<mode><ssescalarmodelower>): ... this.
19005 (vec_init<mode><ssehalfvecmodelower>): New expander.
19006 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
19007 (vec_extractv2sfsf): ... this.
19008 (vec_initv2sf): Renamed to ...
19009 (vec_initv2sfsf): ... this.
19010 (vec_extractv2si): Renamed to ...
19011 (vec_extractv2sisi): ... this.
19012 (vec_initv2si): Renamed to ...
19013 (vec_initv2sisi): ... this.
19014 (vec_extractv4hi): Renamed to ...
19015 (vec_extractv4hihi): ... this.
19016 (vec_initv4hi): Renamed to ...
19017 (vec_initv4hihi): ... this.
19018 (vec_extractv8qi): Renamed to ...
19019 (vec_extractv8qiqi): ... this.
19020 (vec_initv8qi): Renamed to ...
19021 (vec_initv8qiqi): ... this.
19022 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
19023 (vec_init<mode>): Renamed to ...
19024 (vec_init<mode><VEC_base_l>): ... this.
19025 (vec_extract<mode>): Renamed to ...
19026 (vec_extract<mode><VEC_base_l>): ... this.
19027 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
19028 (vec_initv2sfsf): ... this.
19029 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
19030 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
19031 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
19032 element mode after mode in gen_vec_init* calls.
19033 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
19034 (vec_init<mode><Vel>): ... this.
19035 (vec_extract<mode>): Renamed to ...
19036 (vec_extract<mode><Vel>): ... this.
19037 * config/aarch64/iterators.md (Vel): New mode attribute.
19038 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
19039 Add element mode after mode in gen_vec_extract* calls.
19040 * config/s390/vector.md (non_vec_l): New mode attribute.
19041 (vec_extract<mode>): Renamed to ...
19042 (vec_extract<mode><non_vec_l>): ... this.
19043 (vec_init<mode>): Renamed to ...
19044 (vec_init<mode><non_vec_l>): ... this.
19045 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
19046 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
19048 * config/arm/iterators.md (V_elem_l): New mode attribute.
19049 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
19050 (vec_extract<mode><V_elem_l>): ... this.
19051 (vec_extractv2di): Renamed to ...
19052 (vec_extractv2didi): ... this.
19053 (vec_init<mode>): Renamed to ...
19054 (vec_init<mode><V_elem_l>): ... this.
19055 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
19056 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
19057 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
19058 Add element mode after gen_vec_extract* calls.
19059 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
19060 (vec_init<mode><unitmode>): ... this.
19061 (vec_extract<mode>): Renamed to ...
19062 (vec_extract<mode><unitmode>): ... this.
19063 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
19064 (vec_init<mode><unitmode>): ... this.
19065 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
19066 (vec_initv2sfsf): ... this.
19067 (vec_extractv2sf): Renamed to ...
19068 (vec_extractv2sfsf): ... this.
19069 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
19070 Add element mode after gen_vec_extract* calls.
19071 * config/mips/mips.md (unitmode): New mode iterator.
19072 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
19073 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
19074 * config/spu/spu.md (inner_l): New mode attribute.
19075 (vec_init<mode>): Renamed to ...
19076 (vec_init<mode><inner_l>): ... this.
19077 (vec_extract<mode>): Renamed to ...
19078 (vec_extract<mode><inner_l>): ... this.
19079 * config/sparc/sparc.md (veltmode): New mode iterator.
19080 (vec_init<VMALL:mode>): Renamed to ...
19081 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
19082 * config/ia64/vect.md (vec_initv2si): Renamed to ...
19083 (vec_initv2sisi): ... this.
19084 (vec_initv2sf): Renamed to ...
19085 (vec_initv2sfsf): ... this.
19086 (vec_extractv2sf): Renamed to ...
19087 (vec_extractv2sfsf): ... this.
19088 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
19089 (vec_init<mode>): Renamed to ...
19090 (vec_init<mode><VEC_base_l>): ... this.
19091 (vec_extract<mode>): Renamed to ...
19092 (vec_extract<mode><VEC_base_l>): ... this.
19093 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
19094 (vec_initv2sfsf): ... this.
19095 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
19096 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
19097 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
19098 gen_vec_init* calls.
19100 2017-08-01 Richard Biener <rguenther@suse.de>
19102 PR tree-optimization/81297
19103 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
19104 TREE_OVERFLOW from INTEGER_CSTs.
19106 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
19108 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
19110 2017-07-31 Carl Love <cel@us.ibm.com>
19112 * config/rs6000/rs6000-c: Add support for built-in functions
19113 vector signed char vec_xl_be (signed long long, signed char *);
19114 vector unsigned char vec_xl_be (signed long long, unsigned char *);
19115 vector signed int vec_xl_be (signed long long, signed int *);
19116 vector unsigned int vec_xl_be (signed long long, unsigned int *);
19117 vector signed long long vec_xl_be (signed long long, signed long long *);
19118 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
19119 vector signed short vec_xl_be (signed long long, signed short *);
19120 vector unsigned short vec_xl_be (signed long long, unsigned short *);
19121 vector double vec_xl_be (signed long long, double *);
19122 vector float vec_xl_be (signed long long, float *);
19123 * config/rs6000/altivec.h (vec_xl_be): Add #define.
19124 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
19125 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
19127 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
19128 (altivec_expand_builtin): Add switch statement to call
19129 altivec_expand_xl_be for each builtin.
19130 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
19131 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
19132 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
19133 __builtin_vsx_le_be_v16qi.
19134 * doc/extend.texi: Update the built-in documentation file for the
19135 new built-in functions.
19137 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19140 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
19142 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
19144 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19146 * config.gcc: Add z14.
19147 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
19148 CPU model numbers for z13s and z14.
19149 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
19151 * config/s390/s390-opts.h (enum processor_type): Rename
19152 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19153 * config/s390/s390.c (processor_table): Add field for CPU name to
19154 be passed to Binutils.
19155 (s390_asm_output_machine_for_arch): Use the new field in
19156 processor_table for Binutils.
19157 (s390_expand_builtin): Replace arch12 with z14.
19158 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19159 (s390_get_sched_attrmask): Likewise.
19160 (s390_get_unit_mask): Likewise.
19161 * config/s390/s390.opt: Add z14 to processor_type enum.
19163 2017-07-31 Martin Jambor <mjambor@suse.cz>
19166 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
19167 regardless of optimization level.
19169 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
19170 Martin Liska <mliska@suse.cz>
19172 * predict.def: Remove old comment and adjust probability.
19173 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
19174 PREDICT statements.
19176 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19179 * config/i386/i386.c (ix86_function_naked): New function.
19180 (ix86_can_use_return_insn_p): Return false for naked functions.
19181 (ix86_expand_prologue): Skip prologue for naked functions.
19182 (ix86_expand_epilogue): Skip epilogue for naked functions
19183 and emit trap instruction.
19184 (ix86_warn_func_return): New function.
19185 (ix86_attribute_table): Add "naked" attribute specification.
19186 (TARGET_WARN_FUNC_RETURN): Define.
19187 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
19189 2017-07-31 Martin Liska <mliska@suse.cz>
19191 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
19192 (dump_gimple_bb_header): Always dump BB info.
19193 (pp_cfg_jump): Do not append info about BB when dumping a jump.
19195 2017-07-31 Martin Liska <mliska@suse.cz>
19198 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
19199 also with current_function_decl non-null equality.
19201 2017-07-31 Jakub Jelinek <jakub@redhat.com>
19204 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
19205 change type to the element type, instead add eltype variable and
19206 use it where we are interested in the element type.
19208 PR tree-optimization/81603
19209 * ipa-polymorphic-call.c
19210 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
19211 offset arithmetic in offset_int, bail out if the resulting bit offset
19212 doesn't fit into shwi.
19214 2017-07-31 Martin Liska <mliska@suse.cz>
19216 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
19217 (gimplify_save_expr): Fix comment.
19219 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
19222 * config/i386/i386.c (ix86_function_arg): Update arguments for
19224 (ix86_compute_frame_layout): Set the initial stack offset to
19225 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
19226 INCOMING_FRAME_SP_OFFSET.
19227 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
19228 stack before exception handler returns.
19229 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
19230 the 'ERROR_CODE' for exception handler.
19232 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
19234 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
19235 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
19236 (ASM_OUTPUT_REG_POP): Ditto.
19237 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
19238 instead of asm_fprintf to output pure string.
19240 2017-07-29 Jakub Jelinek <jakub@redhat.com>
19242 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
19243 to imported_module_or_decl hook.
19244 (debug_nothing_tree_tree_tree_bool): Remove.
19245 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
19246 * debug.c (do_nothing_debug_hooks): Use
19247 debug_nothing_tree_tree_tree_bool_bool instead of
19248 debug_nothing_tree_tree_tree_bool.
19249 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19250 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
19251 * sdbout.c (sdb_debug_hooks): Likewise.
19252 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
19253 (gen_namespace_die): Add DW_AT_export_symbols attribute if
19255 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
19256 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
19257 attribute, don't add anything.
19259 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19261 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
19262 (fold_build2_stat_loc): Likewise.
19263 (fold_build3_stat_loc): Likewise.
19264 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
19265 (fold_build1_loc): Remove macro.
19266 (fold_build2_loc): Likewise.
19267 (fold_build3_loc): Likewise.
19269 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19271 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
19272 (gimple_build_debug_bind_source_stat): Likewise.
19273 * gimple.h (gimple_build_debug_bind): Remove macro.
19274 (gimple_build_debug_bind_source): Likewise.
19276 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19278 * bitmap.c (bitmap_alloc): Adjust.
19279 (bitmap_gc_alloc): Likewise.
19280 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
19282 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19284 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
19285 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
19286 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
19287 (bitmap_gc_alloc_stat): Likewise.
19288 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
19290 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19292 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
19293 * rtl.h (shallow_copy_rtx): Remove macro.
19295 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19297 * emit-rtl.c (gen_raw_REG): Adjust.
19298 * gengenrtl.c (gendef): Likewise.
19299 * rtl.c (rtx_alloc_stat): Remove _stat from name.
19300 * rtl.h (rtx_alloc): Remove macro.
19302 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19304 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
19305 (build_tree_list_stat): Likewise.
19306 * tree.h (build_tree_list): Remove macro.
19307 (build_tree_list_vec): Likewise.
19309 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19311 * tree.c (make_vector_stat): Remove _stat from name.
19312 (build_vector_stat): Likewise.
19313 * tree.h (make_vector_stat): Remove macro.
19314 (build_vector_stat): Likewise.
19316 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19318 * tree.h (build_var_debug_value): Remove prototype.
19320 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19322 * tree.c (tree_cons_stat): Remove _stat from name.
19323 * tree.h (tree_cons): Remove macro.
19325 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19327 * tree.c (build_vl_exp_stat): Remove _stat from name.
19328 * tree.h (build_vl_exp): Remove macro.
19330 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19332 * tree.c (build_decl_stat): Remove _stat from name.
19333 * tree.h (build_decl): Remove macro.
19335 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19337 * gimple.c (gimple_build_with_ops_stat): Adjust.
19338 (gimple_alloc_stat): Remove _stat from name.
19339 * gimple.h (gimple_alloc): Remove macro.
19341 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19343 * tree.c (make_tree_vec_stat): Remove _stat from name.
19344 (grow_tree_vec_stat): Likewise.
19345 * tree.h (make_tree_vec_stat): Adjust prototype.
19346 (grow_tree_vec_stat): Likewise.
19347 (make_tree_vec): Remove macro.
19348 (grow_tree_vec): Likewise.
19350 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19352 * fold-const.c (fold_build1_stat_loc): Adjust.
19353 (fold_build2_stat_loc): Likewise.
19354 (fold_build3_stat_loc): Likewise.
19355 * tree.c (build0_stat): Remove _stat from name.
19356 (build1_stat): Likewise.
19357 (build2_stat): Likewise.
19358 (build3_stat): Likewise.
19359 (build4_stat): Likewise.
19360 (build5_stat): Likewise.
19361 * tree.h (build1_loc): Remove macro, and rename _stat function
19363 (build2_loc): Likewise.
19364 (build3_loc): Likewise.
19365 (build4_loc): Likewise.
19366 (build5_loc): Likewise.
19368 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19370 * tree.c (make_int_cst_stat): Remove _stat from name.
19371 * tree.h (make_int_cst_stat): Adjust prototype.
19372 (make_int_cst): Remove macro.
19374 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19376 * tree.c (make_tre_binfo_stat): Remove _stat from name.
19377 * tree.h (make_tree_binfo_stat): Adjust prototype.
19378 (make_tree_binfo): Remove.
19380 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19382 * tree.c (copy_node_stat): Rename to copy_node.
19383 (build_distinct_type_copy): Adjust.
19384 * tree.h (copy_node_stat): Adjust prototype.
19385 (copy_node): Remove macro.
19387 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19389 * tree.c (make_node_stat): rename to make_node.
19390 (build_tree_list_stat): Adjust.
19391 (build0_stat): Likewise.
19392 (build2_stat): Likewise.
19393 (build3_stat): Likewise.
19394 (build4_stat): Likewise.
19395 (build5_stat): Likewise.
19396 (build_decl_stat): Likewise.
19397 * tree.h (make_node_stat): Adjust prototype.
19398 (make_node): remove macro.
19400 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
19402 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
19403 (PPC_FEATURE2_SCV): Likewise.
19404 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
19406 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19408 * config/aarch64/aarch64.c
19409 (aarch64_internal_mov_immediate): Add new special pattern.
19410 * config/aarch64/aarch64.md (*movdi_aarch64):
19411 Add reg/32bit const mov case.
19413 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19414 Richard Sandiford <richard.sandiford@linaro.org>
19416 * config/aarch64/aarch64.md (mov<mode>): Generalize.
19417 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
19418 Add integer and movi cases.
19419 (movi-split-hf-df-sf split, fp16): New.
19420 (enabled): Added TARGET_FP_F16INST.
19421 * config/aarch64/iterators.md (GPF_HF): New.
19422 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
19424 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19426 * config/aarch64/aarch64.c
19427 (aarch64_simd_container_mode): Add prototype.
19428 (aarch64_expand_mov_immediate): Add HI support.
19429 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
19430 (aarch64_can_const_movi_rtx_p): New.
19431 (aarch64_preferred_reload_class):
19432 Remove restrictions of using FP registers for certain SIMD operations.
19433 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
19434 (aarch64_valid_floating_const): Add integer move validation.
19435 (aarch64_simd_imm_scalar_p): Remove.
19436 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
19437 (aarch64_legitimate_constant_p): Expand list of supported cases.
19438 * config/aarch64/aarch64-protos.h
19439 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
19440 (aarch64_reinterpret_float_as_int): New.
19441 (aarch64_simd_imm_scalar_p): Remove.
19442 * config/aarch64/constraints.md (Uvi): New.
19443 (Dd): Split into Ds and new Dd.
19444 * config/aarch64/aarch64.md (*movsi_aarch64):
19446 (*movdi_aarch64): Add SIMD mov case.
19448 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19450 * tree-predcom.c: (struct chain): Handle store-store chain in which
19451 stores for elimination only store loop invariant values.
19452 (execute_pred_commoning_chain): Ditto.
19453 (prepare_initializers_chain_store_elim): Ditto.
19454 (prepare_finalizers): Ditto.
19455 (is_inv_store_elimination_chain): New function.
19456 (initialize_root_vars_store_elim_1): New function.
19458 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19460 * tree-predcom.c: Revise general description of the pass.
19461 (enum chain_type): New enum type for store elimination.
19462 (struct chain): New field supporting store elimination.
19463 (struct component): Ditto.
19464 (dump_chain): Dump store-stores chain.
19465 (release_chain): Release resources.
19466 (split_data_refs_to_components): Compute and create component
19467 contains only stores for elimination.
19468 (get_chain_last_ref_at): New function.
19469 (make_invariant_chain): Initialization.
19470 (make_rooted_chain): Specify chain type in parameter and record it.
19471 (add_looparound_copies): Skip for store-stores chain.
19472 (determine_roots_comp): Compute type of chain and pass it to
19474 (initialize_root_vars_store_elim_2): New function.
19475 (finalize_eliminated_stores): New function.
19476 (remove_stmt): Handle store for elimination.
19477 (execute_pred_commoning_chain): Execute predictive commoning on
19478 store-store chains.
19479 (determine_unroll_factor): Skip unroll for store-stores chain.
19480 (prepare_initializers_chain_store_elim): New function.
19481 (prepare_initializers_chain): Hanlde store-store chain.
19482 (prepare_finalizers_chain, prepare_finalizers): New function.
19483 (tree_predictive_commoning_loop): Return integer value indicating
19484 if loop is unrolled or lcssa form is corrupted.
19485 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
19487 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19489 * tree-predcom.c (initialize_root): Delete.
19490 (execute_pred_commoning_chain): Initialize root vars and replace
19491 reference of non-combined chain directly, rather than call above
19494 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19496 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
19497 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
19499 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19501 * tree-predcom.c (struct chain): New field init_seq.
19502 (release_chain): Release init_seq.
19503 (prepare_initializers_chain): Record intialization stmts in above
19505 (insert_init_seqs): New function.
19506 (tree_predictive_commoning_loop): Call insert_init_seqs.
19508 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19510 * tree-predcom.c (determine_roots_comp): Skip trivial components.
19512 2017-07-28 Richard Biener <rguenther@suse.de>
19514 * match.pd: Remove superfluous :c.
19515 * genmatch.c (simplify::id): Add member.
19516 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
19518 (current_id): New global.
19519 (dt_node::parent): Move from ...
19520 (dt_operand::parent): ... here. Add for_id member.
19521 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
19522 (decision_tree::find_node): Relax order requirement when
19523 merging DT_TRUE nodes to ones inbetween the current simplify
19524 and the one we try to merge with. Add diagnostic whenever
19525 we need to enforce pattern order by not merging.
19526 (decision_tree::insert): Set current_id.
19527 (decision_tree::print_node): Dump parent node and for_id.
19528 (parser::last_id): Add member.
19529 (parser::push_simplify): Assign unique id.
19530 (parser::parser): Initialize last_id.
19532 2017-07-28 Martin Liska <mliska@suse.cz>
19535 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
19536 gimple_build_debug_bind.
19538 2017-07-28 Richard Biener <rguenther@suse.de>
19540 PR tree-optimization/81502
19541 * match.pd: Add pattern combining BIT_INSERT_EXPR with
19543 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
19545 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
19546 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
19547 for BIT_FIELD_REF args.
19548 * fold-const.c (make_bit_field_ref): Likewise.
19549 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
19551 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19554 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
19555 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
19556 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
19557 Or it into SANITIZER_UNDEFINED.
19558 * ubsan.c: Include gimple-fold.h and varasm.h.
19559 (ubsan_expand_ptr_ifn): New function.
19560 (instrument_pointer_overflow): New function.
19561 (maybe_instrument_pointer_overflow): New function.
19562 (instrument_object_size): Formatting fix.
19563 (pass_ubsan::execute): Call instrument_pointer_overflow
19564 and maybe_instrument_pointer_overflow.
19565 * internal-fn.c (expand_UBSAN_PTR): New function.
19566 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
19567 * sanitizer.def (__ubsan_handle_pointer_overflow,
19568 __ubsan_handle_pointer_overflow_abort): New builtins.
19569 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
19570 * internal-fn.def (UBSAN_PTR): New internal function.
19571 * opts.c (sanitizer_opts): Add pointer-overflow.
19572 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
19573 * fold-const.c (build_range_check): Compute pointer range check in
19574 integral type if pointer arithmetics would be needed. Formatting
19577 2017-07-28 Martin Liska <mliska@suse.cz>
19580 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
19581 parameters that are of a variable-length.
19583 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
19585 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
19587 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
19588 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
19589 (CRT_CALL_STATIC_FUNCTION): Likewise.
19590 (ASM_DEFAULT_SPEC): New define.
19591 (ASM_SPEC32): Likewise.
19592 (ASM_SPEC64): Likewise.
19593 (ASM_SPEC_COMMON): Likewise.
19594 (ASM_SPEC): Likewise.
19595 (INVALID_64BIT): Likewise.
19596 (LINK_OS_DEFAULT_SPEC): Likewise.
19597 (LINK_OS_SPEC32): Likewise.
19598 (LINK_OS_SPEC64): Likewise.
19599 (POWERPC_LINUX): Likewise.
19600 (PTRDIFF_TYPE): Likewise.
19601 (RESTORE_FP_PREFIX): Likewise.
19602 (RESTORE_FP_SUFFIX): Likewise.
19603 (SAVE_FP_PREFIX): Likewise.
19604 (SAVE_FP_SUFFIX): Likewise.
19605 (SIZE_TYPE): Likewise.
19606 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
19607 (TARGET_64BIT): Likewise.
19608 (TARGET_64BIT): Likewise.
19609 (TARGET_AIX): Likewise.
19610 (WCHAR_TYPE_SIZE): Likewise.
19611 (WCHAR_TYPE): Undefine.
19612 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
19613 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
19614 (CPP_OS_RTEMS_SPEC): Delete.
19615 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
19616 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
19618 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
19620 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19622 PR tree-optimization/81578
19623 * tree-parloops.c (build_new_reduction): Bail out if
19624 reduction_code isn't one of the standard OpenMP reductions.
19625 Move the details printing after that decision.
19627 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19629 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
19630 related to reload_in_progress.
19631 (splat_input_operand): Likewise.
19632 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
19634 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
19636 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
19637 (TARGET_INSTANTIATE_DECLS): Likewise.
19638 (legitimate_indexed_address_p): Delete reload_in_progress code.
19639 (rs6000_debug_legitimate_address_p): Likewise.
19640 (rs6000_eliminate_indexed_memrefs): Likewise.
19641 (rs6000_emit_le_vsx_store): Likewise.
19642 (rs6000_emit_move_si_sf_subreg): Likewise.
19643 (rs6000_emit_move): Likewise.
19644 (register_to_reg_type): Likewise.
19645 (rs6000_pre_atomic_barrier): Likewise.
19646 (rs6000_machopic_legitimize_pic_address): Likewise.
19647 (rs6000_allocate_stack_temp): Likewise.
19648 (rs6000_address_for_fpconvert): Likewise.
19649 (rs6000_address_for_altivec): Likewise.
19650 (rs6000_secondary_memory_needed_rtx): Delete function.
19651 (rs6000_check_sdmode): Likewise.
19652 (rs6000_alloc_sdmode_stack_slot): Likewise.
19653 (rs6000_instantiate_decls): Likewise.
19654 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
19655 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
19656 Delete reload_in_progress.
19657 (*vec_reload_and_plus_<mptrsize>): Likewise.
19658 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
19659 (vsx_div_v2di): Likewise.
19660 (vsx_udiv_v2di): Likewise.
19662 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19664 * config/rs6000/rs6000.opt (mlra): Replace with stub.
19665 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
19666 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
19667 (rs6000_debug_reg_global): Delete print of LRA status.
19668 (rs6000_option_override_internal): Delete dead LRA related code.
19669 (rs6000_lra_p): Delete function.
19670 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
19672 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
19674 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
19675 * config/riscv/rtems.h: New file.
19677 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19678 Sudakshina Das <sudi.das@arm.com>
19680 * config/aarch64/aarch64.md
19681 (define_split for and<mode>3nr_compare): Move
19682 non aarch64_logical_operand to a register.
19683 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
19684 register immediate operand to a register.
19685 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
19687 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19689 PR middle-end/81564
19690 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
19692 2017-07-27 Richard Biener <rguenther@suse.de>
19694 PR tree-optimization/81573
19695 PR tree-optimization/81494
19696 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
19697 multi defuse cycle case.
19699 2017-07-27 Richard Biener <rguenther@suse.de>
19701 PR tree-optimization/81571
19702 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
19705 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
19707 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
19708 earlier and only if MASK_FPU is set. Adjust formatting.
19710 2017-07-27 Martin Liska <mliska@suse.cz>
19712 * opt-functions.awk: Add validation of value of Init.
19713 * optc-gen.awk: Pass new argument.
19715 2017-07-27 Martin Liska <mliska@suse.cz>
19717 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
19718 Fix wrong condition.
19720 2017-07-27 Martin Liska <mliska@suse.cz>
19722 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
19723 BBs and edges seen by autoFDO.
19725 2017-07-27 Richard Biener <rguenther@suse.de>
19727 PR tree-optimization/81502
19728 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
19729 with incompatible but same sized type.
19730 (execute_update_addresses_taken): Likewise.
19732 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
19734 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
19735 flag_tree_loop_vectorize rather than flag_tree_vectorize.
19737 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19740 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
19741 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
19742 Change s_operand to memory_operand.
19744 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
19746 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
19747 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
19748 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
19749 Emit instructions rather than returning an expression. Handle TFmode
19750 and KFmode by casting to TImode.
19751 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
19752 (rs6000_emit_le_vsx_store): Likewise.
19753 * config/rs6000/vsx.md (VSX_TI): New iterator.
19754 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
19755 (*vsx_le_undo_permute_<mode>): Likewise.
19756 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
19757 emit the split sequence.
19758 (*vsx_le_perm_store_<mode>): Likewise.
19760 2017-07-27 Jakub Jelinek <jakub@redhat.com>
19762 PR tree-optimization/81555
19763 PR tree-optimization/81556
19764 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
19765 if true, force CHANGED for the recursive invocation.
19766 (reassociate_bb): Remember original length of ops array, pass
19767 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
19769 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
19770 attributes for noipa attribute. For naked attribute use
19771 lookup_attribute first before lookup_attribute_spec.
19772 * final.c (rest_of_handle_final): Disable IPA RA for functions with
19774 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
19775 for functions with noipa attribute.
19776 (cgraph_externally_visible_p): Return true for functions with noipa
19778 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
19779 for functions with noipa attribute.
19780 * doc/extend.texi: Document noipa function attribute.
19781 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
19782 also for functions with noipa attribute.
19783 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
19785 2017-07-26 Andrew Pinski <apinski@cavium.com>
19787 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
19788 vec_unalign_load_cost and vec_unalign_store_cost.
19790 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
19792 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
19793 -mvsx-small-integer option.
19794 (ISA_3_0_MASKS_IEEE): Likewise.
19795 (OTHER_VSX_VECTOR_MASKS): Likewise.
19796 (POWERPC_MASKS): Likewise.
19797 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
19798 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
19799 code, only testing for DImode being allowed in non-VSX floating
19801 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
19802 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
19804 (rs6000_option_override_internal): Delete -mvsx-small-integer.
19805 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
19806 TARGET_P8_VECTOR test.
19807 (rs6000_secondary_reload_simple_move): Likewise.
19808 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
19809 since TARGET_P9_VECTOR was already tested.
19810 (rs6000_opt_masks): Remove -mvsx-small-integer.
19811 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
19812 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19814 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
19815 test for TARGET_VEXTRACTUB was used, and that uses
19817 (p9 extract splitter): Likewise.
19818 (vsx_extract_<mode>_di_p9): Likewise.
19819 (vsx_extract_<mode>_store_p9): Likewise.
19820 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
19821 for TARGET_P9_VECTOR was used. Delete code that is now dead with
19822 the elimination of TARGET_VSX_SMALL_INTEGER.
19823 (vsx_extract_<mode>_p8): Likewise.
19824 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
19825 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
19826 (vsx_set_<mode>_p9): Likewise.
19827 (vsx_set_v4sf_p9): Likewise.
19828 (vsx_set_v4sf_p9_zero): Likewise.
19829 (vsx_insert_extract_v4sf_p9): Likewise.
19830 (vsx_insert_extract_v4sf_p9_2): Likewise.
19831 * config/rs6000/rs6000.md (sign extend splitter): Change
19832 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
19833 (floatsi<mode>2_lfiwax_mem): Likewise.
19834 (floatunssi<mode>2_lfiwzx_mem): Likewise.
19835 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
19836 since a test for TARGET_P9_VECTOR was used.
19837 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19838 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
19839 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19840 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
19841 TARGET_P8_VECTOR test.
19842 (fix_trunc<mode>si2_stfiwx): Likewise.
19843 (fix_trunc<mode>si2_internal): Likewise.
19844 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
19845 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19847 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19848 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
19849 TARGET_P8_VECTOR test.
19850 (fixuns_trunc<mode>si2_stfiwx): Likewise.
19851 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
19852 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19854 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19855 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
19856 since a test for TARGET_P9_VECTOR was used.
19857 (splitter for loading small constants): Likewise.
19859 2017-07-26 Andrew Pinski <apinski@cavium.com>
19861 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
19864 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
19867 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
19868 (fp_valid_at): Likewise.
19870 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
19872 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
19873 (qdf24xx_addrcost_table): Likewise.
19874 (cortexa57_tunings): Update to use generic_branch_cost.
19875 (cortexa72_tunings): Likewise.
19876 (cortexa73_tunings): Likewise.
19877 (qdf24xx_tunings): Likewise.
19879 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
19881 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
19882 (thunderx2t99_branch_cost): Likewise.
19883 (cortexa35_tunings): Update to use generic_branch_cost.
19884 (cortexa53_tunings): Likewise.
19885 (cortexa57_tunings): Likewise.
19886 (cortexa72_tunings): Likewise.
19887 (cortexa73_tunings): Likewise.
19888 (thunderx2t99_tunings): Likewise.
19890 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19892 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
19893 (sparc_option_override): Honour MASK_FSMULD.
19894 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
19895 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
19896 * config/sparc/sparc.opt (mfsmuld): New option.
19897 * doc/invoke.texi (mfsmuld): Document option.
19899 2017-07-26 Marek Polacek <polacek@redhat.com>
19901 PR middle-end/70992
19902 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
19904 2017-07-26 Richard Biener <rguenther@suse.de>
19906 * gimple-match-head.c (do_valueize): Return OP if valueize
19908 (get_def): New helper to get at the def stmt of a SSA name
19909 if valueize allows.
19910 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
19911 do_valueize to get at the def stmt.
19912 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
19914 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
19916 PR middle-end/46932
19917 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
19919 2017-07-26 Martin Liska <mliska@suse.cz>
19922 * function.c (expand_function_start): Make expansion of
19923 nonlocal_goto_save_area after parm_birth_insn.
19925 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19927 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
19928 from all CPU target flags enable members.
19930 2017-07-26 Richard Biener <rguenther@suse.de>
19932 * genmatch.c (dt_simplify::gen): Make iterator vars const.
19933 (decision_tree::gen): Make 'type' const.
19934 (write_predicate): Likewise.
19936 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
19938 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
19939 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
19940 (rs6000_option_override_internal): Likewise.
19941 (rs6000_expand_vector_set): Likewise.
19942 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
19943 (TARGET_UPPER_REGS_SF): Likewise.
19944 (TARGET_UPPER_REGS_DI): Likewise.
19945 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
19946 (TARGET_DIRECT_MOVE_64BIT): Likewise.
19947 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
19948 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19949 (Splitters for DI constants in Altivec registers): Likewise.
19950 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
19951 (vsx_set_v4sf_p9): Likewise.
19952 (vsx_set_v4sf_p9_zero): Likewise.
19953 (vsx_insert_extract_v4sf_p9): Likewise.
19954 (vsx_insert_extract_v4sf_p9_2): Likewise.
19956 2017-07-25 Carl Love <cel@us.ibm.com>
19958 * doc/extend.texi: Update the built-in documentation file for the
19959 existing built-in functions
19960 vector signed char vec_cnttz (vector signed char);
19961 vector unsigned char vec_cnttz (vector unsigned char);
19962 vector signed short vec_cnttz (vector signed short);
19963 vector unsigned short vec_cnttz (vector unsigned short);
19964 vector signed int vec_cnttz (vector signed int);
19965 vector unsigned int vec_cnttz (vector unsigned int);
19966 vector signed long long vec_cnttz (vector signed long long);
19967 vector unsigned long long vec_cnttz (vector unsigned long long);
19969 2017-07-25 Andrew Pinski <apinski@cavium.com>
19971 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
19972 accesses where the use is for the first operand of a BIT_INSERT.
19974 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19977 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
19978 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
19980 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19982 * config/i386/gstabs.h: Delete.
19983 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
19985 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
19987 * config/i386/i386.c (ix86_decompose_address): Do not check for
19988 register RTX when looking at index_reg or base_reg.
19989 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
19991 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
19993 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
19994 to update EH info here.
19996 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19998 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
20000 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20002 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
20004 2017-07-25 Torsten Duwe <duwe@suse.de>
20006 * common.opt: Introduce -fpatchable-function-entry
20007 command line option, and its variables function_entry_patch_area_size
20008 and function_entry_patch_area_start.
20009 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
20010 including a two-value parser.
20011 * target.def (print_patchable_function_entry): New target hook.
20012 * targhooks.h (default_print_patchable_function_entry): New function.
20013 * targhooks.c (default_print_patchable_function_entry): Likewise.
20014 * toplev.c (process_options): Switch off IPA-RA if
20015 patchable function entries are being generated.
20016 * varasm.c (assemble_start_function): Look at the
20017 patchable-function-entry command line switch and current
20018 function attributes and maybe generate NOP instructions by
20019 calling the print_patchable_function_entry hook.
20020 * doc/extend.texi: Document patchable_function_entry attribute.
20021 * doc/invoke.texi: Document -fpatchable_function_entry
20022 command line option.
20023 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
20025 * doc/tm.texi: Re-generate.
20027 2017-07-25 Jakub Jelinek <jakub@redhat.com>
20030 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
20031 TARGET_AVX512DQ rather than TARGET_AVX512BW.
20033 2017-07-25 Tamar Christina <tamar.christina@arm.com>
20035 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
20037 2017-07-25 Richard Biener <rguenther@suse.de>
20039 PR tree-optimization/81455
20040 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
20041 not walk in cycles when looking for guards.
20043 2017-07-25 Richard Biener <rguenther@suse.de>
20045 PR tree-optimization/81529
20046 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
20047 when optimizing backedge uses.
20049 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
20051 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
20053 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
20054 to dl_section_ref. On AIX, append an expression to subtract
20055 the size of the section length to dl_section_ref.
20057 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
20059 * configure.ac: If any of the config.* scripts fail, exit 1.
20060 * configure: Regenerate.
20062 2017-07-25 Richard Biener <rguenther@suse.de>
20064 PR middle-end/81546
20065 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
20066 of immediate uses, be more verbose on errors.
20068 2017-07-25 Richard Biener <rguenther@suse.de>
20070 PR tree-optimization/81510
20071 * tree-vect-loop.c (vect_is_simple_reduction): When the
20072 reduction stmt is not inside the loop bail out.
20074 2017-07-25 Richard Biener <rguenther@suse.de>
20076 PR tree-optimization/81303
20077 * tree-vect-loop-manip.c (vect_loop_versioning): Build
20078 profitability check against LOOP_VINFO_NITERSM1.
20080 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20082 * domwalk.c (cmp_bb_postorder): Simplify.
20083 (sort_bbs_postorder): New function. Use it...
20084 (dom_walker::walk): ...here to optimize common cases.
20086 2017-07-25 Martin Liska <mliska@suse.cz>
20089 * ipa-visibility.c (function_and_variable_visibility): Make the
20090 redirection just on target that supports aliasing.
20091 Fix GNU coding style.
20093 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
20096 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
20098 * config/arm/rtems.h: Likewise.
20099 * config/bfin/rtems.h: Likewise.
20100 * config/i386/rtemself.h: Likewise.
20101 * config/lm32/rtems.h: Likewise.
20102 * config/m32c/rtems.h: Likewise.
20103 * config/m68k/rtemself.h: Likewise.
20104 * config/microblaze/rtems.h: Likewise.
20105 * config/mips/rtems.h: Likewise.
20106 * config/moxie/rtems.h: Likewise.
20107 * config/nios2/rtems.h: Likewise.
20108 * config/powerpcspe/rtems.h: Likewise.
20109 * config/rs6000/rtems.h: Likewise.
20110 * config/rtems.h: Likewise.
20111 * config/sh/rtems.h: Likewise.
20112 * config/sh/rtemself.h: Likewise.
20113 * config/sparc/rtemself.h: Likewise.
20115 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20118 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
20119 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
20120 * tree-ssa-structalias.c (alias_get_name): Same.
20122 2017-07-25 Bin Cheng <bin.cheng@arm.com>
20125 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
20126 instructions if no du chain is found.
20128 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20130 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
20132 2017-07-25 Richard Biener <rguenther@suse.de>
20134 PR middle-end/81505
20135 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
20138 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
20140 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
20141 upper-regs options.
20142 (ISA_2_7_MASKS_SERVER): Likewise.
20143 (ISA_3_0_MASKS_IEEE): Likewise.
20144 (OTHER_P8_VECTOR_MASKS): Likewise.
20145 (OTHER_VSX_VECTOR_MASKS): Likewise.
20146 (POWERPC_MASKS): Likewise.
20147 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
20148 duplicate list of options.
20149 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
20150 explicit -mupper-regs options.
20151 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
20152 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
20153 alias for -mupper-regs-df.
20154 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20155 (rs6000_init_hard_regno_mode_ok): Likewise.
20156 (rs6000_option_override_internal): Likewise.
20157 (rs6000_opt_masks): Likewise.
20158 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
20159 options in terms of whether -mvsx or -mpower8-vector was used.
20160 (TARGET_UPPER_REGS_DI): Likewise.
20161 (TARGET_UPPER_REGS_SF): Likewise.
20162 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
20163 -mupper-regs-* options.
20165 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
20167 * passes.c (emergency_dump_function): Print some empty lines and a
20168 header before the RTL dump.
20170 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
20172 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
20174 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
20177 * config/aarch64/aarch64.c (aarch64_classify_symbol):
20178 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
20180 2017-07-24 Carl Love <cel@us.ibm.com>
20182 * config/rs6000/rs6000-c.c: Add support for built-in functions
20183 vector float vec_extract_fp32_from_shorth (vector unsigned short);
20184 vector float vec_extract_fp32_from_shortl (vector unsigned short);
20185 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
20186 vec_extract_fp_from_shortl): Add defines for the two builtins.
20187 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
20188 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
20190 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
20191 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
20192 * doc/extend.texi: Update the built-in documentation file for the
20193 new built-in function.
20195 2017-07-24 Jakub Jelinek <jakub@redhat.com>
20198 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
20200 * doc/generic.texi: Likewise.
20201 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
20202 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
20204 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
20206 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
20207 (aarch64_mls_elt_merge<mode>): Likewise.
20209 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
20211 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
20212 having __cxa_atexit.
20214 2017-07-23 Michael Collison <michael.collison@arm.com>
20216 * config/arm/arm.c (arm_option_override): Deprecate
20217 use of -mstructure-size-boundary.
20218 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
20219 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
20221 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20224 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
20225 Reduce cost estimate for direct moves.
20227 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
20230 * config/i386/i386.c (ix86_option_override_internal): Disable
20231 BMI, BMI2 and TBM instructions for -m16.
20233 2017-07-21 Carl Love <cel@us.ibm.com>
20235 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20236 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20237 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20238 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20239 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20240 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20241 VMULOSW): New enum "unspec" values.
20242 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20243 altivec_vmulosw): New patterns.
20244 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20245 VMULOSW): Add definitions.
20247 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
20249 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
20250 (qdf24xx): Likewise.
20251 * config/aarch64/aarch64-options-extensions.def (rdma); New.
20252 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
20253 (AARCH64_FL_V8_1): Renumber.
20254 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
20255 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
20256 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
20257 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
20258 rdma to feature modifiers list.
20260 2017-07-21 Yury Gribov <tetra2005@gmail.com>
20262 PR middle-end/56727
20263 * ipa-visibility (function_and_variable_visibility): Convert
20264 recursive PLT call to direct call if appropriate.
20266 2017-07-21 Andrew Pinski <apinski@cavium.com>
20268 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
20269 operand 1 to see if the types precision matches.
20270 * fold-const.c (operand_equal_p): Likewise.
20272 2017-07-21 Richard Biener <rguenther@suse.de>
20274 PR tree-optimization/81303
20275 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
20276 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
20277 (vect_peeling_hash_get_lowest_cost): Adjust.
20278 (vect_enhance_data_refs_alignment): Likewise. Use
20279 vect_get_peeling_costs_all_drs to compute the penalty for no
20280 peeling to match up costs.
20282 2017-07-21 Richard Biener <rguenther@suse.de>
20284 PR tree-optimization/81500
20285 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
20286 we didn't identify a reduction path.
20288 2017-07-21 Tom de Vries <tom@codesourcery.com>
20289 Cesar Philippidis <cesar@codesourcery.com>
20291 PR gcov-profile/81442
20292 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
20295 2017-07-21 Tom de Vries <tom@codesourcery.com>
20298 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
20300 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
20301 nvptx_override_options_after_change.
20303 2017-07-21 Ulrich Drepper <drepper@redhat.com>
20305 * dwarf2out.c (output_file_names): Avoid double testing for
20306 dwarf_version >= 5.
20308 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
20310 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
20312 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20314 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
20316 (try_crossjump_to_edge): Do not punt on partitioned functions.
20318 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20320 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
20321 Put all BBs reachable only via paths crossing cold region to cold
20323 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
20325 2016-07-21 Richard Biener <rguenther@suse.de>
20327 PR tree-optimization/81303
20328 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
20329 into account prologue and epilogue iterations when raising
20330 min_profitable_iters to sth at least covering one vector iteration.
20332 2017-07-21 Tamar Christina <tamar.christina@arm.com>
20334 * config/arm/arm.c (arm_test_cpu_arch_dat):
20337 2017-07-20 Nathan Sidwell <nathan@acm.org>
20339 Remove TYPE_METHODS.
20340 * tree.h (TYPE_METHODS): Delete.
20341 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
20342 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
20343 (dbxout_type_methods): Scan TYPE_FIELDS.
20344 (dbxout_type): Don't check TYPE_METHODS here.
20345 * function.c (use_register_for_decl): Always ignore register for
20346 class types when not optimizing.
20347 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
20348 * tree.c (free_lang_data_in_type): Stitch out member functions and
20349 templates from TYPE_FIELDS.
20350 (build_distinct_type_copy, verify_type_variant,
20351 verify_type): Member fns are on TYPE_FIELDS.
20352 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
20353 * tree-pretty-print.c (dump_generic_node): Likewise.
20355 2017-07-20 Jakub Jelinek <jakub@redhat.com>
20358 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
20359 V2TImode and V4TImode.
20360 (ix86_expand_vector_extract): Likewise.
20361 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
20362 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
20363 (ssescalarmode): Handle V4TImode and V2TImode.
20364 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
20365 (*vec_extractv2ti, *vec_extractv4ti): New insns.
20366 (VEXTRACTI128_MODE): New mode iterator.
20367 (splitter for *vec_extractv?ti first element): New.
20368 (VEC_INIT_MODE): New mode iterator.
20369 (vec_init<mode>): Consolidate 3 expanders into one using
20370 VEC_INIT_MODE mode iterator.
20372 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20374 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
20375 non_spilled_static_chain_regno_p.
20377 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20379 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
20381 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
20383 * bb-reorder.c (connect_traces): Allow copying of blocks within
20386 2017-07-20 Richard Biener <rguenther@suse.de>
20388 * gimple.h (gimple_phi_result): Add gphi * overload.
20389 (gimple_phi_result_ptr): Likewise.
20390 (gimple_phi_arg): Likewise. Adjust index assert to only
20391 allow actual argument accesses rather than all slots available
20393 (gimple_phi_arg_def): Add gphi * overload.
20394 * tree-phinodes.c (make_phi_node): Initialize only actual
20396 (resize_phi_node): Clear memory not covered by old node,
20397 do not initialize excess argument slots.
20398 (reserve_phi_args_for_new_edge): Initialize new argument slot
20401 2017-07-20 Bin Cheng <bin.cheng@arm.com>
20403 PR tree-optimization/81388
20405 2016-07-21 Bin Cheng <bin.cheng@arm.com>
20407 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
20408 by removing computation of may_be_zero.
20410 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20411 Tom de Vries <tom@codesourcery.com>
20413 PR middle-end/81030
20414 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
20415 when gimple level profile disagrees with what RTL expander did.
20417 2017-07-20 Richard Biener <rguenther@suse.de>
20419 PR tree-optimization/61171
20420 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
20421 (vect_analyze_stmt): Add slp instance parameter.
20422 (vectorizable_reduction): Likewise.
20423 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
20424 (vect_is_simple_reduction): Deal with chains not detected
20425 as SLP reduction chain, specifically not properly associated
20426 chains containing a mix of plus/minus.
20427 (get_reduction_op): Remove.
20428 (get_initial_defs_for_reduction): Simplify, pass in whether
20429 this is a reduction chain, pass in the SLP node for the PHIs.
20430 (vect_create_epilog_for_reduction): Get the SLP instance as
20432 (vectorizable_reduction): Get the SLP instance as arg.
20433 During analysis remember the SLP node with the PHIs in the
20434 instance. Simplify getting at the vectorized reduction PHIs.
20435 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
20436 through SLP instance.
20437 (vect_slp_analyze_operations): Likewise.
20438 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
20439 (vect_transform_stmt): Likewise.
20441 2017-07-20 Tom de Vries <tom@codesourcery.com>
20443 PR tree-optimization/81489
20444 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
20445 read of phi arg location to before loop that modifies phi.
20447 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20449 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
20452 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20454 PR middle-end/81331
20455 * except.c (execute): Fix ordering issue.
20457 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20459 PR rtl-optimization/81423
20460 * combine.c (make_compound_operation_int): Don't try to optimize
20461 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
20463 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20465 PR rtl-optimization/81423
20466 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
20467 with a constant that is -1 in the truncated to mode.
20469 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20471 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
20472 (determine_unlikely_bbs): ... here.
20473 * predict.h (propagate_unlikely_bbs_forward): Declare.
20474 * cfgexpand.c (pass_expand::execute): Use it.
20475 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
20477 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
20478 propagate_unlikely_bbs_forward.
20480 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20482 PR middle-end/81331
20483 * except.c (maybe_add_nop_after_section_switch): New function.
20486 2017-07-19 Tom de Vries <tom@codesourcery.com>
20488 * gimple.h (gimple_phi_set_arg): Make assert more strict.
20490 2017-07-19 Tom de Vries <tom@codesourcery.com>
20492 * gimple.h (gimple_phi_arg): Make assert more strict.
20494 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
20496 * config.gcc (powerpc*-*-*): Add mmintrin.h.
20497 * config/rs6000/mmintrin.h: New file.
20498 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
20500 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20502 PR tree-optimization/81346
20503 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
20505 2017-07-19 Tom de Vries <tom@codesourcery.com>
20507 * config/nvptx/nvptx.md (VECIM): Add V2DI.
20509 2017-07-19 Tom de Vries <tom@codesourcery.com>
20511 * config/nvptx/nvptx-modes.def: Add V2DImode.
20512 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
20513 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
20514 (nvptx_output_mov_insn): Handle lack of mov.b128.
20515 (nvptx_print_operand): Handle 'H' and 'L' codes.
20516 (nvptx_vector_mode_supported): Allow V2DImode.
20517 (nvptx_preferred_simd_mode): New function.
20518 (nvptx_data_alignment): New function.
20519 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
20520 nvptx_preferred_simd_mode.
20521 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
20523 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
20525 2017-07-19 Tom de Vries <tom@codesourcery.com>
20527 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
20528 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
20529 (nvptx_vector_mode_supported): New function. Allow V2SImode.
20530 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
20531 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
20532 (mov<VECIM>_insn): New define_insn.
20533 (define_expand "mov<VECIM>): New define_expand.
20535 2017-07-19 Tom de Vries <tom@codesourcery.com>
20537 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
20539 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20541 PR tree-optimization/81346
20542 * fold-const.h (fold_div_compare, range_check_type): Declare.
20543 * fold-const.c (range_check_type): New function.
20544 (build_range_check): Use range_check_type.
20545 (fold_div_compare): No longer static, rewritten into
20546 a match.pd helper function.
20547 (fold_comparison): Don't call fold_div_compare here.
20548 * match.pd (X / C1 op C2): New optimization using fold_div_compare
20549 as helper function.
20551 2017-07-19 Nathan Sidwell <nathan@acm.org>
20553 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
20554 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
20555 * tree.c (find_decls_types_r, verify_type): Use
20556 TYPE_{MIN,MAX}_VALUE_RAW.
20557 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
20558 (hash_tree): Likewise.
20559 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20561 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20564 2017-07-18 Tom de Vries <tom@codesourcery.com>
20566 PR middle-end/81464
20567 * omp-expand.c (expand_omp_for_static_chunk): Handle
20568 equal-argument loop exit phi.
20570 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
20573 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
20574 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
20575 operand 2 predicate.
20576 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
20577 operand 2 predicate.
20578 (ror,rol -> rorx splitters): Use const_int_operand as
20579 operand 2 predicate.
20581 2017-06-18 Richard Biener <rguenther@suse.de>
20583 PR tree-optimization/81410
20584 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
20585 the gap in the ! slp_perm SLP case after each group.
20587 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20589 PR middle-end/81463
20590 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
20593 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20595 PR middle-end/81462
20596 * predict.c (set_even_probabilities): Cleanup; do not affect
20597 probabilities that are already known.
20598 (combine_predictions_for_bb): Call even when count is set.
20600 2017-07-18 Nathan Sidwell <nathan@acm.org>
20602 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
20605 2017-07-18 Bin Cheng <bin.cheng@arm.com>
20608 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
20609 optimization for loop niter analysis.
20611 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
20614 * config/avr/avr.c (avr_optimize_casesi): Don't use
20615 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
20617 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
20619 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
20620 body_cost_vec from _vect_peel_extended_info.
20621 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
20622 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
20625 2017-07-18 Bin Cheng <bin.cheng@arm.com>
20627 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
20629 2017-07-18 Richard Biener <rguenther@suse.de>
20631 PR tree-optimization/80620
20632 PR tree-optimization/81403
20633 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
20634 info when re-using a VN table entry.
20636 2017-07-18 Richard Biener <rguenther@suse.de>
20638 PR tree-optimization/81418
20639 * tree-vect-loop.c (vectorizable_reduction): Properly compute
20640 vectype_in. Verify that with lane-reducing reduction operations
20641 we have a single def-use cycle.
20643 2017-07-17 Carl Love <cel@us.ibm.com>
20645 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
20647 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20648 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20649 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20650 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20651 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20652 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20653 VMULOSW): New enum "unspec" values.
20654 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20655 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20656 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20657 altivec_vmulosw): New patterns.
20658 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20659 VMULOSW): Add definitions.
20661 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
20663 * config/alpha/alpha.c: Include predict.h.
20665 2017-07-17 Yury Gribov <tetra2005@gmail.com>
20667 * tree-vrp.c (compare_assert_loc): Fix comparison function
20668 to return predictable results.
20670 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20672 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
20674 (subdi3): Likewise.
20675 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
20676 * doc/invoke.texi (mexpand-adddi): Update text.
20678 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20680 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
20681 that also clobbers the CC register. The old expand code is moved
20683 (*arc_clzsi2): ... here.
20684 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
20685 the CC register. The old expand code is moved to ...
20686 (arc_ctzsi2): ... here.
20688 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20690 * config/arc/arc.opt (mindexed-loads): Use initial value
20691 TARGET_INDEXED_LOADS_DEFAULT.
20692 (mauto-modify-reg): Use initial value
20693 TARGET_AUTO_MODIFY_REG_DEFAULT.
20694 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
20695 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
20696 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
20697 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
20699 2017-07-17 Martin Liska <mliska@suse.cz>
20702 * opts.c (finish_options): Do not allow -fgnu-tm
20703 w/ -fsanitize={kernel-,}address. Say sorry.
20705 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20708 * tree-loop-distribution.c (classify_partition): Only assert on
20709 numer of iterations.
20710 (merge_dep_scc_partitions): Delete prameter. Update function call.
20711 (distribute_loop): Remove code handling loop with unknown niters.
20712 (pass_loop_distribution::execute): Skip loop with unknown niters.
20714 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20717 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
20718 function sort_partitions_by_post_order.
20720 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20722 PR tree-optimization/81374
20723 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
20724 the max index of basic blocks, rather than number of basic blocks.
20726 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20728 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
20730 (arc_legitimate_pic_operand_p): Likewise.
20731 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
20733 (arc_needs_pcl_p): Likewise.
20734 (arc_legitimate_pc_offset_p): Likewise.
20735 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
20736 function is also used in constrains.md.
20737 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
20738 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
20739 PLUS. Only return true/false in known cases, otherwise assert.
20740 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
20741 is already called in arc_legitimate_constant_p.
20742 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
20744 (LEGITIMATE_PIC_OPERAND_P): Use
20745 arc_raw_symbolic_reference_mentioned_p function.
20746 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
20751 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20752 Andrew Burgess <andrew.burgess@embecosm.com>
20754 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
20755 (arc_return_address_register): New function.
20756 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
20757 (arc_handle_fndecl_attribute): Add naked attribute.
20758 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
20759 (TARGET_WARN_FUNC_RETURN): Likewise.
20760 (arc_allocate_stack_slots_for_args): New function.
20761 (arc_warn_func_return): Likewise.
20762 (machine_function): Change type fn_type.
20763 (arc_compute_function_type): Consider new naked function type,
20764 change function return type.
20765 (arc_must_save_register): Adapt to handle new
20766 arc_compute_function_type's return type.
20767 (arc_expand_prologue): Likewise.
20768 (arc_expand_epilogue): Likewise.
20769 (arc_return_address_regs): Delete.
20770 (arc_return_address_register): New function.
20771 (arc_epilogue_uses): Use above function.
20772 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
20773 (arc_function_type): Change encoding, add naked type.
20774 (ARC_INTERRUPT_P): Change to handle the new encoding.
20775 (ARC_FAST_INTERRUPT_P): Likewise.
20776 (ARC_NORMAL_P): Define.
20777 (ARC_NAKED_P): Likewise.
20778 (arc_compute_function_type): Delete prototype.
20779 * config/arc/arc.md (in_ret_delay_slot): Use
20780 arc_return_address_register function.
20781 (simple_return): Likewise.
20782 (p_return_i): Likewise.
20784 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20786 PR tree-optimization/81428
20787 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
20788 can't be built for those types.
20790 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
20792 Remove stuff dead since r239246.
20794 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
20795 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
20796 (avr_inform_devices): Remove dead stuff.
20798 2017-07-17 Tamar Christina <tamar.christina@arm.com>
20800 * config/arm/arm_neon.h: Fix softp typo.
20802 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20804 PR tree-optimization/81365
20805 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
20806 aggregate moves onto bb predecessor edges, make sure there are no
20807 loads that could alias the lhs in between the start of bb and the
20810 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
20813 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
20814 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
20815 [LSHIFTRT, outer_code = TRUNCATE]: Same.
20817 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20819 PR tree-optimization/81396
20820 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
20821 (init_symbolic_number): Initialize it to 1.
20822 (perform_symbolic_merge): Add n_ops from both operands into the new
20824 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
20825 without base_addr as useless if they need more than one operation.
20826 (bswap_replace): Handle !bswap case for NULL base_addr.
20828 2017-07-17 Tom de Vries <tom@codesourcery.com>
20831 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
20834 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
20836 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
20837 conditional builtin define __FIX_LEON3FT_B2BST.
20839 2017-07-17 Daniel Cederman <cederman@gaisler.com>
20841 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
20842 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
20845 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20847 PR rtl-optimization/81424
20848 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
20849 to remove potential trapping from operands if -fnon-call-exceptions.
20851 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20853 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
20854 profile_proability for scalling.
20855 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
20857 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20859 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
20861 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20863 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
20864 fixpoint arithmetics.
20866 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20868 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
20869 fixpoint arithmetics.
20871 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20873 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
20874 fixpoint arithmetics.
20876 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20878 * profile-count.h (profile_probability::from_reg_br_prob_note,
20879 profile_probability::to_reg_br_prob_note): New functions.
20880 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
20881 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
20882 * predict.c (probability_reliable_p): Update.
20883 (edge_probability_reliable_p): Update.
20884 (br_prob_note_reliable_p): Update.
20885 (invert_br_probabilities): Update.
20886 (add_reg_br_prob_note): New function.
20887 (combine_predictions_for_insn): Update.
20888 * asan.c (asan_clear_shadow): Update.
20889 * cfgbuild.c (compute_outgoing_frequencies): Update.
20890 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
20891 (update_br_prob_note): Update.
20892 (rtl_verify_edges): Update.
20893 (purge_dead_edges): Update.
20894 (fixup_reorder_chain): Update.
20895 * emit-rtl.c (try_split): Update.
20896 * ifcvt.c (cond_exec_process_insns): Update.
20897 (cond_exec_process_if_block): Update.
20898 (dead_or_predicable): Update.
20899 * internal-fn.c (expand_addsub_overflow): Update.
20900 (expand_neg_overflow): Update.
20901 (expand_mul_overflow): Update.
20902 * loop-doloop.c (doloop_modify): Update.
20903 * loop-unroll.c (compare_and_jump_seq): Update.
20904 * optabs.c (emit_cmp_and_jump_insn_1): Update.
20905 * predict.h: Update.
20906 * reorg.c (mostly_true_jump): Update.
20908 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
20909 * config/alpha/alpha.c (emit_unlikely_jump): Update.
20910 * config/arc/arc.c: (emit_unlikely_jump): Update.
20911 * config/arm/arm.c: (emit_unlikely_jump): Update.
20912 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
20913 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
20914 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
20915 (ix86_print_operand): Update.
20916 (ix86_split_fp_branch): Update.
20917 (predict_jump): Update.
20918 * config/ia64/ia64.c (ia64_print_operand): Update.
20919 * config/mmix/mmix.c (mmix_print_operand): Update.
20920 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
20921 (rs6000_expand_split_stack_prologue): Update.
20922 * config/rs6000/rs6000.c: Update.
20923 * config/s390/s390.c (s390_expand_vec_strlen): Update.
20924 (s390_expand_vec_movstr): Update.
20925 (s390_expand_cs_tdsi): Update.
20926 (s390_expand_split_stack_prologue): Update.
20927 * config/sh/sh.c (sh_print_operand): Update.
20928 (expand_cbranchsi4): Update.
20929 (expand_cbranchdi4): Update.
20930 * config/sparc/sparc.c (output_v9branch): Update.
20931 * config/spu/spu.c (get_branch_target): Update.
20932 (ea_load_store_inline): Update.
20933 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
20934 * config/tilepro/tilepro.c: Update.
20936 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20938 * gimplify.c (mostly_copy_tree_r): Revert latest change.
20939 (gimplify_save_expr): Likewise.
20941 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20943 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
20945 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20947 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
20949 * timevar.def (TV_IPA_FNSUMMARY): Define.
20951 2017-07-16 Daniel Cederman <cederman@gaisler.com>
20953 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
20954 to back store errata sensitive sequence from being generated.
20955 (sqrtdf2_fix): Likewise.
20957 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20959 * tree-ssa-threadupdate.c (compute_path_counts,
20960 update_joiner_offpath_counts): Use profile_probability.
20962 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
20965 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20967 * config/arm/arm-c.c (arm_cpu_builtins): Define
20968 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20970 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
20972 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20973 array entries to represent __ieee128 versions of the
20974 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
20975 scalar_extract_sig, and scalar_insert_exp built-in functions.
20976 (altivec_resolve_overloaded_builtin): Add special case handling
20977 for the __builtin_scalar_insert_exp function, as represented by
20978 the P9V_BUILTIN_VEC_VSIEDP constant.
20979 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
20980 exponent support for __ieee128 argument.
20981 (VSESQP): Add scalar extract signature support for __ieee128
20983 (VSTDCNQP): Add scalar test negative support for __ieee128
20985 (VSIEQP): Add scalar insert exponent support for __int128 argument
20986 with __ieee128 result.
20987 (VSIEQPF): Add scalar insert exponent support for __ieee128
20988 argument with __ieee128 result.
20989 (VSTDCQP): Add scalar test data class support for __ieee128
20991 (VSTDCNQP): Add overload support for scalar test negative with
20992 __ieee128 argument.
20993 (VSTDCQP): Add overload support for scalar test data class
20994 __ieee128 argument.
20995 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
20996 UNSPEC_VSX_SXSIGDP.
20997 (UNSPEC_VSX_SIEXPQP): New constant.
20998 (xsxexpqp): New insn for VSX scalar extract exponent quad
21000 (xsxsigqp): New insn for VSX scalar extract significand quad
21002 (xsiexpqpf): New insn for VSX scalar insert exponent quad
21003 precision with floating point argument.
21004 (xststdcqp): New expand for VSX scalar test data class quad
21006 (xststdcnegqp): New expand for VSX scalar test negative quad
21008 (xststdcqp): New insn to match expansions for VSX scalar test data
21009 class quad precision and VSX scalar test negative quad precision.
21010 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
21011 special case operand checking to enforce that second operand of
21012 VSX scalar test data class with quad precision argument is a 7-bit
21014 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21015 prototypes and descriptions of __ieee128 versions of
21016 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
21017 scalar_test_data_class, and scalar_test_neg built-in functions.
21019 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21021 PR tree-optimization/81162
21022 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
21023 replace a negate with an add.
21025 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
21027 * doc/invoke.texi (arm/-mcpu): Document +crypto.
21029 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21031 * config/arm/arm-c.c (arm_cpu_builtins): Define
21032 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
21034 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21036 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
21037 (armv8-r): Set ARM Cortex-R52 as default CPU.
21038 * config/arm/arm-tables.opt: Regenerate.
21039 * config/arm/arm-tune.md: Regenerate.
21040 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
21042 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
21043 extension for -mcpu=cortex-r52.
21045 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21047 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
21048 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
21049 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
21050 (fp-armv8): Define it as FP_ARMv8 only.
21051 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
21052 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
21054 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
21055 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
21056 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
21057 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
21058 than TARGET_FPU_ARMV8.
21059 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
21060 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
21061 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
21063 * config/arm/neon.md (neon_vrint): Likewise.
21064 (neon_vcvt): Likewise.
21065 (neon_<fmaxmin_op><mode>): Likewise.
21066 (<fmaxmin><mode>3): Likewise.
21067 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
21068 * config/arm/predicates.md (arm_cond_move_operator): Check against
21069 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
21071 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
21073 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
21074 to top of function.
21076 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21078 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
21079 loop in comment with memset.
21081 2017-07-14 Martin Liska <mliska@suse.cz>
21083 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
21084 * dwarf2out.c (is_java): Remove the function.
21085 (output_pubname): Remove usage of the function.
21086 (lower_bound_default): Remove usage of DW_LANG_Java.
21087 (gen_compile_unit_die): Likewise.
21088 * gcc.c: Remove compiler defaults for .java and .zip files.
21089 * gimple-expr.c (remove_suffix): Change as there's no longer
21090 extension than 4-letter one.
21091 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
21092 (gimplify_save_expr): Likewise.
21093 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
21094 as it's possible even for other languages than Java.
21095 * langhooks.h (struct lang_hooks): Remove Java from a comment.
21096 * lto-opts.c (lto_write_options): Remove reference to Java.
21097 * opts.c (strip_off_ending): Update file extension handling.
21098 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
21099 * tree-eh.c (lower_resx): Likewise.
21100 * tree.c (free_lang_data_in_type): Remove dead code.
21101 (find_decls_types_r): Likewise.
21102 (build_common_builtin_nodes): Remove Java from a comment.
21103 (verify_type): Remove dead code.
21104 * varasm.c (assemble_external): Remove Java from a comment.
21106 2017-07-14 Martin Liska <mliska@suse.cz>
21108 * opts.c (finish_options): Add quotes.
21109 (common_handle_option): Likewise.
21111 2017-07-14 Martin Liska <mliska@suse.cz>
21113 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
21114 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
21115 Remove N_SO_PASCAL.
21116 * dwarf2out.c (lower_bound_default): Do not handle
21118 (gen_compile_unit_die): Likewise.
21119 * gcc.c: Remove default extension binding for GNU Pascal.
21120 * stmt.c: Remove Pascal language from a comment.
21121 * xcoffout.c: Likewise.
21123 2017-07-13 David Malcolm <dmalcolm@redhat.com>
21126 * diagnostic-show-locus.c (fixit_cmp): New function.
21127 (layout::layout): Sort m_fixit_hints.
21128 (column_range::column_range): Assert that the values are valid.
21129 (struct char_span): New struct.
21130 (correction::overwrite): New method.
21131 (struct source_line): New struct.
21132 (line_corrections::add_hint): Add assertions. Reimplement memcpy
21133 calls in terms of classes source_line and char_span, and
21134 correction::overwrite.
21135 (selftest::test_overlapped_fixit_printing_2): New function.
21136 (selftest::diagnostic_show_locus_c_tests): Call it.
21138 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
21140 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
21141 early if there is no lhs.
21143 2017-07-13 Martin Liska <mliska@suse.cz>
21145 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
21146 (gen_reference_type_die): Likewise.
21147 * stor-layout.c: Remove Pascal-related comment.
21149 2017-07-13 Martin Liska <mliska@suse.cz>
21151 * opts.c (finish_options): Add quotes to error messages.
21152 (parse_sanitizer_options): Likewise.
21154 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
21156 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
21158 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
21160 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
21162 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
21164 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
21166 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
21168 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
21171 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
21172 provides the hardware capability bits, define the macro
21173 __BUILTIN_CPU_SUPPORTS__.
21174 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
21175 if GLIBC does not provide the hardware capability bits. Add a
21176 gcc_unreachable call if the built-in cpu function is neither
21177 __builtin_cpu_is nor __builtin_cpu_supports.
21178 (rs6000_get_function_versions_dispatcher): Change the warning
21179 that an old GLIBC is used which does not export the capability
21180 bits to be an error.
21181 * doc/extend.texi (target_clones attribute): Document the
21182 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
21183 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
21184 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
21185 the macros defined by GCC if the newer GLIBC is available.
21187 2017-07-12 Jeff Law <law@redhat.com>
21189 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
21190 remaining includes slightly.
21191 * config/riscv/riscv-builtins.c: Include profile-count.h.
21193 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21196 * config/avr/avr.c (avr_set_current_function): In diagnostic
21197 messages: Quote keywords and (parts of) identifiers.
21198 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
21201 2017-07-12 Carl Love <cel@us.ibm.com>
21203 * config/rs6000/rs6000-c.c: Add support for built-in functions
21204 vector bool char vec_revb (vector bool char);
21205 vector bool short vec_revb (vector short char);
21206 vector bool int vec_revb (vector bool int);
21207 vector bool long long vec_revb (vector bool long long);
21208 * doc/extend.texi: Update the built-in documentation file for the
21209 new built-in functions.
21211 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21213 * config/s390/s390.md: Remove movcc splitter.
21215 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21217 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
21218 load/store on condition.
21220 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21223 * config/avr/avr.c (avr_encode_section_info)
21224 [progmem && !TREE_READONLY]: Error if progmem object needs
21227 2017-07-11 Michael Collison <michael.collison@arm.com>
21229 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
21232 2017-07-11 Carl Love <cel@us.ibm.com>
21234 * config/rs6000/rs6000-c.c: Add support for builtins
21235 vector unsigned int vec_parity_lsbb (vector signed int);
21236 vector unsigned int vec_parity_lsbb (vector unsigned int);
21237 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
21238 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
21239 vector unsigned long long vec_parity_lsbb (vector signed long long);
21240 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
21241 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
21242 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
21243 * doc/extend.texi: Update the built-in documentation file for the
21244 new built-in functions.
21246 2017-07-11 David Malcolm <dmalcolm@redhat.com>
21248 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
21249 (layout::m_primary_loc): New field.
21250 (layout::layout): Initialize new field. Move location filtering
21251 logic from here to...
21252 (layout::maybe_add_location_range): ...this new method. Add
21253 support for filtering to just the lines already specified by other
21255 (layout::will_show_line_p): New method.
21256 (gcc_rich_location::add_location_if_nearby): New method.
21257 (selftest::test_add_location_if_nearby): New test function.
21258 (selftest::diagnostic_show_locus_c_tests): Call it.
21259 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
21262 2017-07-11 Tom de Vries <tom@codesourcery.com>
21264 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
21265 (bb_first_real_insn): New function.
21266 (nvptx_single): Add extra initialization of broadcasted condition
21269 2017-07-11 Nathan Sidwell <nathan@acm.org>
21271 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
21273 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
21275 * doc/extend.texi (AVR Function Attributes): Remove weblink to
21276 Binutils doc as TEXI will mess them up.
21277 * doc/invoke.texi (AVR Options): Same here.
21279 2017-07-11 Daniel Cederman <cederman@gaisler.com>
21281 * config/sparc/sparc.opt (mfix-ut700): New option.
21282 (mfix-gr712rc): Likewise.
21283 (sparc_fix_b2bst): New variable.
21284 * doc/invoke.texi (SPARC options): Document them.
21285 (ARM options): Fix warnings.
21286 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
21287 instructions to prevent sequences that can trigger the store-store
21288 errata for certain LEON3FT processors.
21289 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
21290 (sparc_option_override): Set sparc_fix_b2bst appropriately.
21291 * config/sparc/sparc.md (fix_b2bst): New attribute.
21292 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
21294 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
21297 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
21299 (*rsqrtsf2_sse): Ditto.
21301 (div<mode>3): Macroize insn from divdf3 and divsf3
21302 using MODEF mode iterator.
21304 2017-07-10 Martin Sebor <msebor@redhat.com>
21306 PR tree-optimization/80397
21307 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
21308 instead of testing for equality to INTEGER_TYPE.
21310 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
21312 * config.gcc: Remove uclibc from arc target spec.
21314 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
21316 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
21318 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21321 * lto-wrapper.c (remove_option): New function.
21322 (merge_and_complain): Merge PIC/PIE options more realistically.
21324 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21326 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
21330 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
21331 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
21332 * config.in: Regenerate.
21333 * configure: Regenerate.
21334 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
21335 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
21336 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
21337 (TARGET_GASISR_PROLOGUES): ...target mask.
21338 * common/config/avr/avr-common.c
21339 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
21340 Set -mgas-isr-prologues.
21341 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
21342 INSERT_PASS_BEFORE for it.
21343 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
21344 * config/avr/avr.c (avr_option_override)
21345 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
21346 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
21347 (avr_attribute_table) <no_gccisr>: Add new function attribute.
21348 (avr_set_current_function) <is_no_gccisr>: Init machine field.
21349 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
21351 (make_avr_pass_pre_proep): New function.
21352 (emit_push_sfr) <treg>: Add argument to function and use it
21353 instead of TMP_REG.
21354 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
21355 and set machine->gasisr.yes.
21356 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
21357 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
21358 __gcc_isr.n_pushed to .L__stack_usage.
21359 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
21360 (avr_asm_final_postscan_insn): ...this new static function.
21361 * config/avr/avr.h (machine_function)
21362 <is_no_gccisr, use_L__stack_usage>: New fields.
21363 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
21364 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
21365 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
21366 (gasisr, *gasisr): New expander and insn.
21367 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21368 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
21369 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
21371 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
21373 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
21376 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21378 Move jump-tables out of .text again.
21381 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
21382 (ASM_OUTPUT_ADDR_VEC): New function.
21383 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
21384 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
21385 INSN_ADDRESSes as asm comment.
21386 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
21387 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
21388 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
21389 * config/avr/avr.md (*tablejump): Adjust comment.
21390 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
21391 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
21393 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
21394 (avr_output_addr_vec): New proto.
21395 (avr_log_t) <insn_addresses>: New field.
21397 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
21400 * config/i386/i386.c (ix86_function_arg_advance): Set
21401 outgoing_args_on_stack to true if there are outgoing arguments
21403 (ix86_function_arg): Likewise.
21404 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
21405 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
21406 * config/i386/i386.h (machine_function): Add
21407 outgoing_args_on_stack.
21409 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
21411 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
21412 supporting pthreds.
21413 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
21415 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
21417 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
21418 (REAL_H): Remove $(MACHMODE_H).
21419 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
21421 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
21422 $(MACHMODE_H) and double-int.h.
21423 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
21425 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
21428 2017-07-07 Andrew Pinski <apinski@cavium.com>
21430 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
21431 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
21433 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21435 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
21436 Add warning if GCC was not configured to link against a GLIBC that
21437 exports the hardware capability bits.
21438 (make_resolver_func): Make resolver function private and not a
21439 COMDAT function. Create the name with clone_function_name instead
21440 of make_unique_name.
21443 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
21444 correct operand in doing the split.
21446 2017-07-07 Carl Love <cel@us.ibm.com>
21448 * config/rs6000/rs6000-c: Add support for built-in function
21449 vector unsigned short vec_pack_to_short_fp32 (vector float,
21451 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
21452 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
21453 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
21454 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
21455 (convert_4f32_8i16): Add define_expand.
21456 * doc/extend.texi: Update the built-in documentation file for the
21457 new built-in function.
21459 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21461 * config/sparc/m8.md: New file.
21462 * config/sparc/sparc.md: Include m8.md.
21464 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21466 * config/sparc/sparc.opt: New option -mvis4b.
21467 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
21468 (sparc_option_override): Handle VIS4B.
21469 (enum sparc_builtins): Define
21470 SPARC_BUILTIN_DICTUNPACK{8,16,32},
21471 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
21472 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
21473 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
21474 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
21475 (check_constant_argument): New function.
21476 (sparc_vis_init_builtins): Define builtins
21477 __builtin_vis_dictunpack{8,16,32},
21478 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
21479 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
21480 __builtin_vis_fpcmpde{8,16,32}shl and
21481 __builtin_vis_fpcmpur{8,16,32}shl.
21482 (sparc_expand_builtin): Check that the constant operands to
21483 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
21484 constant and in range.
21485 * config/sparc/sparc-c.c (sparc_target_macros): Handle
21487 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
21488 (SPARC_IMM5_P): Likewise.
21489 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
21490 (enabled): Handle vis4b.
21491 (UNSPEC_DICTUNPACK): New unspec.
21492 (UNSPEC_FPCMPSHL): Likewise.
21493 (UNSPEC_FPUCMPSHL): Likewise.
21494 (UNSPEC_FPCMPDESHL): Likewise.
21495 (UNSPEC_FPCMPURSHL): Likewise.
21496 (cpu_feature): New CPU feature `vis4b'.
21497 (dictunpack{8,16,32}): New insns.
21498 (FPCSMODE): New mode iterator.
21499 (fpcscond): New code iterator.
21500 (fpcsucond): Likewise.
21501 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
21502 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
21503 (fpcmpde{8,16,32}{si,di}shl): Likewise.
21504 (fpcmpur{8,16,32}{si,di}shl): Likewise.
21505 * config/sparc/constraints.md: Define constraints `q' for unsigned
21506 2-bit integer constants and `t' for unsigned 5-bit integer
21508 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
21510 (imm5_operand_dictunpack16): Likewise.
21511 (imm5_operand_dictunpack32): Likewise.
21512 (imm2_operand): Likewise.
21513 * doc/invoke.texi (SPARC Options): Document -mvis4b.
21514 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
21515 ditunpack* and fpcmp*shl builtins.
21517 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21519 * config.gcc: Handle m8 in --with-{cpu,tune} options.
21520 * config.in: Add HAVE_AS_SPARC6 define.
21521 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
21523 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
21525 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
21526 (CPP_CPU_SPEC): Handle m8.
21527 (ASM_CPU_SPEC): Likewise.
21528 * config/sparc/sparc-opts.h (enum processor_type): Add
21530 * config/sparc/sparc.c (m8_costs): New struct.
21531 (sparc_option_override): Handle TARGET_CPU_m8.
21532 (sparc32_initialize_trampoline): Likewise.
21533 (sparc64_initialize_trampoline): Likewise.
21534 (sparc_issue_rate): Likewise.
21535 (sparc_register_move_cost): Likewise.
21536 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
21537 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
21538 (ASM_CPU64_DEFAULT_SPEC): Likewise.
21539 (CPP_CPU_SPEC): Handle M8.
21540 (ASM_CPU_SPEC): Likewise.
21541 (AS_M8_FLAG): Define.
21542 * config/sparc/sparc.md: Add m8 to the cpu attribute.
21543 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
21544 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
21546 * configure: Regenerate.
21547 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
21550 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21552 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
21554 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
21555 ("*movdi_insn_sp32"): Do not set v3pipe.
21556 ("*movsi_insn"): Likewise.
21557 ("*movdi_insn_sp64"): Likewise.
21558 ("*movsf_insn"): Likewise.
21559 ("*movdf_insn_sp32"): Likewise.
21560 ("*movdf_insn_sp64"): Likewise.
21561 ("*zero_extendsidi2_insn_sp64"): Likewise.
21562 ("*sign_extendsidi2_insn"): Likewise.
21563 ("*mov<VM32:mode>_insn"): Likewise.
21564 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21565 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21566 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21567 ("<vlop:code><VL:mode>3"): Likewise.
21568 ("*not_<vlop:code><VL:mode>3"): Likewise.
21569 ("*nand<VL:mode>_vis"): Likewise.
21570 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
21571 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
21572 ("one_cmpl<VL:mode>2"): Likewise.
21573 ("faligndata<VM64:mode>_vis"): Likewise.
21574 ("alignaddrsi_vis"): Likewise.
21575 ("alignaddrdi_vis"): Likweise.
21576 ("alignaddrlsi_vis"): Likewise.
21577 ("alignaddrldi_vis"): Likewise.
21578 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21579 ("bmaskdi_vis"): Likewise.
21580 ("bmasksi_vis"): Likewise.
21581 ("bshuffle<VM64:mode>_vis"): Likewise.
21582 ("cmask8<P:mode>_vis"): Likewise.
21583 ("cmask16<P:mode>_vis"): Likewise.
21584 ("cmask32<P:mode>_vis"): Likewise.
21585 ("pdistn<P:mode>_vis"): Likewise.
21586 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21588 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21590 * config/sparc/sparc.md ("subtype"): New insn attribute.
21591 ("*wrgsr_sp64"): Set insn subtype.
21592 ("*rdgsr_sp64"): Likewise.
21593 ("alignaddrsi_vis"): Likewise.
21594 ("alignaddrdi_vis"): Likewise.
21595 ("alignaddrlsi_vis"): Likewise.
21596 ("alignaddrldi_vis"): Likewise.
21597 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21598 ("fexpand_vis"): Likewise.
21599 ("fpmerge_vis"): Likewise.
21600 ("faligndata<VM64:mode>_vis"): Likewise.
21601 ("bshuffle<VM64:mode>_vis"): Likewise.
21602 ("cmask8<P:mode>_vis"): Likewise.
21603 ("cmask16<P:mode>_vis"): Likewise.
21604 ("cmask32<P:mode>_vis"): Likewise.
21605 ("fchksm16_vis"): Likewise.
21606 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
21607 ("fmean16_vis"): Likewise.
21608 ("fp<plusminus_insn>64_vis"): Likewise.
21609 ("<plusminus_insn>v8qi3"): Likewise.
21610 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21611 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
21612 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
21613 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
21614 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
21615 ("*movqi_insn"): Likewise.
21616 ("*movhi_insn"): Likewise.
21617 ("*movsi_insn"): Likewise.
21618 ("movsi_pic_gotdata_op"): Likewise.
21619 ("*movdi_insn_sp32"): Likewise.
21620 ("*movdi_insn_sp64"): Likewise.
21621 ("movdi_pic_gotdata_op"): Likewise.
21622 ("*movsf_insn"): Likewise.
21623 ("*movdf_insn_sp32"): Likewise.
21624 ("*movdf_insn_sp64"): Likewise.
21625 ("*zero_extendhisi2_insn"): Likewise.
21626 ("*zero_extendqihi2_insn"): Likewise.
21627 ("*zero_extendqisi2_insn"): Likewise.
21628 ("*zero_extendqidi2_insn"): Likewise.
21629 ("*zero_extendhidi2_insn"): Likewise.
21630 ("*zero_extendsidi2_insn_sp64"): Likewise.
21631 ("ldfsr"): Likewise.
21632 ("prefetch_64"): Likewise.
21633 ("prefetch_32"): Likewise.
21634 ("tie_ld32"): Likewise.
21635 ("tie_ld64"): Likewise.
21636 ("*tldo_ldub_sp32"): Likewise.
21637 ("*tldo_ldub1_sp32"): Likewise.
21638 ("*tldo_ldub2_sp32"): Likewise.
21639 ("*tldo_ldub_sp64"): Likewise.
21640 ("*tldo_ldub1_sp64"): Likewise.
21641 ("*tldo_ldub2_sp64"): Likewise.
21642 ("*tldo_ldub3_sp64"): Likewise.
21643 ("*tldo_lduh_sp32"): Likewise.
21644 ("*tldo_lduh1_sp32"): Likewise.
21645 ("*tldo_lduh_sp64"): Likewise.
21646 ("*tldo_lduh1_sp64"): Likewise.
21647 ("*tldo_lduh2_sp64"): Likewise.
21648 ("*tldo_lduw_sp32"): Likewise.
21649 ("*tldo_lduw_sp64"): Likewise.
21650 ("*tldo_lduw1_sp64"): Likewise.
21651 ("*tldo_ldx_sp64"): Likewise.
21652 ("*mov<VM32:mode>_insn"): Likewise.
21653 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21654 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21656 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21658 * config/sparc/sparc.md ("type"): New insn type viscmp.
21659 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
21661 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
21662 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
21663 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21664 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
21666 ("n7_vis_logical_11cycle"): Likewise.
21667 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
21668 * config/sparc/niagara2.md ("niag3_vis": Likewise.
21669 * config/sparc/niagara.md ("niag_vis"): Likewise.
21670 * config/sparc/ultra3.md ("us3_fga"): Likewise.
21671 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
21673 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21675 * config/sparc/sparc.md: New instruction type `bmask'.
21676 (bmaskdi_vis): Use the `bmask' type.
21677 (bmasksi_vis): Likewise.
21678 * config/sparc/ultra3.md (us3_array): Likewise.
21679 * config/sparc/niagara7.md (n7_array): Likewise.
21680 * config/sparc/niagara4.md (n4_array): Likewise.
21681 * config/sparc/niagara2.md (niag2_vis): Likewise.
21682 (niag3_vis): Likewise.
21683 * config/sparc/niagara.md (niag_vis): Likewise.
21685 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21687 * ipa-comdats.c: Remove optimize check from gate.
21688 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
21689 for functions not optimized.
21690 (ipa_fn_summary_read): Skip optimize check.
21691 (ipa_fn_summary_write): Likewise.
21692 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
21694 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
21696 (can_inline_edge_p): Check flag_pcc_struct_return for match.
21697 (check_callers): Give up on caller which is not optimized.
21698 (inline_small_functions): Likewise.
21699 (ipa_inline): Do not give up when not optimizing.
21700 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
21701 away unoptimizes cdtors.
21702 (whole_program_function_and_variable_visibility): Do
21703 ipa_discover_readonly_nonaddressable_vars in LTO mode.
21704 * ipa.c (process_references): Do not check optimize.
21705 (symbol_table::remove_unreachable_nodes): Update optimize check.
21706 (set_writeonly_bit): Update optimize check.
21707 (pass_ipa_cdtor_merge::gate): Do not check optimize.
21708 (pass_ipa_single_use::gate): Remove.
21710 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21712 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
21713 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
21714 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
21715 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
21716 permute_load, permute_store, adjust_extract, adjust_splat,
21717 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
21718 replace_swap_with_copy, dump_swap_insn_table,
21719 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
21720 recombine_lvx_pattern, recombine_stvx_pattern,
21721 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
21722 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
21723 to file rs6000-p8swap.c.
21724 * config/rs6000/rs6000-p8swap.c: New file.
21725 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
21726 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
21727 and rs6000*-*-* targets.
21729 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21731 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
21733 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21735 * lto-wrapper.c (merge_and_complain): Do not merge
21736 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
21737 fsigned_zeros, ftrapping_math, fwrapv.
21738 (append_compiler_options): Do not track these options.
21739 (append_linker_options): Likewie
21741 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21743 * cgraphunit.c (cgraph_node::finalize_function): When
21744 !flag_toplevel_reorde set no_reorder flag.
21745 (varpool_node::finalize_decl): Likewise.
21746 (symbol_table::compile): Drop no toplevel reorder path.
21748 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21750 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
21751 edges; zero probability is not better than uninitialized.
21753 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
21755 * asan.h (asan_sanitize_allocas_p): Declare.
21756 * asan.c (asan_sanitize_allocas_p): New function.
21757 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
21758 (handle_builtin_alloca): Likewise.
21759 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
21760 if !asan_sanitize_allocas_p.
21761 * params.def (asan-instrument-allocas): Add new option.
21762 * params.h (ASAN_PROTECT_ALLOCAS): Define.
21763 * opts.c (common_handle_option): Disable allocas sanitization for
21766 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
21768 * asan.c: Include gimple-fold.h.
21769 (get_last_alloca_addr): New function.
21770 (handle_builtin_stackrestore): Likewise.
21771 (handle_builtin_alloca): Likewise.
21772 (asan_emit_allocas_unpoison): Likewise.
21773 (get_mem_refs_of_builtin_call): Add new parameter, remove const
21774 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
21775 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
21776 (instrument_builtin_call): Pass gimple iterator to
21777 get_mem_refs_of_builtin_call.
21778 (last_alloca_addr): New global.
21779 * asan.h (asan_emit_allocas_unpoison): Declare.
21780 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
21781 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
21782 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
21783 if function calls alloca.
21784 * gimple-fold.c (replace_call_with_value): Remove static keyword.
21785 * gimple-fold.h (replace_call_with_value): Declare.
21786 * internal-fn.c: Include asan.h.
21787 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
21788 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
21790 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21792 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
21793 (C_SELFTEST_FLAGS): New.
21794 (CPP_SELFTEST_FLAGS): New.
21795 (SELFTEST_DEPS): New, from deps of s-selftest.
21796 (C_SELFTEST_DEPS): New, from deps of s-selftest.
21797 (CPP_SELFTEST_DEPS): New.
21798 (selftest): Add dependency on s-selftest-c++.
21799 (s-selftest): Rename to...
21800 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
21801 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
21802 than SELFTEST_FLAGS.
21803 (selftest-gdb): Rename to...
21804 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
21806 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
21807 (selftest-valgrind): Rename to...
21808 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
21810 (selftest-valgrind): Reintroduce as an alias for
21811 selftest-c-valgrind.
21812 (s-selftest-c++): New.
21813 (selftest-c++-gdb): New.
21814 (selftest-c++-valgrind): New.
21816 2017-07-06 Olivier Hainque <hainque@adacore.com>
21818 * gcc.c (process_command): When deciding if undefined variables
21819 should be ignored when processing specs, accept "gcc -v" as well.
21821 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21823 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
21824 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
21826 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21828 * config/arm/arm-cpus.in (armv8-r): Add new entry.
21829 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
21830 * config/arm/arm-tables.opt: Regenerate.
21831 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
21833 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
21835 2017-07-06 Carl Love <cel@us.ibm.com>
21837 * ChangeLog: Clean up from mid air collision
21839 2017-07-06 Carl Love <cel@us.ibm.com>
21841 * config/rs6000/rs6000-c.c: Add support for built-in functions
21842 vector signed int vec_subc (vector signed int, vector signed int);
21843 vector signed __int128 vec_subc (vector signed __int128,
21844 vector signed __int128);
21845 vector unsigned __int128 vec_subc (vector unsigned __int128,
21846 vector unsigned __int128);
21847 vector signed int vec_sube (vector signed int, vector signed int,
21848 vector signed int);
21849 vector unsigned int vec_sube (vector unsigned int,
21850 vector unsigned int,
21851 vector unsigned int);
21852 vector signed __int128 vec_sube (vector signed __int128,
21853 vector signed __int128,
21854 vector signed__int128);
21855 vector unsigned __int128 vec_sube (vector unsigned __int128,
21856 vector unsigned __int128,
21857 vector unsigned __int128);
21858 vector signed int vec_subec (vector signed int, vector signed int,
21859 vector signed int);
21860 vector unsigned int vec_subec (vector unsigned int,
21861 vector unsigned int,
21862 vector unsigned int);
21863 vector signed __int128 vec_subec (vector signed __int128,
21864 vector signed __int128,
21865 vector signed__int128);
21866 vector unsigned __int128 vec_subec (vector unsigned __int128,
21867 vector unsigned __int128,
21868 vector unsigned __int128);
21869 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
21870 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
21871 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
21872 BU_ALTIVEC_OVERLOAD_X definitions.
21873 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
21874 * doc/extend.texi: Update the built-in documentation file for the new
21875 built-in functions.
21877 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21880 * diagnostic-show-locus.c (layout::layout): Use start and finish
21881 spelling location for the start and finish of each range.
21882 * genmatch.c (linemap_client_expand_location_to_spelling_point):
21883 Add unused aspect param.
21884 * input.c (expand_location_1): Add "aspect" param, and use it
21885 to access the correct part of the location.
21886 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
21888 (expand_location_to_spelling_point): Likewise.
21889 (linemap_client_expand_location_to_spelling_point): Add "aspect"
21890 param, and pass it to expand_location_1.
21892 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
21894 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
21895 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
21896 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
21897 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
21898 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
21899 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
21900 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
21901 _mm_maskz_getmant_ss): New intrinsics.
21902 (__builtin_ia32_getexpss128_mask): Changed to ...
21903 __builtin_ia32_getexpss128_round ... this.
21904 (__builtin_ia32_getexpsd128_mask): Changed to ...
21905 __builtin_ia32_getexpsd128_round ... this.
21906 * config/i386/i386-builtin-types.def
21907 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
21908 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
21909 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
21910 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
21911 __builtin_ia32_getmantss_mask_round): New builtins.
21912 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
21913 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
21914 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
21915 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
21916 * config/i386/sse.md
21917 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
21918 avx512f_sgetexp<mode><mask_scalar_name>
21919 <round_saeonly_scalar_name> ... this.
21920 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
21921 %0, %1, %2<round_saeonly_op3>}): Changed to ...
21922 vgetexp<ssescalarmodesuffix>
21923 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21924 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
21925 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
21926 avx512f_vgetmant<mode><mask_scalar_name>
21927 <round_saeonly_scalar_name> ... this.
21928 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
21929 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
21930 vgetmant<ssescalarmodesuffix>
21931 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
21932 %0<mask_scalar_operand4>, %1, %2
21933 <round_saeonly_scalar_mask_op4>, %3} ... this.
21934 * config/i386/subst.md (mask_scalar_operand4,
21935 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
21936 round_saeonly_scalar_nimm_predicate): New subst attributes.
21938 2017-07-06 Julia Koval <julia.koval@intel.com>
21940 * config/i386/i386.c (ix86_erase_embedded_rounding):
21941 Remove code for old rounding pattern.
21943 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
21945 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
21947 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
21949 * doc/sourcebuild.texi (Test Directives, Variants of
21950 dg-require-support): Add documentation for dg-require-stack-check.
21952 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
21954 * config/i386/subst.md (mask_scalar, round_scalar,
21955 round_saeonly_scalar): New meta-templates.
21956 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
21957 round_scalar_mask_operand3, round_scalar_mask_op3,
21958 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
21959 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
21960 round_saeonly_scalar_constraint,
21961 round_saeonly_scalar_prefix): New subst attribute.
21962 * config/i386/sse.md
21963 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
21964 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
21965 <round_scalar_name> ... this.
21966 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
21967 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
21968 <round_scalar_name> ... this.
21969 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
21970 <sse>_vm<code><mode>3<mask_scalar_name>
21971 <round_saeonly_scalar_name> ... this.
21972 (v<plusminus_mnemonic><ssescalarmodesuffix>
21973 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21974 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21975 v<plusminus_mnemonic><ssescalarmodesuffix>
21976 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21977 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21978 (v<multdiv_mnemonic><ssescalarmodesuffix>
21979 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21980 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21981 v<multdiv_mnemonic><ssescalarmodesuffix>
21982 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21983 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21984 (v<maxmin_float><ssescalarmodesuffix>
21985 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
21986 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
21987 v<maxmin_float><ssescalarmodesuffix>
21988 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21989 %0<mask_scalar_operand3>, %1, %<iptr>2
21990 <round_saeonly_scalar_mask_op3>} ... this.
21992 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
21994 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
21995 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
21997 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
21998 Alan Hayward <alan.hayward@arm.com>
21999 David Sherwood <david.sherwood@arm.com>
22001 * combine.c (simplify_if_then_else): Remove "enum" before
22003 * compare-elim.c (can_eliminate_compare): Likewise.
22004 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
22006 (aarch64_lookup_simd_builtin_type): Likewise.
22007 (aarch64_simd_builtin_type): Likewise.
22008 (aarch64_init_simd_builtin_types): Likewise.
22009 (aarch64_simd_expand_args): Likewise.
22010 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
22012 (aarch64_reverse_mask): Likewise.
22013 (aarch64_simd_emit_reg_reg_move): Likewise.
22014 (aarch64_gen_adjusted_ldpstp): Likewise.
22015 (aarch64_ccmp_mode_to_code): Likewise.
22016 (aarch64_operands_ok_for_ldpstp): Likewise.
22017 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
22018 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
22020 (aarch64_min_divisions_for_recip_mul): Likewise.
22021 (aarch64_reassociation_width): Likewise.
22022 (aarch64_get_condition_code_1): Likewise.
22023 (aarch64_simd_emit_reg_reg_move): Likewise.
22024 (aarch64_simd_attr_length_rglist): Likewise.
22025 (aarch64_reverse_mask): Likewise.
22026 (aarch64_operands_ok_for_ldpstp): Likewise.
22027 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
22028 (aarch64_gen_adjusted_ldpstp): Likewise.
22029 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
22031 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
22032 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
22033 (arm_lookup_simd_builtin_type): Likewise.
22034 (arm_simd_builtin_type): Likewise.
22035 (arm_init_simd_builtin_types): Likewise.
22036 (arm_expand_builtin_args): Likewise.
22037 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
22038 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
22039 (ft32_setup_incoming_varargs): Likewise.
22040 (ft32_function_arg): Likewise.
22041 (ft32_function_arg_advance): Likewise.
22042 (ft32_pass_by_reference): Likewise.
22043 (ft32_arg_partial_bytes): Likewise.
22044 (ft32_valid_pointer_mode): Likewise.
22045 (ft32_addr_space_pointer_mode): Likewise.
22046 (ft32_addr_space_legitimate_address_p): Likewise.
22047 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
22049 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
22050 (ix86_emit_outlined_ms2sysv_restore): Likewise.
22051 (iamcu_alignment): Likewise.
22052 (canonicalize_vector_int_perm): Likewise.
22053 (ix86_noce_conversion_profitable_p): Likewise.
22054 (ix86_mpx_bound_mode): Likewise.
22055 (ix86_operands_ok_for_move_multiple): Likewise.
22056 * config/microblaze/microblaze-protos.h
22057 (microblaze_expand_conditional_branch_reg): Likewise.
22058 * config/microblaze/microblaze.c
22059 (microblaze_expand_conditional_branch_reg): Likewise.
22060 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
22062 (rs6000_reassociation_width): Likewise.
22063 (rs6000_invalid_binary_op): Likewise.
22064 (fusion_p9_p): Likewise.
22065 (emit_fusion_p9_load): Likewise.
22066 (emit_fusion_p9_store): Likewise.
22067 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
22069 (riscv_hard_regno_mode_ok_p): Likewise.
22070 (riscv_address_insns): Likewise.
22071 (riscv_split_symbol): Likewise.
22072 (riscv_legitimize_move): Likewise.
22073 (riscv_function_value): Likewise.
22074 (riscv_hard_regno_nregs): Likewise.
22075 (riscv_expand_builtin): Likewise.
22076 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
22077 (riscv_build_integer): Likewise.
22078 (riscv_split_integer): Likewise.
22079 (riscv_legitimate_constant_p): Likewise.
22080 (riscv_cannot_force_const_mem): Likewise.
22081 (riscv_regno_mode_ok_for_base_p): Likewise.
22082 (riscv_valid_base_register_p): Likewise.
22083 (riscv_valid_offset_p): Likewise.
22084 (riscv_valid_lo_sum_p): Likewise.
22085 (riscv_classify_address): Likewise.
22086 (riscv_legitimate_address_p): Likewise.
22087 (riscv_address_insns): Likewise.
22088 (riscv_load_store_insns): Likewise.
22089 (riscv_force_binary): Likewise.
22090 (riscv_split_symbol): Likewise.
22091 (riscv_force_address): Likewise.
22092 (riscv_legitimize_address): Likewise.
22093 (riscv_move_integer): Likewise.
22094 (riscv_legitimize_const_move): Likewise.
22095 (riscv_legitimize_move): Likewise.
22096 (riscv_address_cost): Likewise.
22097 (riscv_subword): Likewise.
22098 (riscv_output_move): Likewise.
22099 (riscv_canonicalize_int_order_test): Likewise.
22100 (riscv_emit_int_order_test): Likewise.
22101 (riscv_function_arg_boundary): Likewise.
22102 (riscv_pass_mode_in_fpr_p): Likewise.
22103 (riscv_pass_fpr_single): Likewise.
22104 (riscv_pass_fpr_pair): Likewise.
22105 (riscv_get_arg_info): Likewise.
22106 (riscv_function_arg): Likewise.
22107 (riscv_function_arg_advance): Likewise.
22108 (riscv_arg_partial_bytes): Likewise.
22109 (riscv_function_value): Likewise.
22110 (riscv_pass_by_reference): Likewise.
22111 (riscv_setup_incoming_varargs): Likewise.
22112 (riscv_print_operand): Likewise.
22113 (riscv_elf_select_rtx_section): Likewise.
22114 (riscv_save_restore_reg): Likewise.
22115 (riscv_for_each_saved_reg): Likewise.
22116 (riscv_register_move_cost): Likewise.
22117 (riscv_hard_regno_mode_ok_p): Likewise.
22118 (riscv_hard_regno_nregs): Likewise.
22119 (riscv_class_max_nregs): Likewise.
22120 (riscv_memory_move_cost): Likewise.
22121 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
22122 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
22123 (rl78_addr_space_address_mode): Likewise.
22124 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22126 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
22127 (rs6000_reassociation_width): Likewise.
22128 (rs6000_invalid_binary_op): Likewise.
22129 (fusion_p9_p): Likewise.
22130 (emit_fusion_p9_load): Likewise.
22131 (emit_fusion_p9_store): Likewise.
22132 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
22133 (ok_for_simple_move_operands): Likewise.
22134 (ok_for_simple_move_strict_operands): Likewise.
22135 (ok_for_simple_arith_logic_operands): Likewise.
22136 (visium_legitimize_reload_address): Likewise.
22137 (visium_select_cc_mode): Likewise.
22138 (output_cbranch): Likewise.
22139 (visium_split_double_move): Likewise.
22140 (visium_expand_copysign): Likewise.
22141 (visium_expand_int_cstore): Likewise.
22142 (visium_expand_fp_cstore): Likewise.
22143 * config/visium/visium.c (visium_pass_by_reference): Likewise.
22144 (visium_function_arg): Likewise.
22145 (visium_function_arg_advance): Likewise.
22146 (visium_libcall_value): Likewise.
22147 (visium_setup_incoming_varargs): Likewise.
22148 (visium_legitimate_constant_p): Likewise.
22149 (visium_legitimate_address_p): Likewise.
22150 (visium_legitimize_address): Likewise.
22151 (visium_secondary_reload): Likewise.
22152 (visium_register_move_cost): Likewise.
22153 (visium_memory_move_cost): Likewise.
22154 (prepare_move_operands): Likewise.
22155 (ok_for_simple_move_operands): Likewise.
22156 (ok_for_simple_move_strict_operands): Likewise.
22157 (ok_for_simple_arith_logic_operands): Likewise.
22158 (visium_function_value_1): Likewise.
22159 (rtx_ok_for_offset_p): Likewise.
22160 (visium_legitimize_reload_address): Likewise.
22161 (visium_split_double_move): Likewise.
22162 (visium_expand_copysign): Likewise.
22163 (visium_expand_int_cstore): Likewise.
22164 (visium_expand_fp_cstore): Likewise.
22165 (visium_split_cstore): Likewise.
22166 (visium_select_cc_mode): Likewise.
22167 (visium_split_cbranch): Likewise.
22168 (output_cbranch): Likewise.
22169 (visium_print_operand_address): Likewise.
22170 * expmed.c (flip_storage_order): Likewise.
22171 * expmed.h (emit_cstore): Likewise.
22172 (flip_storage_order): Likewise.
22173 * genrecog.c (validate_pattern): Likewise.
22174 * hsa-gen.c (gen_hsa_addr): Likewise.
22175 * internal-fn.c (expand_arith_overflow): Likewise.
22176 * ira-color.c (allocno_copy_cost_saving): Likewise.
22177 * lra-assigns.c (find_hard_regno_for_1): Likewise.
22178 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
22179 (process_invariant_for_inheritance): Likewise.
22180 * lra-eliminations.c (move_plus_up): Likewise.
22181 * omp-low.c (lower_oacc_reductions): Likewise.
22182 * simplify-rtx.c (simplify_subreg): Likewise.
22183 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
22184 (TARGET_CHKP_BOUND_MODE): Likewise..
22185 * targhooks.c (default_chkp_bound_mode): Likewise.
22186 (default_setup_incoming_vararg_bounds): Likewise.
22187 * targhooks.h (default_chkp_bound_mode): Likewise.
22188 (default_setup_incoming_vararg_bounds): Likewise.
22189 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
22190 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
22191 (have_whole_vector_shift): Likewise.
22192 * tree-vect-stmts.c (vectorizable_load): Likewise.
22193 * doc/tm.texi: Regenerate.
22195 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22197 Graceful degrade if Binutils PR21472 is not available.
22200 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
22201 .rodata in flash test fails.
22202 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
22203 * confgure: Regenerate.
22204 * config.in: Regenerate.
22205 * config/avr/avr.c (avr_asm_named_section)
22206 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
22207 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
22208 (avr_asm_init_sections): Same.
22210 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22212 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
22213 (fma<VH:mode>4_intrinsic): Likewise.
22214 (*fmsub<VCVTF:mode>4): Likewise.
22215 (*fmsub<VH:mode>4_intrinsic): Likewise.
22217 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22220 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
22221 Don't depend on "optimize > 0".
22222 (out_movhi_r_mr, out_movqi_mr_r): Same.
22223 (out_movhi_mr_r, out_movqi_r_mr): Same.
22224 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
22225 io_address_operand on "optimize > 0".
22227 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22229 * tree-loop-distribution.c: Add general explanantion on the pass.
22230 (generate_loops_for_partition): Mark distributed loop.
22231 (pg_add_dependence_edges): New parameter. Handle alias data
22232 dependence specially and record it in the parameter if asked.
22233 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
22234 (init_partition_graph_vertices, add_partition_graph_edge): New.
22235 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
22236 (free_partition_graph_vdata, build_partition_graph): New.
22237 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
22238 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
22239 (data_ref_segment_size, latch_dominated_by_data_ref): New.
22240 (compute_alias_check_pairs, version_loop_by_alias_check): New.
22241 (version_for_distribution_p, finalize_partitions): New.
22242 (distribute_loop): Handle alias data dependence specially. Factor
22243 out loop fusion code as functions and call these functions.
22245 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22247 * tree-loop-distribution.c (classify_partition): New parameter and
22248 better handle reduction statement.
22249 (rdg_build_partitions): Revise comment.
22250 (distribute_loop): Compute statements in all partitions and pass it
22251 to classify_partition.
22253 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22255 * tree-loop-distribution.c (enum partition_type): New.
22256 (struct partition): New field type.
22257 (partition_merge_into): Add parameter. Update partition type.
22258 (data_dep_in_cycle_p, update_type_for_merge): New functions.
22259 (build_rdg_partition_for_vertex): Compute partition type.
22260 (rdg_build_partitions): Dump partition type.
22261 (distribute_loop): Update calls to partition_merge_into.
22263 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22265 * tree-loop-distribution.c (struct ddr_hasher): New.
22266 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
22268 (classify_partition): Call get_data_dependence.
22269 (pg_add_dependence_edges): Ditto.
22270 (distribute_loop): Release data dependence hash table.
22272 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22274 * tree-loop-distribution.c (ref_base_address): Delete.
22275 (similar_memory_accesses): Rename ...
22276 (share_memory_accesses): ... to this. Check if partitions access
22277 the same memory reference.
22278 (distribute_loop): Call share_memory_accesses.
22280 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22282 * tree-loop-distribution.c (struct partition): New field recording
22283 its data reference.
22284 (partition_alloc, partition_free): Init and release data refs.
22285 (partition_merge_into): Merge data refs.
22286 (build_rdg_partition_for_vertex): Collect data refs for partition.
22287 (pg_add_dependence_edges): Change parameters from vector to bitmap.
22289 (distribute_loop): Remve data refs from vertice data of partition
22292 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22294 * tree-loop-distribution.c (params.h): Include header file.
22295 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
22296 (datarefs_vec): New global var.
22297 (create_rdg_vertices): Use datarefs_vec directly.
22298 (free_rdg): Don't free data references.
22299 (build_rdg): Update use. Don't free data references.
22300 (distribute_loop): Compute global variable for data references.
22301 Bail out if there are too many data references.
22303 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22305 * tree-loop-distribution.c (loop_nest): New global var.
22306 (build_rdg): Use loop directly, rather than loop nest.
22307 (pg_add_dependence_edges): Remove loop nest parameter. Use global
22309 (distribute_loop): Compute global variable loop nest. Update use.
22311 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22313 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
22314 (partition_merge_into): New parameter. Dump reason for fusion.
22315 (distribute_loop): Update use of partition_merge_into.
22317 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22319 * tree-loop-distribution.c (bb_top_order_index): New.
22320 (bb_top_order_index_size, bb_top_order_cmp): New.
22321 (stmts_from_loop): Use topological order.
22322 (pass_loop_distribution::execute): Compute and release topological
22323 order for basic blocks.
22325 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22327 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
22330 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22332 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
22333 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
22334 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
22335 * internal-fn.def (LOOP_DIST_ALIAS): New.
22336 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
22337 (fold_loop_internal_call): ... this.
22338 (vect_loop_dist_alias_call): New function.
22339 (set_uid_loop_bbs): Call fold_loop_internal_call.
22340 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
22343 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22346 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
22347 Require dead FLAGS_REG at the beginning of a peephole.
22349 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22352 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
22353 arguments in the call to __builtin_ia32_sbb_u32.
22354 (_subborrow_u64): Swap _X and _Y arguments in the call to
22355 __builtin_ia32_sbb_u64.
22357 2017-07-04 Jakub Jelinek <jakub@redhat.com>
22360 * tree-vrp.c (compare_assert_loc): Turn into a function template
22361 with stable template parameter. Only test if a->e is NULL,
22362 !a->e == !b->e has been verified already. Use e == NULL or
22363 e != NULL instead of e or ! e tests. If stable is true, don't use
22364 iterative_hash_expr, on the other side allow a or b or both NULL
22365 and sort the NULLs last.
22366 (process_assert_insertions): Sort using compare_assert_loc<false>
22367 instead of compare_assert_loc, later sort using
22368 compare_assert_loc<true> before calling process_assert_insertions_for
22369 in a loop. Use break instead of continue once seen NULL pointer.
22371 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22373 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
22374 Cortex-R7 and Cortex-R8 processors.
22376 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22378 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
22379 uninitialized while src is not.
22381 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
22383 * common/config/arm/arm-common.c: Adjust include path for
22385 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
22386 (arm-cpu.h): Create in build directory. Adjust dependency rules.
22387 (arm-cpu-data.h): Likewise.
22388 (arm-cpu-cdata.h): Likewise.
22389 * config/arm/arm-cpu.h: Delete.
22390 * config/arm/arm-cpu-cdata.h: Delete.
22391 * config/arm/arm-cpu-data.h: Delete.
22393 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
22395 * config/arm/arm-cpus.in (cortex-a55): New.
22396 (cortex-a75): Likewise.
22397 (cortex-a75.cortex-a55): Likewise.
22398 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
22400 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
22401 * config/arm/arm-cpu-cdata.h: Regenerate.
22402 * config/arm/arm-cpu-data.h: Regenerate.
22403 * config/arm/arm-cpu.h: Regenerate.
22404 * config/arm/arm-tables.opt: Regenerate.
22405 * config/arm/arm-tune.md: Regenerate.
22407 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22409 * haifa-sched.c (sched_create_recovery_edges): Update profile.
22411 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22413 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
22416 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
22418 PR tree-optimization/81292
22419 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
22420 full_string_p, also call adjust_related_strinfos if the adjustment
22421 is simple, otherwise invalidate related strinfos.
22423 2017-07-04 Martin Liska <mliska@suse.cz>
22426 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
22427 newly created variable as DECL_IGNORED_P.
22429 2017-07-04 Martin Liska <mliska@suse.cz>
22432 * ipa-inline.c (inline_small_functions):
22433 Use xstrdup_for_dump.
22435 2017-07-04 Tom de Vries <tom@codesourcery.com>
22437 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
22439 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22442 * config/darwin.c (darwin_function_switched_text_sections):
22445 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
22447 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
22449 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22451 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
22453 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22455 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
22456 min_profitable_iters, and th as inclusive lower bounds.
22457 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
22458 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
22459 for min_profitable_iters and min_profitable_estimate.
22460 (vect_transform_loop): Treat th as an inclusive lower bound.
22461 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22463 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22466 * config/darwin.c (darwin_function_switched_text_sections):
22467 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
22468 in two pieces, and suppress the use of buf.
22470 2017-07-03 Nathan Sidwell <nathan@acm.org>
22472 * hash-table.h (hash_table_mod1): Fix indentation.
22474 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22476 PR middle-end/81290
22477 * predict.c (force_edge_cold): Be more careful about propagation
22479 * profile-count.h (profile_probability::guessed,
22480 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
22482 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
22484 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
22486 * doc/invoke.texi (rcpc architecture extension): Document it.
22488 2017-07-03 Richard Biener <rguenther@suse.de>
22490 PR tree-optimization/60510
22491 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
22492 the scalar reduction PHI and use it.
22493 (vectorizable_reduction): Properly guard the single_defuse_cycle
22494 path for non-SLP reduction chains where we cannot use it.
22495 Rework reduc_def/index and vector type deduction. Rework
22496 vector operand gathering during reduction op code-gen.
22497 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
22498 chains dissolve the chain and leave it to non-SLP reduction
22501 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22503 * tree-data-ref.h (dr_alignment): Declare.
22504 * tree-data-ref.c (dr_alignment): New function.
22505 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
22506 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
22508 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
22510 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22512 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
22513 and base_misalignment fields.
22514 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
22515 * tree-data-ref.c: Include builtins.h.
22516 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
22517 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
22518 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
22519 * tree-vect-data-refs.c: Include tree-cfg.h.
22520 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
22521 fields instead of calculating an alignment here.
22522 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
22523 innermost_loop_behavior fields.
22525 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22527 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
22529 (DR_STEP_ALIGNMENT): New macro.
22530 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
22531 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
22532 (create_data_ref): Print it.
22533 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
22534 to tell whether the step preserves vector (mis)alignment.
22535 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22536 Move the check for an integer step and generalise to all INTEGER_CST.
22537 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
22538 Print the outer step alignment.
22540 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22542 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
22543 with offset_alignment.
22544 (DR_ALIGNED_TO): Delete.
22545 (DR_OFFSET_ALIGNMENT): New macro.
22546 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
22547 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
22548 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
22549 (create_data_ref): Likewise.
22550 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22551 (vect_analyze_data_refs): Likewise.
22552 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
22553 creating dummy innermost behavior.
22555 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22557 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
22558 with a "innermost_loop_behavior *" and refeence tree.
22559 * tree-data-ref.c (dr_analyze_innermost): Likewise.
22560 (create_data_ref): Update call accordingly.
22561 * tree-predcom.c (find_looparound_phi): Likewise.
22563 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22565 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
22566 fields with dr_wrt_vec_loop.
22567 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
22568 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
22569 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
22570 (vect_dr_behavior): New function.
22571 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22572 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
22573 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
22574 track whether the step preserves the misalignment.
22575 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22576 Use vect_dr_behavior.
22577 (vect_setup_realignment): Update call accordingly.
22578 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
22579 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
22580 call to vect_create_addr_base_for_vector_ref.
22581 (vect_create_cond_for_align_checks): Likewise.
22582 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
22583 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
22584 (vect_recog_mask_conversion_pattern): Likewise.
22585 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
22586 (new_stmt_vec_info): Remove redundant zeroing.
22588 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22590 * common/config/arm/arm-common.c (arm_be8_option): New function.
22591 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
22592 (ISA_ARMv6): Add isa_bit_be8.
22593 * config/arm/arm.h (arm_be8_option): Add prototype.
22594 (BE8_SPEC_FUNCTION): New define.
22595 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
22596 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
22597 (mlittle-endian): Similarly.
22598 (mbe8, mbe32): New options.
22599 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
22600 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
22602 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22604 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
22606 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22608 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
22609 (cleanup_tree_cfg_bb): Use it.
22610 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
22612 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
22614 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22617 * loop-doloop.c (add_test): Update profile.
22619 2017-07-03 Martin Liska <mliska@suse.cz>
22622 * sanopt.c (rewrite_usage_of_param): New function.
22623 (sanitize_rewrite_addressable_params): Likewise.
22624 (pass_sanopt::execute): Call rewrite_usage_of_param.
22626 2017-07-03 Richard Biener <rguenther@suse.de>
22628 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
22629 back to using VIEW_CONVERT_EXPR.
22631 2017-07-03 Martin Liska <mliska@suse.cz>
22634 * doc/extend.texi: Document when a resolver function is
22635 generated for target_clones.
22637 2017-07-03 Martin Liska <mliska@suse.cz>
22639 * asan.c (asan_emit_stack_protection): Unpoison just red zones
22640 and shadow memory of auto variables which are subject of
22641 use-after-scope sanitization.
22642 (asan_expand_mark_ifn): Add do set only when is_poison.
22644 2016-07-03 Richard Biener <rguenther@suse.de>
22646 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
22648 (vect_force_simple_reduction): Record reduction def -> phi mapping.
22649 (vectorizable_reduction): Perform reduction PHI creation when
22650 visiting a reduction PHI and adjust and simplify code generation
22651 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
22652 (vect_transform_loop): Visit reduction PHIs.
22653 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
22654 defs into the SLP tree.
22655 (vect_build_slp_tree): Reduction defs terminate the recursion.
22656 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
22658 (vect_get_vec_defs_for_stmt_copy): Export.
22659 (vect_get_vec_defs): Likewise.
22660 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
22662 (vect_get_vec_defs_for_stmt_copy): Declare.
22663 (vect_get_vec_defs): Likewise.
22665 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22667 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
22668 parameter with a "loop" parameter and use it instead of the
22669 loop containing DR_STMT. Don't check simple_iv when doing
22670 BB analysis. Describe the two analysis modes in the comment.
22672 2017-07-03 Tom de Vries <tom@codesourcery.com>
22674 PR tree-optimization/69468
22675 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
22676 (find_same_succ_bb): Handle ignore_edge_flags.
22678 2017-07-03 Tom de Vries <tom@codesourcery.com>
22680 PR tree-optimization/81192
22681 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
22683 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
22685 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
22687 2017-07-03 Tom de Vries <tom@codesourcery.com>
22689 PR tree-optimization/81192
22690 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
22691 BB_SAME_SUCC (bb) == NULL.
22693 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22695 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
22698 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22700 * dumpfile.c: Include profile-count.h
22701 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
22703 (insert_cond_bb): Update profile.
22704 * tree-cfg.h (insert_cond_bb): Update prototype.
22705 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
22706 * tree-dump.c: Do not include tree-cfg.
22708 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22710 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
22712 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22714 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
22717 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22719 * tree-complex.c (expand_complex_div_wide): update profile.
22721 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22722 Alan Hayward <alan.hayward@arm.com>
22723 David Sherwood <david.sherwood@arm.com>
22725 * Makefile.in (MACHMODE_H): Remove insn-modes.h
22726 (CORETYPES_H): New define.
22727 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
22728 (insn-modes-inline.h, s-modes-inline-h): New rules.
22729 (generated_files): Add insn-modes-inline.h.
22730 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
22731 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
22732 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
22733 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
22734 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
22735 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
22736 (build/gencodes.o, build/genconditions.o): Likewise.
22737 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
22738 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
22739 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
22740 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
22741 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
22742 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
22743 * coretypes.h: Include everything up to real.h for generators.
22744 Include insn-modes.h first. Include wide-int-print.h after
22745 wide-int.h. Include insn-modes-inline.h and then machmode.h.
22746 * machmode.h: Don't include insn-modes.h here.
22747 * function-tests.c: Remove includes of signop.h, machmode.h,
22748 double-int.h and wide-int.h.
22750 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
22752 * optc-save-gen.awk: Likewise.
22753 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
22754 * godump.c: Remove include of wide-int-print.h.
22755 * pretty-print.h: Likewise.
22756 * wide-int-print.cc: Likewise.
22757 * wide-int.cc: Likewise.
22758 * hash-map-tests.c: Remove include of signop.h.
22759 * hash-set-tests.c: Likewise.
22760 * rtl-tests.c: Likewise.
22761 * mkconfig.sh: Remove include of machmode.h.
22762 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
22764 (emit_insn_modes_inline_h): ...this new function. Emit the code
22765 into an insn-modes-inline.h header file, adding appropriate
22766 include guards and end comments.
22767 (emit_insn_modes_c_header): Remove include of machmode.h.
22768 (emit_min_insn_modes_c_header): Include coretypes.h rather than
22770 (main): Handle -i flag and call emit_insn_modes_inline_h when
22773 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22775 * tree-ssa-strlen.c (strinfo): Rename the length field to
22776 nonzero_chars. Add a full_string_p field.
22777 (compare_nonzero_chars, zero_length_string_p): New functions.
22778 (get_addr_stridx): Add an offset_out parameter.
22779 Use compare_nonzero_chars.
22780 (get_stridx): Update accordingly. Use compare_nonzero_chars.
22781 (new_strinfo): Update after above changes to strinfo.
22782 (set_endptr_and_length): Set full_string_p.
22783 (get_string_length): Update after above changes to strinfo.
22784 (unshare_strinfo): Update call to new_strinfo.
22785 (maybe_invalidate): Likewise.
22786 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
22787 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
22788 as a uhwi instead of an shwi. Update after above changes to
22789 strinfo and new_strinfo.
22790 (zero_length_string): Assert that chainsi contains full strings.
22791 Use zero_length_string_p. Update call to new_strinfo.
22792 (adjust_related_strinfos): Update after above changes to strinfo.
22793 Copy full_string_p from origsi.
22794 (adjust_last_stmt): Use zero_length_string_p.
22795 (handle_builtin_strlen): Update after above changes to strinfo and
22796 new_strinfo. Install the lhs as the string length if the previous
22797 entry didn't describe a full string.
22798 (handle_builtin_strchr): Update after above changes to strinfo
22800 (handle_builtin_strcpy): Likewise.
22801 (handle_builtin_strcat): Likewise.
22802 (handle_builtin_malloc): Likewise.
22803 (handle_pointer_plus): Likewise.
22804 (handle_builtin_memcpy): Likewise. Track nonzero characters
22805 that aren't necessarily followed by a nul terminator.
22806 (handle_char_store): Likewise.
22808 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22810 PR tree-optimization/80769
22811 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
22812 for malloc and calloc. Document the new invariant that all related
22813 strinfos have delayed lengths or none do.
22814 (verify_related_strinfos): Move earlier in file.
22815 (set_endptr_and_length): New function, split out from...
22816 (get_string_length): ...here. Also set the lengths of related
22818 (zero_length_string): Assert that chainsi has known (rather than
22820 (adjust_related_strinfos): Likewise.
22822 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22824 PR tree-optimization/81136
22825 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
22826 assert that two references with the same misalignment have the same
22827 compile-time misalignment if those compile-time misalignments
22830 2017-07-01 Andi Kleen <ak@linux.intel.com>
22832 * print-tree.c (print_node): Print all attributes.
22834 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22836 * cfg.c (scale_bbs_frequencies): New function.
22837 * cfg.h (scale_bbs_frequencies): Declare it.
22838 * cfgloopanal.c (single_likely_exit): Cleanup.
22839 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
22841 (scale_loop_profile): Likewise.
22842 (loop_version): Likewise.
22843 (create_empty_loop_on_edge): Update.
22844 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
22845 scale_loop_frequencies, scale_loop_profile, loopify,
22846 loop_version): Update prototypes.
22847 * modulo-sched.c (sms_schedule): Update.
22848 * predict.c (unlikely_executed_edge_p): Also check probability.
22849 (probably_never_executed_edge_p): Fix typo.
22850 * tree-if-conv.c (version_loop_for_if_conversion): Update.
22851 * tree-parloops.c (gen_parallel_loop): Update.
22852 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
22853 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22854 * tree-ssa-loop-split.c (split_loop): Update.
22855 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
22856 * tree-vect-loop-manip.c (vect_do_peeling): Update.
22857 (vect_loop_versioning): Update.
22858 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22860 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22862 * trans-mem.c (split_bb_make_tm_edge): Update profile.
22864 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22866 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
22867 to keep profile consistent.
22869 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22871 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
22872 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
22873 * profile-count.h (max_safe_multiplier): Make unsigned.
22874 (profile_count::guessed_zero): New.
22876 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22878 * bb-reorder.c (fix_up_crossing_landing_pad,
22879 fix_crossing_conditional_branches): Use make_single_succ_edge
22880 to keep profile consistent.
22882 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22884 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
22887 2017-07-01 Jakub Jelinek <jakub@redhat.com>
22890 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
22891 the right scopes, make sure cond_jump isn't preserved between multiple
22892 iterations. Search for fallthru edge whenever there are 3+ edges and
22893 use find_fallthru_edge for it.
22895 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22897 Patch by Alexander Monakov <amonakov@ispras.ru>
22898 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
22899 probabilities consistently.
22901 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22903 * pa.c (pa_expand_compare_and_swap_loop): Update call of
22904 emit_cmp_and_jump_insns.
22906 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22909 * tree-inline.c (expand_call_inline): Combine profile statuses.
22911 2017-06-30 Andrew Pinski <apinski@cavium.com>
22913 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
22914 fold_stmt returned true.
22916 2017-06-30 Nathan Sidwell <nathan@acm.org>
22918 * ggc.h (empty_string): Delete.
22919 * cfgexpand.c (expand_asm_stmt): Use plain "".
22920 * optabs.c (expand_asm_memory_barrier): Likewise.
22921 * stringpool.c (empty_string): Delete.
22922 (digit_vector, digit_string): Delete.
22923 (ggc_alloc_string): Use plain "", don't optimize single digit
22924 strings. Use ggc_alloc_atomic.
22926 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
22928 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
22929 comparison set and one other set, use the cost of the non-comparison
22932 2017-06-30 Nathan Sidwell <nathan@acm.org>
22934 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
22937 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
22939 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
22940 loops. Remove now unneeded calls to gimple_switch_set_label() that
22941 just set removed labels to NULL_TREE.
22943 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
22945 * tree-ssanames.c (set_range_info_raw): Abstract from ...
22946 (set_range_info): ...here. Only call set_range_info_raw if domain
22948 (set_nonzero_bits): Call set_range_info_raw.
22949 * tree-ssanames.h (set_range_info_raw): New.
22951 2017-06-30 Jakub Jelinek <jakub@redhat.com>
22954 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
22955 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
22956 of nonimmediate_operand and <store_mask_constraint> instead of m
22957 for the input operand. For V8FI iterator, always split if input
22958 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
22959 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
22960 <store_mask_predicate> instead of register_operand and
22961 <store_mask_constraint> instead of v for the input operand. Make
22962 sure both operands aren't MEMs for if not <mask_applied>.
22964 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
22966 * lto-wrapper.c (copy_file) Close both file descriptors before
22969 2017-06-30 Martin Liska <mliska@suse.cz>
22972 * multiple_target.c (create_dispatcher_calls): Make ifunc
22973 also for function that don't have calls or are not referenced.
22975 2017-06-30 Richard Biener <rguenther@suse.de>
22977 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
22978 analyze the first scalar stmt. Move vector type computation
22979 for the BB case here from ...
22980 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
22981 live operation processing in the SLP case properly.
22983 2017-06-30 Richard Biener <rguenther@suse.de>
22985 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
22987 2017-06-30 Martin Liska <mliska@suse.cz>
22990 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
22991 before BUILT_IN_UNWIND_RESUME when ASAN is used.
22993 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
22995 * doc/invoke.texi (AArch64): Add missing options and remove redundant
22998 2017-06-30 Richard Biener <rguenther@suse.de>
23000 PR tree-optimization/81249
23001 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
23002 condition reduction result to original scalar type.
23004 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23006 * profile-count.h (enum profile_quality): Fix typos and whitespace
23009 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23011 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
23012 type for branch probabilities.
23014 2017-06-29 Julian Brown <julian@codesourcery.com>
23015 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23017 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
23018 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
23019 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
23020 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
23022 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23024 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
23025 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
23026 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
23027 CC usage from generic code to here.
23028 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
23029 CC usage into the target macros.
23031 2017-06-29 Maya Rashish <coypu@sdf.org>
23033 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
23036 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23038 * arm/arm-builtins.c: Include profile-count.h
23039 * except.c (sjlj_emit_function_enter): Use
23040 profile_probability::unlikely.
23042 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23044 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
23045 and tocrel_offset be pointer args rather than implicitly using
23047 (legitimate_constant_pool_address_p, rs6000_emit_move,
23048 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
23049 tocrel_offset and use in toc_relative_expr_p call.
23050 (print_operand, print_operand_address): Use static tocrel_base_oac
23051 and tocrel_offset_oac.
23052 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
23055 2017-06-29 Maya Rashish <coypu@sdf.org>
23057 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
23059 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
23061 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
23062 objects, take into account only the alignment of 'op0' and 'mode1' if
23065 2017-06-29 Steve Ellcey <sellcey@cavium.com>
23067 * ccmp.c (ccmp_tree_comparison_p): New function.
23068 (ccmp_candidate_p): Update to use above function.
23069 (get_compare_parts): New function.
23070 (expand_ccmp_next): Update to use new functions.
23071 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
23073 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
23074 take mode as argument.
23075 * ccmp.h (expand_ccmp_expr): Add mode as argument.
23076 * expr.c (expand_expr_real_1): Pass mode as argument.
23078 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
23080 * combine.c (combine_instructions): Print insns to dump_file, together
23083 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23085 * asan.c (asan_emit_stack_protection): Update.
23086 (create_cond_insert_point): Update.
23087 * auto-profile.c (afdo_propagate_circuit): Update.
23088 * basic-block.h (struct edge_def): Turn probability to
23089 profile_probability.
23090 (EDGE_FREQUENCY): Update.
23091 * bb-reorder.c (find_traces_1_round): Update.
23092 (better_edge_p): Update.
23093 (sanitize_hot_paths): Update.
23094 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
23095 (make_single_succ_edge): Update.
23096 (check_bb_profile): Update.
23097 (dump_edge_info): Update.
23098 (update_bb_profile_for_threading): Update.
23099 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
23100 probabilitycount to 0.
23101 * cfgbuild.c (compute_outgoing_frequencies): Update.
23102 * cfgcleanup.c (try_forward_edges): Update.
23103 (outgoing_edges_match): Update.
23104 (try_crossjump_to_edge): Update.
23105 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
23106 (expand_gimple_tailcall): Update.
23107 (construct_init_block): Use make_single_succ_edge.
23108 (construct_exit_block): Use make_single_succ_edge.
23109 * cfghooks.c (verify_flow_info): Update.
23110 (redirect_edge_succ_nodup): Update.
23111 (split_edge): Update.
23112 (account_profile_record): Update.
23113 * cfgloopanal.c (single_likely_exit): Update.
23114 * cfgloopmanip.c (scale_loop_profile): Update.
23115 (set_zero_probability): Remove.
23116 (duplicate_loop_to_header_edge): Update.
23117 * cfgloopmanip.h (loop_version): Update prototype.
23118 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
23119 (force_nonfallthru_and_redirect): Update.
23120 (update_br_prob_note): Update.
23121 (rtl_verify_edges): Update.
23122 (purge_dead_edges): Update.
23123 (rtl_lv_add_condition_to_bb): Update.
23124 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
23125 * cgraphunit.c (init_lowered_empty_function): Update.
23126 (cgraph_node::expand_thunk): Update.
23127 * cilk-common.c: Include profile-count.h
23128 * dojump.c (inv): Remove.
23129 (jumpifnot): Update.
23130 (jumpifnot_1): Update.
23131 (do_jump_1): Update.
23133 (do_jump_by_parts_greater_rtx): Update.
23134 (do_compare_rtx_and_jump): Update.
23135 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
23136 do_jump_1. do_compare_rtx_and_jump): Update prototype.
23137 * dwarf2cfi.c: Include profile-count.h
23138 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
23139 (sjlj_emit_dispatch_table): Likewise.
23140 * explow.c: Include profile-count.h
23141 * expmed.c (emit_store_flag_force): Update.
23142 (do_cmp_and_jump): Update.
23143 * expr.c (compare_by_pieces_d::generate): Update.
23144 (compare_by_pieces_d::finish_mode): Update.
23145 (emit_block_move_via_loop): Update.
23146 (store_expr_with_bounds): Update.
23147 (store_constructor): Update.
23148 (expand_expr_real_2): Update.
23149 (expand_expr_real_1): Update.
23150 * expr.h (try_casesi, try_tablejump): Update prototypes.
23151 * gimple-pretty-print.c (dump_probability): Update.
23152 (dump_profile): New.
23153 (dump_gimple_label): Update.
23154 (dump_gimple_bb_header): Update.
23155 * graph.c (draw_cfg_node_succ_edges): Update.
23156 * hsa-gen.c (convert_switch_statements): Update.
23157 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
23158 (find_if_case_1): Update.
23159 (find_if_case_2): Update.
23160 * internal-fn.c (expand_arith_overflow_result_store): Update.
23161 (expand_addsub_overflow): Update.
23162 (expand_neg_overflow): Update.
23163 (expand_mul_overflow): Update.
23164 (expand_vector_ubsan_overflow): Update.
23165 * ipa-cp.c (good_cloning_opportunity_p): Update.
23166 * ipa-split.c (split_function): Use make_single_succ_edge.
23167 * ipa-utils.c (ipa_merge_profiles): Update.
23168 * loop-doloop.c (add_test): Update.
23169 (doloop_modify): Update.
23170 * loop-unroll.c (compare_and_jump_seq): Update.
23171 (unroll_loop_runtime_iterations): Update.
23172 * lra-constraints.c (lra_inheritance): Update.
23173 * lto-streamer-in.c (input_cfg): Update.
23174 * lto-streamer-out.c (output_cfg): Update.
23175 * mcf.c (adjust_cfg_counts): Update.
23176 * modulo-sched.c (sms_schedule): Update.
23177 * omp-expand.c (expand_omp_for_init_counts): Update.
23178 (extract_omp_for_update_vars): Update.
23179 (expand_omp_ordered_sink): Update.
23180 (expand_omp_for_ordered_loops): Update.
23181 (expand_omp_for_generic): Update.
23182 (expand_omp_for_static_nochunk): Update.
23183 (expand_omp_for_static_chunk): Update.
23184 (expand_cilk_for): Update.
23185 (expand_omp_simd): Update.
23186 (expand_omp_taskloop_for_outer): Update.
23187 (expand_omp_taskloop_for_inner): Update.
23188 * omp-simd-clone.c (simd_clone_adjust): Update.
23189 * optabs.c (expand_doubleword_shift): Update.
23190 (expand_abs): Update.
23191 (emit_cmp_and_jump_insn_1): Update.
23192 (expand_compare_and_swap_loop): Update.
23193 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
23194 * predict.c (predictable_edge_p): Update.
23195 (edge_probability_reliable_p): Update.
23196 (set_even_probabilities): Update.
23197 (combine_predictions_for_insn): Update.
23198 (combine_predictions_for_bb): Update.
23199 (propagate_freq): Update.
23200 (estimate_bb_frequencies): Update.
23201 (force_edge_cold): Update.
23202 * profile-count.c (profile_count::dump): Add missing space into dump.
23203 (profile_count::debug): Add newline.
23204 (profile_count::differs_from_p): Explicitly convert to unsigned.
23205 (profile_count::stream_in): Update.
23206 (profile_probability::dump): New member function.
23207 (profile_probability::debug): New member function.
23208 (profile_probability::differs_from_p): New member function.
23209 (profile_probability::differs_lot_from_p): New member function.
23210 (profile_probability::stream_in): New member function.
23211 (profile_probability::stream_out): New member function.
23212 * profile-count.h (profile_count_quality): Rename to ...
23213 (profile_quality): ... this one.
23214 (profile_probability): New.
23215 (profile_count): Update.
23216 * profile.c (compute_branch_probabilities): Update.
23217 * recog.c (peep2_attempt): Update.
23218 * sched-ebb.c (schedule_ebbs): Update.
23219 * sched-rgn.c (find_single_block_region): Update.
23220 (compute_dom_prob_ps): Update.
23221 (schedule_region): Update.
23222 * sel-sched-ir.c (compute_succs_info): Update.
23223 * stmt.c (struct case_node): Update.
23224 (do_jump_if_equal): Update.
23225 (get_outgoing_edge_probs): Update.
23226 (conditional_probability): Update.
23227 (emit_case_dispatch_table): Update.
23228 (expand_case): Update.
23229 (expand_sjlj_dispatch_table): Update.
23230 (emit_case_nodes): Update.
23231 * targhooks.c: Update.
23232 * tracer.c (better_p): Update.
23233 (find_best_successor): Update.
23234 * trans-mem.c (expand_transaction): Update.
23235 * tree-call-cdce.c: Update.
23236 * tree-cfg.c (gimple_split_edge): Upate.
23237 (move_sese_region_to_fn): Upate.
23238 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
23239 * tree-eh.c (lower_resx): Upate.
23240 (cleanup_empty_eh_move_lp): Upate.
23241 * tree-if-conv.c (version_loop_for_if_conversion): Update.
23242 * tree-inline.c (copy_edges_for_bb): Update.
23243 (copy_cfg_body): Update.
23244 * tree-parloops.c (gen_parallel_loop): Update.
23245 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
23246 (gimple_gen_time_profiler): Update.
23247 * tree-ssa-dce.c (remove_dead_stmt): Update.
23248 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
23249 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
23250 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
23251 (unloop_loops): Update.
23252 (try_peel_loop): Update.
23253 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23254 * tree-ssa-loop-split.c (connect_loops): Update.
23255 (split_loop): Update.
23256 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
23257 (hoist_guard): Update.
23258 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
23259 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
23260 (value_replacement): Update.
23261 * tree-ssa-reassoc.c (branch_fixup): Update.
23262 * tree-ssa-tail-merge.c (replace_block_by): Update.
23263 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
23264 (create_edge_and_update_destination_phis): Update.
23265 (compute_path_counts): Update.
23266 (recompute_probabilities): Update.
23267 (update_joiner_offpath_counts): Update.
23268 (freqs_to_counts_path): Update.
23269 (duplicate_thread_path): Update.
23270 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
23271 (struct switch_conv_info): Update.
23272 (gen_inbound_check): Update.
23273 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23274 (vect_do_peeling): Update.
23275 (vect_loop_versioning): Update.
23276 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23277 (optimize_mask_stores): Update.
23278 * ubsan.c (ubsan_expand_null_ifn): Update.
23279 * value-prof.c (gimple_divmod_fixed_value): Update.
23280 (gimple_divmod_fixed_value_transform): Update.
23281 (gimple_mod_pow2): Update.
23282 (gimple_mod_pow2_value_transform): Update.
23283 (gimple_mod_subtract): Update.
23284 (gimple_mod_subtract_transform): Update.
23285 (gimple_ic): Update.
23286 (gimple_stringop_fixed_value): Update.
23287 (gimple_stringops_transform): Update.
23288 * value-prof.h: Update.
23290 2017-06-29 Carl Love <cel@us.ibm.com>
23292 * config/rs6000/rs6000-c.c: Add support for built-in functions
23293 vector signed int vec_signed (vector float);
23294 vector signed long long vec_signed (vector double);
23295 vector signed int vec_signed2 (vector double, vector double);
23296 vector signed int vec_signede (vector double);
23297 vector signed int vec_signedo (vector double);
23298 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
23299 instruction generator.
23300 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23301 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
23302 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
23304 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
23305 vunsignede_v2df): Add define_expands.
23306 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
23307 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
23308 VEC_UNSIGNEDO): Add definitions.
23309 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23310 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
23311 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
23312 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
23313 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
23314 * config/rs6000/altivec.h (vec_signed, vec_signed2,
23315 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
23316 vec_unsignede, vec_unsignedo): Add builtin defines.
23317 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
23319 * doc/extend.texi: Update the built-in documentation file for the
23320 new built-in functions.
23322 2017-06-29 Richard Biener <rguenther@suse.de>
23324 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
23325 reduction chains to LOOP_VINFO_REDUCTIONS.
23326 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
23327 SLP reductions after processing reduction chains.
23329 2017-06-29 Nathan Sidwell <nathan@acm.org>
23331 * builtins.c (fold_builtin_FUNCTION): Use
23332 lang_hooks.decl_printable_name.
23334 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
23336 PR middle-end/81194
23337 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
23338 with only one label.
23339 * stmt.c (expand_case): Assert NCASES is greater than one.
23341 2017-06-29 Richard Biener <rguenther@suse.de>
23343 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
23345 (group_case_labels): Likewise.
23346 (find_taken_edge): Push sanity checking on val to workers...
23347 (find_taken_edge_cond_expr): ... here
23348 (find_taken_edge_switch_expr): ... and here, handle cases
23349 with just a default label.
23350 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
23351 (group_case_labels): Likewise.
23352 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
23353 group_case_labels does anything cleanup the CFG again.
23355 2017-06-29 Bin Cheng <bin.cheng@arm.com>
23357 PR tree-optimization/81196
23358 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
23359 exit condition comparing two IVs.
23361 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
23363 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
23364 profile to the dummy entry at the end of the list of architectures.
23365 * config/arm/arm-cpu-cdata.h: Regenerated.
23367 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23368 Michael Collison <michael.collison@arm.com>
23371 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
23373 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
23374 (*aarch64_reg_<mode>3_minus_mask): New pattern.
23375 (*aarch64_<optab>_reg_di3_mask2): New pattern.
23376 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
23377 of shift when the shift amount is masked with constant equal to
23378 the size of the mode.
23379 * config/aarch64/predicates.md (subreg_lowpart_operator): New
23382 2017-06-29 Martin Liska <mliska@suse.cz>
23384 * config/i386/i386.opt: Change range from [1,5] to [0,5].
23386 2017-06-29 Yury Gribov <tetra2005@gmail.com>
23389 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
23392 (edge_growth_cache_entry::edge_growth_cache_entry): New
23394 (reset_edge_growth_cache): Update to use constructor.
23396 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23398 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
23399 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23400 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23402 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
23404 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
23405 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
23407 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
23409 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
23410 (*-linux-uclibc*): Add t-uclibc tmake_file.
23411 * config/t-musl: New.
23412 * config/t-uclibc: New.
23414 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
23416 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
23418 (gen_comm_data): Emit architectural setting of arch_prof.
23419 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
23421 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
23422 (armv8-m.base, armv8-m.main): Likewise.
23423 * arm-protos.h (arm_build_target): Add profile field.
23424 (arch_option): Likewise.
23425 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
23427 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
23428 arm_active_target.profile.
23430 2017-06-28 Richard Biener <rguenther@suse.de>
23432 PR middle-end/81227
23433 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
23434 TYPE_OVERFLOW_WRAPS.
23435 * match.pd (negate_expr_p): Likewise.
23436 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
23437 fold_build2, not fold_binary.
23439 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23441 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
23442 Convert memory address to Pmode.
23443 (aarch64_print_operand): Assert MEM operands are always Pmode.
23445 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23448 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
23449 Remove redundant if.
23450 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
23451 * config/arm/aarch-common-protos.h
23452 (aarch_forward_to_shift_is_not_shifted_re): Remove.
23453 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
23455 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
23458 * multiple_target.c (create_dispatcher_calls): Set the default
23459 clone to be static, not public.
23461 2017-06-28 Richard Biener <rguenther@suse.de>
23463 * tree-vect-loop.c (vectorizable_reduction): Move special
23464 cond reduction IV var creation ...
23465 (vect_create_epilog_for_reduction): ... here. Remove induction_index
23466 parameter. Use STMT_VINFO_VECTYPE.
23467 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
23470 2017-06-28 Martin Liska <mliska@suse.cz>
23473 * ipa-visibility.c (non_local_p): Handle visibility.
23475 2017-06-28 Martin Liska <mliska@suse.cz>
23478 * common.opt: Add IntegerRange to various options.
23479 * opt-functions.awk (integer_range_info): New function.
23480 * optc-gen.awk: Add integer_range_info to cl_options struct.
23481 * opts-common.c (decode_cmdline_option): Handle
23482 CL_ERR_INT_RANGE_ARG.
23483 (cmdline_handle_error): Likewise.
23484 * opts.c (print_filtered_help): Show valid interval in
23485 when --help is provided.
23486 * opts.h (struct cl_option): Add range_min and range_max fields.
23487 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
23489 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
23491 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
23492 (x * C EQ/NE y * C): New transformation.
23494 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
23496 * genmultilib (combination_space): Accept '+' in option names.
23498 2017-06-28 Martin Liska <mliska@suse.cz>
23501 * asan.c (instrument_derefs): Bail out inner references
23502 that are hard register variables.
23504 2017-06-28 Jakub Jelinek <jakub@redhat.com>
23507 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
23508 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
23510 2017-06-28 Richard Biener <rguenther@suse.de>
23512 * tree-vectorizer.h (vect_get_vec_defs): Remove.
23513 (vect_get_slp_defs): Adjust.
23514 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
23516 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
23518 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
23519 get_initial_defs_for_reduction instead of vect_get_vec_defs.
23520 (vectorizable_reduction): Adjust.
23521 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
23523 (vect_get_slp_defs): Likewise.
23524 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
23525 (vectorizable_bswap): Adjust.
23526 (vectorizable_call): Likewise.
23527 (vectorizable_conversion): Likewise.
23528 (vectorizable_assignment): Likewise.
23529 (vectorizable_shift): Likewise.
23530 (vectorizable_operation): Likewise.
23531 (vectorizable_store): Likewise.
23532 (vectorizable_condition): Likewise.
23533 (vectorizable_comparison): Likewise.
23535 2017-06-28 Michael Collison <michael.collison@arm.com>
23538 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
23540 (arm_gen_movmemqi): Removed unused variable 'i'.
23541 Convert 'for' loop into 'while' loop.
23542 (arm_expand_prologue): Remove last unnecessary set of insn.
23543 (thumb_pop): Remove unused variable 'pushed_words'.
23544 (thumb_exit): Remove last unnecessary set of regs_to_pop.
23546 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23548 * config/s390/predicates.md: Use s390_rel_address_ok_p.
23549 * config/s390/s390-protos.h: Add prototype of
23550 s390_rel_address_ok_p.
23551 * config/s390/s390.c (s390_got_symbol): New function.
23552 (s390_rel_address_ok_p): New function.
23553 (legitimize_pic_address): Use s390_rel_address_ok_p.
23554 (s390_load_got): Use s390_got_symbol.
23555 (s390_option_override): Issue error if
23556 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
23557 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
23559 * config/s390/s390.opt: New option mpic-data-is-text-relative.
23561 2017-06-27 Andrew Pinski <apinski@cavium.com>
23563 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
23564 (X * copysign (1.0, X)): New pattern.
23565 (X * copysign (1.0, -X)): New pattern.
23566 (copysign (-1.0, CST)): New pattern.
23568 2017-06-27 Joseph Myers <joseph@codesourcery.com>
23570 * genmultilib (combination_space): Remove variable.
23571 Validate reuse rules against regular expression for any sequence
23572 of multilib options in any order.
23574 2017-06-27 Michael Collison <michael.collison@arm.com>
23576 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
23577 call aarch64_split_simd_combine.
23578 * (aarch64_combine_internal<mode>): Delete pattern.
23579 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
23580 Allow register and subreg operands.
23582 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23584 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
23585 specific need, just fallback on defaults.
23586 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
23588 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23589 Olivier Hainque <hainque@adacore.com>
23591 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
23593 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
23594 targets. Pick a default if no particular attempt applied.
23595 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
23598 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23600 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
23601 (x86_64-wrs-vxworks7): Likewise.
23603 2017-06-27 Marek Polacek <polacek@redhat.com>
23606 * ubsan.c (instrument_null): Check get_base_address's result for null.
23608 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
23610 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
23612 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
23614 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
23615 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
23616 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
23617 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
23618 New function types.
23619 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
23620 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
23621 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
23622 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
23623 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
23624 BUILT_IN_FEUPDATEENV): New builtins.
23625 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
23626 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
23627 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
23628 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
23630 (builtin_structptr_types): Adjust size.
23631 * tree.c (builtin_structptr_types): Add four entries.
23633 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23634 Olivier Hainque <hainque@adacore.com>
23636 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
23637 (TLS_SYM): New local macro, forcing reference to __tls__ on
23638 link command lines for VxWorks 7 RTPs, triggering initialization
23640 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
23641 OS features TLS support, true for RTPs on VxWorks 7.
23642 * config/vxworks.c (vxworks_override_options): Setup emutls
23645 2017-06-27 Jakub Jelinek <jakub@redhat.com>
23647 * predict.c (test_prediction_value_range): Use -1U instead of -1
23648 to avoid narrowing conversion warning.
23649 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
23650 to avoid narrowing conversion warning.
23651 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
23653 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
23655 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23657 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
23658 64bit configurations.
23659 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
23660 (SIZE_TYPE): Likewise.
23661 * config/vxworks.c (vxworks_emutls_var_fields): Use
23662 long_unsigned_type_node instead of unsigned_type_node as the offset
23663 field type, which is "pointer" mode in emutls.c.
23665 2017-06-27 Jakub Jelinek <jakub@redhat.com>
23668 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
23670 PR middle-end/81207
23671 * gimple-fold.c (replace_call_with_call_and_fold): Handle
23672 gimple_vuse copying separately from gimple_vdef copying.
23674 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23676 * value-prof.c (free_hist): Remove call to memset and the enclosing if
23679 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
23680 Olivier Hainque <hainque@adacore.com>
23682 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
23683 for all vxworks7 targets.
23684 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
23685 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
23686 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
23687 variations for VX6/VX7 and 32/64bits later on in ...
23688 (VXWORKS_LIB_SPEC): Leverage new macros.
23689 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
23690 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
23692 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
23694 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
23695 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
23697 2017-06-26 Carl Love <cel@us.ibm.com>
23699 * config/rs6000/rs6000-c.c: Add support for built-in functions
23700 vector bool char vec_reve (vector bool char);
23701 vector signed char vec_reve (vector signed char);
23702 vector unsigned char vec_reve (vector unsigned char);
23703 vector bool int vec_reve (vector bool int);
23704 vector signed int vec_reve (vector signed int);
23705 vector unsigned int vec_reve (vector unsigned int);
23706 vector bool long long vec_reve (vector bool long long);
23707 vector signed long long vec_reve (vector signed long long);
23708 vector unsigned long long vec_reve (vector unsigned long long);
23709 vector bool short vec_reve (vector bool short);
23710 vector signed short vec_reve (vector signed short);
23711 vector double vec_reve (vector double);
23712 vector float vec_reve (vector float);
23713 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
23714 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
23715 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
23716 (altivec_vreve): New pattern.
23717 * config/rs6000/altivec.h (vec_reve): New define.
23718 * doc/extend.texi (vec_rev): Update the built-in documentation file
23719 for the new built-in functions.
23721 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23723 PR tree-optimization/71815
23724 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
23726 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
23728 (slsr_process_phi): Likewise.
23729 (replace_uncond_cands_and_profitable_phis): Don't replace a
23730 multiply candidate with a stride of 1 (copy or cast).
23731 (phi_incr_cost): Call uses_consumed_by_stmt rather than
23733 (lowest_cost_path): Likewise.
23734 (total_savings): Likewise.
23736 2017-06-26 Richard Biener <rguenther@suse.de>
23739 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
23740 Use def_builtin_pure for all gather builtins.
23742 2017-06-26 Richard Biener <rguenther@suse.de>
23744 PR tree-optimization/81203
23745 * tree-tailcall.c (find_tail_calls): Do not move stmts into
23746 non-dominating BBs.
23748 2017-06-26 Marek Polacek <polacek@redhat.com>
23751 * doc/invoke.texi: Document -Wmultistatement-macros.
23753 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
23755 * doc/sourcebuild.texi (ARM-specific attributes): Document new
23756 arm_neon_ok_no_float_abi effective target.
23758 2017-06-26 Richard Biener <rguenther@suse.de>
23760 PR tree-optimization/80928
23761 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
23762 (copy_bbs): Set BB_DUPLICATED flag early.
23763 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
23765 (execute_on_shrinking_pred): Likewise.
23766 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
23767 BB_DUPLICATED blocks.
23768 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
23769 iterate over all PHIs considering removal of *gsi.
23771 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
23773 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
23776 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23778 * config/rs6000/rs6000-string.c: (expand_block_clear,
23779 do_load_for_compare, select_block_compare_mode,
23780 compute_current_alignment, expand_block_compare,
23781 expand_strncmp_align_check, expand_strn_compare,
23782 expand_block_move, rs6000_output_load_multiple)
23783 Move functions related to string/block move/compare
23784 to a separate file.
23785 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
23786 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
23787 for this function which is now used in two files.
23788 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
23789 * config.gcc: Add rs6000-string.o to extra_objs for
23790 targets powerpc*-*-* and rs6000*-*-*.
23792 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
23795 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
23796 32-bit, since indexed is not valid for DImode.
23797 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
23798 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
23799 (define_peephole2 for Altivec d-form load): Add 32-bit support.
23800 (define_peephole2 for Altivec d-form store): Likewise.
23803 * multiple_target.c (create_dispatcher_calls): Only create the
23804 dispatcher call if the function is the default clone of a
23805 versioned function.
23807 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
23809 PR middle-end/80902
23810 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
23811 a call, force the call to not be a tail call.
23813 2017-06-23 Jeff Law <law@redhat.com>
23815 * doc/contrib.texi: Add entry for Steven Pemberton's work on
23818 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
23820 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
23821 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
23822 handling for early expansion of vector shifts (sl,sr,sra,rl).
23823 (builtin_function_type): Add vector shift right instructions
23824 to the unsigned argument list.
23826 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
23828 rtl-optimizatoin/79286
23829 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
23830 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
23831 trap. PIC register plus a const unspec without offset can never trap.
23833 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
23835 * tree.h (builtin_structptr_type): New type.
23836 (builtin_structptr_types): Declare new array.
23837 * tree.c (builtin_structptr_types): New array.
23838 (free_lang_data, build_common_tree_nodes): Use it.
23840 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
23843 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
23846 2017-06-22 Matt Turner <mattst88@gmail.com>
23848 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
23849 Lake models to skylake case. Assume skylake for unknown
23850 models with clflushopt.
23852 2017-06-22 Jeff Law <law@redhat.com>
23854 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
23855 frame sizes that do not satisfy aarch64_uimm12_shift.
23857 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
23859 * profile-count.h (apply_probability,
23860 apply_scale, probability_in): Fix checks for zero.
23862 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23864 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
23865 * doc/cppdiropts.texi (-I @var{dir}): Document it.
23867 2016-06-22 Richard Biener <rguenther@suse.de>
23869 * tree-vect-loop.c (vect_model_reduction_cost): Handle
23870 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
23871 REDUC_MAX_EXPR support.
23872 (vectorizable_reduction): Likewise.
23873 (vect_create_epilog_for_reduction): Likewise.
23875 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
23877 * match.pd (A / (1 << B) -> A >> B): New.
23878 * generic-match-head.c: Include optabs-tree.h.
23879 * gimple-match-head.c: Likewise.
23880 * optabs-tree.h (target_supports_op_p): New.
23881 * optabs-tree.c (target_supports_op_p): New.
23883 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23885 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
23886 $gcc_cv_ld --help output.
23887 (gcc_cv_ld_demangle): Likewise.
23888 (gcc_cv_ld_eh_frame_hdr): Likewise.
23889 (gcc_cv_ld_pie): Likewise.
23890 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
23891 (gcc_cv_ld_buildid): Likewise.
23892 (gcc_cv_ld_sysroot): Likewise.
23893 (ld_bndplt_support): Likewise.
23894 (ld_pushpopstate_support): Likewise.
23895 * configure: Regenerate.
23896 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
23898 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23901 * config/i386/sse.md (round<mode>2): Renumber match_dup and
23902 operands indexes to avoid gap between operands and match_dups.
23904 2017-06-21 Andrew Pinski <apinski@cavium.com>
23906 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
23907 Increment Arith_shift and Arith_shift_reg by 1.
23908 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
23910 * config/aarch64/aarch64.c (thunderx_tunings): Enable
23911 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
23912 (aarch64_strip_extend): Add new argument and test for it.
23913 (aarch64_cheap_mult_shift_p): New function.
23914 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
23915 add a cost if it is true.
23916 Update calls to aarch64_strip_extend.
23917 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
23919 2017-06-21 Andrew Pinski <apinski@cavium.com>
23921 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
23923 (thunderxt88): Likewise.
23924 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
23925 (thunderx_prefetch_tune): New variable.
23926 (thunderx2t99_prefetch_tune): Update for the correct values.
23927 (thunderxt88_tunings): New variable.
23928 (thunderx_tunings): Use thunderx_prefetch_tune instead of
23929 generic_prefetch_tune.
23930 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
23932 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23934 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
23935 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
23936 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
23937 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
23938 (aarch64_atomic_cas<mode>, GPI): Likewise.
23940 2017-06-21 Martin Liska <mliska@suse.cz>
23942 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
23943 statements on cold and hot labels.
23944 * predict.c (tree_estimate_probability_bb): Remove the
23945 prediction from this place.
23947 2017-06-21 Martin Liska <mliska@suse.cz>
23949 PR tree-optimization/79489
23950 * gimplify.c (maybe_add_early_return_predict_stmt): New
23952 (gimplify_return_expr): Call the function.
23953 * predict.c (tree_estimate_probability_bb): Remove handling
23955 * predict.def: Update comment about early return predictor.
23956 * gimple-predict.h (is_gimple_predict): New function.
23957 * predict.def: Change default value of early return to 66.
23958 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
23960 * passes.def: Put pass_strip_predict_hints to the beginning of
23963 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
23965 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
23966 FUNCTION_DECL declarations.
23967 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
23969 (dwaf2out_decl): Likewise.
23970 * godump.c (go_early_global_decl): Skip call to the real debug hook
23971 for FUNCTION_DECL declarations.
23972 * passes.c (rest_of_decl_compilation): Skip call to the
23973 early_global_decl debug hook for FUNCTION_DECL declarations, unless
23974 -fdump-go-spec is passed.
23976 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23978 * config/i386/i386.c (struct builtin_isa): New field pure_p.
23979 Reorder for compactness.
23980 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
23981 (def_builtin_pure, def_builtin_pure2): New functions.
23982 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
23984 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23986 * match.pd (nop_convert): New predicate.
23987 ((A +- CST1) +- CST2): Allow some NOP conversions.
23989 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23992 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
23993 with ctors/dtors if GOVD_SHARED is set.
23995 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23997 * config/aarch64/aarch64.md (movti_aarch64):
23998 Emit mov rather than orr.
23999 (movtf_aarch64): Likewise.
24000 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
24001 Emit mov rather than orr.
24003 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24005 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
24006 Swap alternatives, make integer dup more expensive.
24008 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24010 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
24011 Return true for non-tls symbols.
24013 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
24015 * config/aarch64/aarch64-cores.def (cortex-a55): New.
24016 (cortex-a75): Likewise.
24017 (cortex-a75.cortex-a55): Likewise.
24018 * config/aarch64/aarch64-tune.md: Regenerate.
24019 * doc/invoke.texi (-mtune): Document new values for -mtune.
24021 2017-06-21 Tom de Vries <tom@codesourcery.com>
24023 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
24024 stack_size feature.
24025 (Effective-Target Keywords, Other attributes): Suggest using
24026 dg-add-options stack_size feature to get stack limit in stack_size
24027 effective target documentation.
24029 2017-06-21 Julian Brown <julian@codesourcery.com>
24030 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24032 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
24033 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
24034 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
24036 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
24037 attribute type list for neon_multiply.
24038 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
24039 attribute type list for neon_multiply.
24040 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
24041 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
24042 attribute type list for neon_multiply.
24043 * config/arm/types.md (crypto_pmull): Add.
24044 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
24045 attribute type list.
24047 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
24049 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
24052 2017-06-20 Jakub Jelinek <jakub@redhat.com>
24054 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
24055 to make sure not to dereference a NULL cost_classes_ptr pointer.
24057 2017-06-20 Carl Love <cel@us.ibm.com>
24059 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24060 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
24061 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
24062 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
24063 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
24064 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
24065 VMULOSW): New enum "unspec" values.
24066 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
24067 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
24068 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
24069 altivec_vmulosw): New patterns.
24070 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
24071 VMULOSW): Add definitions.
24073 2017-06-20 Julia Koval <julia.koval@intel.com>
24075 * config/i386/i386.c: Fix rounding expand for new pattern.
24076 * config/i386/subst.md: Fix pattern (parallel -> unspec).
24078 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24080 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
24081 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
24083 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24085 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
24088 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24090 * config/aarch64/aarch64-cores.def: Rearrange to sort by
24091 architecture, then by implementer ID.
24092 * config/aarch64/aarch64-tune.md: Regenerate.
24094 2017-06-20 Richard Biener <rguenther@suse.de>
24096 PR middle-end/81097
24097 * fold-const.c (split_tree): Fold to type before negating.
24099 2017-06-20 David Malcolm <dmalcolm@redhat.com>
24101 * diagnostic-show-locus.c
24102 (selftest::test_fixit_deletion_affecting_newline): New function.
24103 (selftest::diagnostic_show_locus_c_tests): Call it.
24105 2017-06-20 Andreas Schwab <schwab@suse.de>
24108 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
24111 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24113 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
24114 __ARM_FEATURE_COPROC according to support.
24116 2017-06-20 Jakub Jelinek <jakub@redhat.com>
24118 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
24119 Rewritten to avoid overflow for > 32-bit pointers.
24122 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
24123 by removing enum keyword.
24124 (ubsan_type_descriptor): Likewise. Formatting fix.
24127 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
24128 splitter): Require TARGET_SSE2 in the condition.
24130 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
24133 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
24134 for doing vector set of SFmode on ISA 3.0.
24135 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
24136 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
24138 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
24139 SFmode value into a V4SF variable that was extracted from another
24140 V4SF variable without converting the element to double precision
24141 and back to single precision vector format.
24142 (vsx_insert_extract_v4sf_p9_2): Likewise.
24144 2017-06-19 Jakub Jelinek <jakub@redhat.com>
24146 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
24147 in UWHI to avoid undefined overflow.
24150 * ubsan.h (enum ubsan_encode_value_phase): New.
24151 (ubsan_encode_value): Change second argument to
24152 enum ubsan_encode_value_phase with default value of
24153 UBSAN_ENCODE_VALUE_GENERIC.
24154 * ubsan.c (ubsan_encode_value): Change second argument to
24155 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
24156 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
24157 create_tmp_var_raw instead of create_tmp_var and use a
24159 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
24160 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
24161 ubsan_encode_value callers.
24164 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
24165 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
24166 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
24168 2017-06-19 Richard Biener <rguenther@suse.de>
24170 PR middle-end/81118
24171 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
24172 estimates if we changed anything.
24174 2017-06-19 Richard Biener <rguenther@suse.de>
24176 PR tree-optimization/80887
24177 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
24178 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
24179 simplified lookups, then reset mprts_hook.
24180 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
24182 (try_to_simplify): Likewise.
24184 2017-06-19 Martin Liska <mliska@suse.cz>
24187 * gimplify.c (gimplify_switch_expr):
24188 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
24190 2017-06-19 Martin Liska <mliska@suse.cz>
24192 * doc/install.texi: Document that PGO runs in 4 stages.
24194 2017-06-19 Martin Liska <mliska@suse.cz>
24197 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
24198 to dispatcher function name.
24199 * multiple_target.c (replace_function_decl): New function.
24200 (create_dispatcher_calls): Redirect both edges and references.
24202 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
24204 * profile-count.c (profile_count::dump): Dump quality.
24205 (profile_count::differs_from_p): Update for unsigned val.
24206 * profile-count.h (profile_count_quality): New enum.
24207 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
24209 2017-06-19 Richard Biener <rguenther@suse.de>
24211 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
24212 struct function as arg.
24213 (estimate_numbers_of_iterations): Export overload with loop arg.
24214 (free_numbers_of_iterations_estimates_loop): Use an overload of
24215 free_numbers_of_iterations_estimates instead.
24216 * tree-cfg.c (remove_bb): Adjust.
24217 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
24218 * tree-parloops.c (gen_parallel_loop): Likewise.
24219 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
24221 (tree_unroll_loops_completely): Likewise.
24222 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
24223 Use an overload instead and export.
24224 (estimated_loop_iterations): Adjust.
24225 (max_loop_iterations): Likewise.
24226 (likely_max_loop_iterations): Likewise.
24227 (estimate_numbers_of_iterations): Take struct function as arg
24229 (loop_exits_before_overflow): Adjust.
24230 (free_numbers_of_iterations_estimates_loop): Use an overload.
24231 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
24232 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
24234 2017-06-19 Richard Biener <rguenther@suse.de>
24237 * ipa-prop.c (find_constructor_constant_at_offset): Handle
24238 RANGE_EXPR conservatively.
24240 2017-06-16 Carl Love <cel@us.ibm.com>
24242 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24243 definitions for vec_float, vec_float2, vec_floato,
24244 vec_floate built-ins.
24245 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
24246 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
24248 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
24249 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
24250 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
24251 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
24252 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
24253 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
24254 vec_floato): Add builtin defines.
24255 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
24256 Update the built-in documentation file for the new built-in
24259 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24261 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
24262 (mthumb): Mark as the negative of -marm.
24264 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24266 * doc/invoke.texi (ARM Options, -mcpu): Document supported
24268 (ARM Options, -mtune): Document that this accepts the same
24269 extension options as -mcpu.
24270 (ARM Options, -mfpu): Document addition of -mfpu=auto.
24272 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24274 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
24275 permitted extensions.
24277 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24279 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
24280 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
24281 (armv8-m.main): Add option +nodsp.
24282 * config/arm/arm-cpu-cdata.h: Regenerated.
24284 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24286 * config/arm/t-fuchsia: New file.
24287 * config.gcc (arm*-*-fuchsia*): Use it.
24289 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24291 * config/arm/t-symbian: Rewrite for new option infrastructure.
24293 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24295 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
24296 (MULTILIB_REQUIRED): Likewise.
24298 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24300 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
24301 (MULTILIB_RESUE): Likewise.
24302 (MULTILIB_MATCHES): Likewise.
24303 (MULTLIB_REQUIRED): Likewise.
24305 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24307 * config/arm/t-rtems: Rewrite for new option framework.
24309 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24311 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
24312 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
24313 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
24314 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
24315 * config/arm/t-multilib: ... here.
24316 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
24317 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
24318 armv7-a and armv8*-a when A-profile libraries have not been built.
24319 * config/arm/t-rmprofile: Rewrite.
24321 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24323 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
24324 with a backslash. Remove the backslash after substituting unescaped
24326 * doc/fragments.texi (MULTILIB_REUSE): Document it.
24328 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24330 * config.gcc: (arm*-*-*): When building a-profile libraries, force
24331 the driver to pass through the default setting of -mfloat-abi.
24332 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
24334 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
24335 (all_feat_combs): New rule.
24336 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
24338 * config/arm/t-aprofile: Rewrite.
24340 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24342 * config/arm/arm.h (FPUTYPE_AUTO): Define.
24343 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
24344 fpu is not specified by the user/command-line.
24345 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
24346 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
24347 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
24348 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
24349 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
24350 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
24352 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24354 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
24355 * config/arm/t-arm-elf: Rewritten.
24357 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24359 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
24360 have some floating-point instructions.
24361 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
24362 (TARGET_MAYBE_HARD_FLOAT): New macro.
24363 * config/arm/arm-builtins.c (arm_init_builtins): Use
24364 TARGET_MAYBE_HARD_FLOAT.
24365 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
24367 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24369 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
24370 (configargs.h): Include it.
24371 (arm_print_hint_for_fpu_option): New function.
24372 (arm_parse_fpu_option): New function.
24373 (candidate_extension): New class.
24374 (arm_canon_for_multilib): New function.
24375 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
24376 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
24377 (ARCH_CANONICAL_SPECS): New macro.
24378 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
24380 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24382 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
24383 are set after handling multilib fragments. Set target_cpu_default2
24386 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24388 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
24390 (arm*-*-*): Set target_cpu_default2 to a quoted string.
24391 * config/arm/parsecpu.awk (check_cpu): Validate any extension
24393 (check_arch): Likewise.
24394 * config/arm/arm.c (arm_configure_build_target): Handle
24395 TARGET_CPU_DEFAULT being a string constant. Scan any feature
24396 options in the default.
24398 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24400 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
24401 when an option is an alias of another.
24402 * config/arm/parsecpu.awk (optalias): New parser token.
24403 (gen_comm_data): Mark non-alias options as such. Emit entries
24404 for extension aliases.
24405 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
24406 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
24407 (armv6kz, armv6zk, armv6t2): Likewise.
24408 (armv7): Make vfpv3-d16 an alias.
24409 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
24411 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
24412 Sort in canonical order.
24413 (armv8-a): Sort in canonical order.
24414 (armv8.1-a, armv8.2-a): Likewise.
24415 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
24417 (cortex-a9): Sort in canonical order.
24418 * config/arm/arm.c (selftests.h): Include it.
24419 (arm_test_cpu_arch_data): New function.
24420 (arm_run_self_tests): New function.
24421 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
24422 (targetm): Move declaration to the end of the file.
24423 * arm-cpu-cdata.h: Regenerated.
24425 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24427 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
24428 call to target_mode_check describing the type of option passed.
24429 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
24430 (arm_target_thumb_only): Use arm_parse_arch_option_name or
24431 arm_parse_cpu_option_name to match parameters against list of
24433 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
24434 arm_arch_core_flags data structure.
24435 * config/arm/arm-cpu_cdata.h: Regenerated.
24437 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24439 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
24441 (arm_print_hint_for_cpu_option): Likewise.
24442 (arm_print_hint_for_arch_option): Likewise.
24443 (arm_parse_cpu_option_name): Likewise.
24444 (arm_parse_arch_option_name): Likewise.
24445 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
24446 of entries in the all_fpus list.
24447 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
24448 (arm_parse_cpu_option_name): Declare.
24449 (arm_parse_arch_option_name): Declare.
24450 (arm_parse_option_features): Declare.
24451 (arm_intialize_isa): Declare.
24452 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
24454 (gen_comm_data): ... here. Make definitions non-static.
24455 * config/arm/arm-cpu-data.h: Regenerated.
24456 * config/arm/arm-cpu-cdata.h: Regenerated.
24458 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24460 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
24461 (cpu_arch_extension): New structure.
24462 (cpu_arch_option, arch_option, cpu_option): New structures.
24463 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
24464 architecture types.
24465 (gen_data): Generate new format data tables.
24466 * config/arm/arm.c (cpu_tune): New structure.
24467 (cpu_option, processors): Delete.
24468 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
24469 (arm_print_hint_for_cpu_option): ... this and ...
24470 (arm_print_hint_for_arch_option): ... this.
24471 (arm_parse_arch_cpu_name): Delete. Replace with ...
24472 (arm_parse_cpu_option_name): ... this and ...
24473 (arm_parse_arch_option_name): ... this.
24474 (arm_unrecognized_feature): Change type of target parameter to
24476 (arm_parse_arch_cpu_features): Delete. Replace with ...
24477 (arm_parse_option_features): ... this.
24478 (arm_configure_build_target): Rework to use new configuration data
24480 (arm_print_tune_info): Rework for new configuration data tables.
24481 * config/arm/arm-cpu-data.h: Regenerated.
24482 * config/arm/arm-cpu.h: Regenerated.
24484 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24486 * Makefile.in (OBJS): Move sbitmap.o from here ...
24487 (OBJS-libcommon): ... to here.
24489 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24491 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
24492 (ISA_ALL_CRYPTO): New macro.
24493 (ISA_ALL_SIMD): New macro
24494 (ISA_ALL_FP): New macro.
24495 * config/arm/arm.c (fpu_bitlist): Update initializer.
24496 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
24498 (arm9e): Add fpu. Add option for nofp
24499 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
24500 (arm926ej-s, arm1026ej-s): Likewise.
24501 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
24502 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
24503 neon-fp16, neon-vfpv4, nofp and nosimd.
24504 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
24505 (cortex-a8): Add fpu. Add option for nofp.
24506 (cortex-a9): Add fpu. Add options for nosimd and nofp.
24507 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
24508 (cortex-r4f): Add fpu.
24509 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
24510 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
24512 (cortex-r8): Likewise.
24513 (cortex-m4): Add fpu. Add option for nofp.
24514 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
24515 (cortex-a17.cortex-a7): Likewise.
24516 (cortex-a32): Add fpu. Add options for crypto and nofp.
24517 (cortex-a35, cortex-a53): Likewise.
24518 (cortex-a57): Add fpu. Add option for crypto.
24519 (cortex-a72, cortex-a73): Likewise.
24520 (exynos-m1): Likewise.
24521 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
24522 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
24523 (cortex-m33): Add fpu. Add option for nofp.
24524 * config/arm/arm-cpu-cdata.h: Regenerated
24525 * config/arm/arm-cpu-data.h: Regenerated.
24527 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24529 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
24530 (armv5te, armv5tej): Likewise.
24531 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
24532 (armv7): Add options fp and vfpv3-d16.
24533 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
24534 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
24536 (armv7ve): Likewise.
24537 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
24538 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
24539 (armv8-a): Add nocrypto option.
24540 (armv8.1-a, armv8.2-a): Likewise.
24541 (armv8-m.main): add options fp, fp.dp and nofp.
24543 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24545 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
24547 (armv8-a+crc): Delete.
24548 (armv8.1-a): Add options simd, crypto and nofp.
24549 (armv8.2-a): Add options fp16, simd, crypto and nofp.
24550 (armv8.2-a+fp16): Delete.
24551 (armv8-m.main): Add option dsp.
24552 (armv8-m.main+dsp): Delete.
24553 (cortex-a8): Add fpu. Add nofp option.
24554 (cortex-a9): Add fpu. Add nofp and nosimd options.
24555 * config/arm/parsecpu.awk (gen_data): Generate option tables and
24556 link to main cpu and architecture data structures.
24557 (gen_comm_data): Only put isa attributes from the main architecture
24559 (option): New statement for architecture and CPU entries.
24560 * arm.c (struct cpu_option): New structure.
24561 (struct processors): Add entry for options.
24562 (arm_unrecognized_feature): New function.
24563 (arm_parse_arch_cpu_name): Ignore any characters after the first
24565 (arm_parse_arch_cpu_feature): New function.
24566 (arm_configure_build_target): Separate out any CPU and architecture
24567 features and parse separately. Don't error out if -mfpu=auto is
24568 used with only an architecture string.
24569 (arm_print_asm_arch_directives): New function.
24570 (arm_file_start): Call it.
24571 * config/arm/arm-cpu-cdata.h: Regenerated.
24572 * config/arm/arm-cpu-data.h: Likewise.
24573 * config/arm/arm-tables.opt: Likewise.
24575 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24577 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
24578 assembler when it is not -mfpu=auto.
24580 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24582 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
24583 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
24584 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
24585 (ASM_CPU_SPEC): Rewrite.
24586 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
24587 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
24588 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
24589 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
24590 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
24591 copied string is NUL-terminated. Also strip any characters prefixed
24593 (arm_rewrite_selected_arch): New function.
24594 (arm_rewrite_march): New function.
24596 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24598 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
24599 (x_arm_cpu_string, x_arm_tune_string): Likewise.
24600 (march, mcpu, mtune): Convert to string-based options.
24601 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
24602 (arm_parse_arch_cpu_name): New function.
24603 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
24604 identify selected architecture or CPU.
24605 (arm_option_save): New function.
24606 (TARGET_OPTION_SAVE): Redefine.
24607 (arm_option_restore): Restore string options.
24608 (arm_option_print): Print string options.
24610 2017-06-16 Martin Sebor <msebor@redhat.com>
24612 PR tree-optimization/80933
24613 PR tree-optimization/80934
24614 * builtins.c (fold_builtin_3): Do not handle bcmp here.
24615 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
24616 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
24617 (gimple_fold_builtin): Call them.
24619 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24621 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
24622 as unlikely; update profile.
24624 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24626 * predict.c (force_edge_cold): Handle declaring edges impossible
24629 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24631 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
24633 (try_unroll_loop_completely): Fix reporting.
24635 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24637 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
24639 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
24642 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
24643 if given a non-constant argument for an intrinsic which requires a
24646 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24648 * profile.c (compare_freqs): New function.
24649 (branch_prob): Sort edge list.
24650 (find_spanning_tree): Assume that the list is priority sorted.
24652 2017-06-16 Richard Biener <rguenther@suse.de>
24654 PR tree-optimization/81090
24655 * passes.def (pass_record_bounds): Remove.
24656 * tree-pass.h (make_pass_record_bounds): Likewise.
24657 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
24658 make_pass_record_bounds): Likewise.
24659 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
24660 not free niter estimates at the beginning but at the end.
24661 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
24663 2017-06-16 Richard Biener <rguenther@suse.de>
24665 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
24666 initializer to workaround ICE in host GCC 4.8.
24668 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24670 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
24672 (clone_inlined_nodes): Update.
24674 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24676 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
24677 prefetch settings, and enable prefetching by default at -O3.
24679 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24681 * config/aarch64/aarch64.c (aarch64_override_options_internal):
24682 Set flag_prefetch_loop_arrays according to tuning data.
24684 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24686 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
24687 New tune structure.
24688 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
24689 [Unrelated to main purpose of the patch] Place the pointer field last
24690 to enable type checking errors when tune structure are wrongly merged.
24691 * config/aarch64/aarch64.c (generic_prefetch_tune,)
24692 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
24693 (thunderx2t99_prefetch_tune): New tune constants.
24694 (tune_params *_tunings): Update all tunings (no functional change).
24695 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
24696 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
24697 from tunings structures.
24699 2017-06-16 Jakub Jelinek <jakub@redhat.com>
24702 * ubsan.c (instrument_null): Add T argument, use it instead
24703 of computing it based on IS_LHS.
24704 (instrument_object_size): Likewise.
24705 (pass_ubsan::execute): Adjust instrument_null and
24706 instrument_object_size callers to pass gimple_get_lhs or
24707 gimple_assign_rhs1 result to it. Use instrument_null instead of
24708 calling get_base_address and instrument_mem_ref. Handle
24709 aggregate call arguments for object-size sanitization.
24711 2017-06-16 Yury Gribov <tetra2005@gmail.com>
24713 PR tree-optimization/81089
24714 * tree-vrp.c (is_masked_range_test): Validate operands of
24717 2017-06-15 Martin Sebor <msebor@redhat.com>
24720 * dumpfile.c (dump_register): Avoid calling memset to initialize
24721 a class with a default ctor.
24722 * gcc.c (struct compiler): Remove const qualification.
24723 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
24724 * hash-table.h: Ditto.
24725 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
24727 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
24728 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
24730 * params.h (struct param_info): Make struct members non-const.
24731 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
24732 with default initialization.
24733 * vec.h (vec_copy_construct, vec_default_construct): New helper
24735 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
24736 with vec_copy_construct.
24737 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
24738 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
24739 * doc/invoke.texi (-Wclass-memaccess): Document.
24741 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24743 * emit-rtl.h (is_leaf): Update comment about local
24744 register allocator.
24746 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
24749 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
24750 for a variable to have a section before checking if the section has a
24752 Set section to.persistent if persistent attribute is set.
24753 Warn if .persistent attribute is used on an automatic variable.
24755 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
24757 PR rtl-optimization/80474
24758 * reorg.c (update_block): Do not ignore instructions in a delay slot.
24760 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
24762 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
24765 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
24767 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
24768 (casesi): Emit bounds checking as RTL.
24769 (casesi_internal_mips16_<mode>): Remove bounds checking.
24771 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
24773 * config/xtensa/xtensa.c (xtensa_option_override): Append
24774 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
24775 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
24776 xtensa_doloop_hooks): Define unconditionally.
24777 (xtensa_reorg_loops): Only call reorg_loops in the presence of
24779 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
24780 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
24781 for it in xtensa_option_override.
24782 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
24783 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
24785 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
24787 * doc/cppopts.texi: Document '-' special value to -MF.
24789 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
24791 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
24792 (cortex_a53_fconst): Likewise.
24793 (cortex_a53_fpmul): Likewise.
24794 (cortex_a53_f_load_64): Likewise.
24795 (cortex_a53_f_load_many): Likewise.
24796 (cortex_a53_advsimd_alu): Likewise.
24797 (cortex_a53_advsimd_alu_q): Likewise.
24798 (cortex_a53_advsimd_mul): Likewise.
24799 (cortex_a53_advsimd_mul_q): Likewise.
24800 (fpmac bypass): Add new bypass for fpmac-fpmac case.
24801 Add missing fmul, r2f_cvt and fconst cases.
24803 2017-06-14 Richard Biener <rguenther@suse.de>
24805 PR middle-end/81088
24806 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
24808 (fold_binary_loc): When associating do not treat pre-existing
24809 TREE_OVERFLOW on literal constants as a reason to allow
24810 TREE_OVERFLOW on associated literal constants.
24812 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24814 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
24815 (MASK_FEATURES): New macro.
24816 * config/sparc/sparc.c (sparc_option_override): Remove the special
24817 handling of -mfpu and generalize it to all MASK_FEATURES switches.
24819 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24821 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
24822 a division of 0 if non-call exceptions are enabled.
24824 2017-06-14 Andrew Pinski <apinski@cavium.com>
24825 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24828 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
24829 Improve vector initialization code gen for only variable case.
24831 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24833 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
24835 2017-06-14 Richard Biener <rguenther@suse.de>
24837 PR tree-optimization/81083
24838 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
24841 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24843 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
24844 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
24845 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
24846 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
24847 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
24848 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
24850 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24852 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
24853 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
24855 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24857 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
24859 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24861 * config/rs6000/t-rtems: Don't handle SPE.
24863 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24865 * config/rs6000/t-linux: Don't handle SPE.
24867 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24869 * config/rs6000/eabispe.h: Delete file.
24871 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24873 * config/rs6000/t-spe: Delete file.
24875 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24877 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
24878 (rs6000_legitimate_offset_address_p): Return false for anything in
24879 V2SImode or V2SFmode.
24881 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24883 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
24884 except V2SF and V2SI. Rearrange the vector modes, and add comments.
24885 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
24887 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
24888 (rs6000_legitimate_offset_address_p): Ditto.
24889 (rs6000_emit_move): Ditto.
24890 (rs6000_init_builtins): Remove V4HI_type_node.
24892 2017-06-13 Martin Liska <mliska@suse.cz>
24895 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
24896 (gate_asan): Likewise.
24897 * asan.h (asan_no_sanitize_address_p): Remove the function.
24898 (sanitize_flags_p): New function.
24899 * builtins.def: Fix coding style.
24900 * common.opt: Use renamed enum value.
24901 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
24902 * doc/extend.texi: Document no_sanitize attribute.
24903 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
24904 to SANITIZE_UNDEFINED_NONDEFAULT.
24905 * gcc.c (sanitize_spec_function): Use the renamed enum value.
24906 * gimple-fold.c (optimize_atomic_compare_exchange_p):
24907 Use sanitize_flags_p.
24908 * gimplify.c (gimplify_function_tree): Likewise.
24909 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
24910 * opts.c (parse_no_sanitize_attribute): New function.
24911 (common_handle_option): Use renamed enum value.
24912 * opts.h (parse_no_sanitize_attribute): Declare.
24913 * tree.c (sanitize_flags_p): New function.
24914 * tree.h: Declared here.
24915 * tsan.c: Use sanitize_flags_p.
24916 * ubsan.c (ubsan_expand_null_ifn): Likewise.
24917 (instrument_mem_ref): Likewise.
24918 (instrument_bool_enum_load): Likewise.
24919 (do_ubsan_in_current_function): Remove the function.
24920 (pass_ubsan::execute): Use sanitize_flags_p.
24921 * ubsan.h: Remove do_ubsan_in_current_function
24922 * tree-cfg.c (print_no_sanitize_attr_value): New function.
24923 (dump_function_to_file): Use it here.
24925 2017-06-13 Martin Jambor <mjambor@suse.cz>
24927 PR tree-optimization/80803
24928 PR tree-optimization/81063
24929 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
24930 (propagate_subaccesses_across_link): Enqueue subtree whenever
24931 necessary instead of relying on the caller.
24933 2017-06-13 Martin Jambor <mjambor@suse.cz>
24935 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
24936 that have a first_link.
24937 (sort_and_splice_var_accesses): Do not check first_link before
24939 (subtree_mark_written_and_enqueue): Likewise.
24940 (propagate_all_subaccesses): Likewise and do not stop at first
24941 parent with a first_link.
24943 2017-06-13 Martin Jambor <mjambor@suse.cz>
24945 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
24948 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24950 * match.pd: New pattern.
24952 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24954 * tree-vrp.c (is_masked_range_test): New function.
24955 (register_edge_assert_for): Determine ranges for
24958 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24960 PR tree-optimization/67328
24961 * fold-const.c (maskable_range_p): New function.
24962 (build_range_check): Generate bittests if possible.
24964 2017-06-13 Martin Liska <mliska@suse.cz>
24966 * gimple-pretty-print.c (dump_probability): Add new argument.
24967 (dump_edge_probability): Dump both probability and count.
24968 (dump_gimple_label): Likewise.
24969 (dump_gimple_bb_header): Likewise.
24971 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
24974 * config/avr/avr-devices.c: Fix indentation.
24975 * config/avr/gen-avr-mmcu-specs.c: Dito.
24977 2017-06-13 Richard Biener <rguenther@suse.de>
24979 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
24980 instead get vector type from stmt_info.
24981 (vectorizable_reduction): Adjust. Remove dead code.
24983 2017-06-13 Richard Biener <rguenther@suse.de>
24985 PR middle-end/81065
24986 * fold-const.c (extract_muldiv_1): Remove bogus distribution
24987 case of C * (x * C2 + C3).
24988 (fold_addr_of_array_ref_difference): Properly fold index difference.
24990 2017-06-12 David S. Miller <davem@davemloft.net>
24993 * config/sparc/sparc.md (return expander): Emit frame blockage if
24994 function uses alloca.
24996 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
24998 * combine.c (make_field_assignment): Check len rather than the mode
24999 precision when calling force_to_mode.
25001 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
25003 Support multilibs and devices that see flash in RAM address range.
25006 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
25007 (avr_mcu_t) <flash_pm_offset>: New field.
25008 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
25009 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
25010 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
25011 (AVR_TINY_PM_OFFSET): Remove macro.
25012 * config/avr/avr.opt (-mshort-calls): New option.
25013 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
25014 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
25015 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
25016 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
25017 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
25018 instead of avr_arch->have_jmp_call.
25019 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
25020 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
25021 avr_arch->flash_pm_offset to define.
25022 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
25023 new field flash_pm_offset. Add entry for avrxmega3.
25024 (avr_texinfo): Add entry for avrxmega3.
25025 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
25026 attiny212, attiny214,
25027 attiny412, attiny414, attiny416, attiny417,
25028 attiny814, attiny816, attiny817,
25029 attiny1614, attiny1616, attiny1617,
25030 attiny3214, attiny3216, attiny3217.
25031 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
25032 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
25033 (avr_print_operand_address) [AVR_TINY]: Same.
25034 (avr_asm_init_sections) <readonly_data_section>: Only patch
25035 callback if avr_arch->flash_pm_offset = 0.
25036 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
25037 for rodata if avr_arch->flash_pm_offset != 0.
25038 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
25039 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
25040 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
25041 (m_options): Append opt_rcall.
25042 (m_dirnames): Append dir_rcall.
25043 * config/avr/t-multilib: Regenerate.
25045 * configure.ac [target=avr]: Check whether avrxmega3 default
25046 linker description file works as needed.
25047 * configure: Regenerate.
25048 * doc/avr-mmcu.texi: Regenerate.
25049 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
25050 <__AVR_ARCH__>: Document avrxmega3 and 103.
25051 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
25052 <__AVR_SHORT_CALLS__>: Document it.
25053 <__AVR_PM_BASE_ADDRESS__>: Document it.
25054 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
25055 (AVR Variable Attributes) <progmem>: Document this is
25056 not needed for avrxmega3.
25057 (AVR Named Address Spaces) <__flash>: Dito.
25059 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
25061 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
25063 2017-06-12 Doug Rupp <rupp@adacore.com>
25065 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
25066 Append vxworks-stdint.h to the tm_file list.
25067 * config/vxworks-stdint.h: New file.
25069 2017-06-12 Martin Liska <mliska@suse.cz>
25071 PR tree-optimization/81041
25072 * tree-profile.c (gimple_gen_ic_func_profiler):
25073 Create an extra BB in profile-generate
25074 (gimple_gen_time_profiler): Likewise.
25076 2017-06-12 Jakub Jelinek <jakub@redhat.com>
25078 PR tree-optimization/81003
25079 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
25080 (update_range_test): Use it instead of force_gimple_operand_gsi.
25082 2017-06-12 Richard Biener <rguenther@suse.de>
25084 PR tree-optimization/81053
25085 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
25086 with backedge value not defined in loop. Simplify def stmt
25089 2017-06-11 Tom de Vries <tom@codesourcery.com>
25092 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
25094 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
25095 nvptx_cannot_force_const_mem.
25097 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25099 * opts.c (finish_options): Move test for flag_split_stack after
25100 it has been initialized.
25102 2017-06-11 Jason Merrill <jason@redhat.com>
25104 * tree.h (id_equal): New.
25105 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
25106 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
25107 instead of strcmp of IDENTIFIER_POINTER.
25109 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25111 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
25112 (mark_all_inlined_calls_cdtor): Fix formating.
25113 (inline_transform): Rescale profile before inlining.
25115 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25117 * cgraph.h (cgraph_edge::clone): Update prototype.
25118 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
25119 (cgraph_node::create_clone): Update.
25120 (cgraph_node::create_version_clone): Update.
25121 * tree-inline.c (copy_bb): Update.
25122 (expand_call_inline): Update.
25124 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25126 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
25127 factored out from ...
25128 (rs6000_emit_prologue): ... here.
25130 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25132 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
25133 factored out from ...
25134 (rs6000_emit_prologue): ... here.
25136 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25138 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
25140 (handle_missing_profiles): Fix computation of tp_first_run.
25141 (counts_to_freqs): Do not touch freqs when count is 0.
25143 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25145 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
25148 2017-06-10 Tom de Vries <tom@codesourcery.com>
25150 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
25151 attributes): Document signal effective target.
25153 2017-06-10 Tom de Vries <tom@codesourcery.com>
25155 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
25156 Document effective target stack_size.
25158 2017-06-09 David Malcolm <dmalcolm@redhat.com>
25160 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
25161 to the edit_context if they can be auto-applied.
25163 2017-06-9 Ian Lance Taylor <iant@golang.org>
25165 * opts.c (finish_options): If -fsplit-stack, disable implicit
25166 -forder-blocks-and-partition.
25167 * doc/invoke.texi (Optimize Options): Document that when using
25168 -fsplit-stack -forder-blocks-and-partition is not implicitly
25171 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25173 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
25174 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
25175 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
25176 * builtins.def (abort, trap, unreachable): Declare cold.
25177 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
25178 * tree-core.h (ECF_COLD): New.
25179 * tree.c (set_call_expr_flags): Handle ECF_COLD.
25180 (build_common_builtin_nodes): Mark unreachable and abort as cold.
25182 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25184 * predict.c (unlikely_executed_stmt_p): Cleanup.
25186 2017-06-09 Richard Biener <rguenther@suse.de>
25188 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
25189 model if the ref is always written to.
25191 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25193 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
25195 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25197 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
25200 2017-06-09 Tom de Vries <tom@codesourcery.com>
25203 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
25204 "target cannot support label values" when encountering LABEL_REF.
25206 2017-06-09 Martin Liska <mliska@suse.cz>
25208 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
25209 (gimple_gen_ic_func_profiler): Emit direct comparison
25210 of __gcov_indirect_call_callee with NULL.
25211 (gimple_gen_time_profiler): Change probability from
25212 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
25214 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25216 * profile.c (edge_gcov_counts): Turn to pointer.
25217 (compute_branch_probabilities, compute_branch_probabilities): Update.
25218 (branch_prob): Do not clear edge_gcov_count.
25219 * profile.h (edge_gcov_counts): Turn to pointer.
25220 (edge_gcov_count): Update.
25222 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25224 * gimple.h (gimple_check_failed): Mark cold.
25226 2017-06-09 Richard Biener <rguenther@suse.de>
25228 PR tree-optimization/66623
25229 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
25230 refactor check_reduction into two parts, properly computing
25231 whether we have to check reduction validity for outer loop
25234 2017-06-09 Richard Biener <rguenther@suse.de>
25236 PR tree-optimization/79483
25237 * graphite-scop-detection.c (order): New global.
25238 (get_order): Compute bb to order mapping that satisfies code
25239 generation constraints.
25240 (cmp_pbbs): New helper.
25241 (build_scops): Start domwalk at entry block, sort generated
25244 2017-06-09 Richard Biener <rguenther@suse.de>
25246 PR middle-end/81007
25247 * ipa-polymorphic-call.c
25248 (ipa_polymorphic_call_context::restrict_to_inner_class):
25249 Skip FIELD_DECLs with error_mark_node type.
25250 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
25253 2017-06-09 Martin Liska <mliska@suse.cz>
25255 * predict.c (struct branch_predictor): New struct.
25256 (test_prediction_value_range): New test.
25257 (predict_c_tests): New function.
25258 * selftest-run-tests.c (selftest::run_tests): Run the function.
25259 * selftest.h: Declare new tests.
25261 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
25264 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
25265 gen_add3_insn did not fail.
25266 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
25267 r0, construct that number in a temporary reg and add that reg to r0.
25268 If asked to put the result in r0 as well, fail.
25270 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
25272 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25273 for early expansion of vec_eqv.
25275 2017-06-08 Jakub Jelinek <jakub@redhat.com>
25277 PR middle-end/81005
25278 * ubsan.c (instrument_null): Avoid pointless code temporary.
25279 (pass_ubsan::execute): Instrument aggregate arguments of calls.
25281 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
25285 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
25288 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
25291 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25293 * predict.c (unlikely_executed_edge_p): Move ahead.
25294 (probably_never_executed_edge_p): Use it.
25296 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25298 PR middle-end/79988
25299 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
25300 gimple_call_builtin_p call.
25302 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25304 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
25305 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
25306 rtl_check_failed_type2, rtl_check_failed_code1,
25307 rtl_check_failed_code2, rtl_check_failed_code_mode,
25308 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
25309 rtvec_check_failed_bounds, rtl_check_failed_flag,
25310 _fatal_insn_not_found, _fatal_insn): Likewise.
25311 * tree.h (tree_contains_struct_check_failed,
25312 tree_check_failed, tree_not_check_failed,
25313 tree_class_check_failed, tree_range_check_failed,
25314 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
25315 tree_vec_elt_check_failed, phi_node_elt_check_failed,
25316 tree_operand_check_failed, omp_clause_check_failed,
25317 omp_clause_operand_check_failed, omp_clause_range_check_failed):
25320 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25322 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
25323 flag_branch_probabilities.
25324 * ipa-inline.c (edge_badness): Likewise.
25325 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
25326 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
25327 * predict.c (maybe_hot_frequency_p): Likewise.
25328 (probably_never_executed): Likewise.
25329 * sched-ebb.c (schedule_ebbs): Likewise.
25330 * sched-rgn.c (find_single_block_region): Likewise.
25331 * tracer.c (tail_duplicate): Likewise.
25333 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25335 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
25336 longer requires x_flag_profile_use.
25338 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25340 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
25341 instead of flag_reorder_blocks_and_partition.
25342 * dbxout.c (dbxout_function_end): Likewise.
25343 * dwarf2out.c (gen_subprogram_die): Likewise.
25344 * haifa-sched.c (sched_create_recovery_edges): Likewise.
25345 * hw-doloop.c (reorg_loops): Likewise.
25346 * varasm.c (assemble_start_function,
25347 assemble_end_function): Likewise.
25348 (decide_function_section): Do not check for
25349 flag_reorder_blocks_and_partition.
25351 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25353 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
25355 (chkp_get_hard_register_fake_addr_expr): Ditto.
25356 (chkp_build_addr_expr): Add check for hard reg case.
25357 (chkp_parse_array_and_component_ref): Ditto.
25358 (chkp_find_bounds_1): Ditto.
25359 (chkp_process_stmt): Don't generate bounds store for
25362 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25364 * predict.c (maybe_hot_bb_p): Do not check profile status.
25365 (maybe_hot_edge_p): Likewise.
25366 (probably_never_executed): Check for zero counts even if profile
25368 (unlikely_executed_edge_p): New function.
25369 (unlikely_executed_stmt_p): New function.
25370 (unlikely_executed_bb_p): New function.
25371 (set_even_probabilities): Use unlikely predicates.
25372 (combine_predictions_for_bb): Likewise.
25373 (predict_paths_for_bb): Likewise.
25374 (predict_paths_leading_to_edge): Likewise.
25375 (determine_unlikely_bbs): New function.
25376 (estimate_bb_frequencies): Use it.
25377 (compute_function_frequency): Use zero counts even if profile is
25379 * profile-count.h: Fix typo.
25381 2017-08-08 Julia Koval <julia.koval@intel.com>
25383 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
25384 _mm512_mask_cvtsepi16_storeu_epi8,
25385 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
25386 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
25387 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
25388 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
25389 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
25390 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
25391 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
25392 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
25393 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
25394 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
25395 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
25396 __builtin_ia32_pmovuswb256mem_mask,
25397 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
25398 __builtin_ia32_pmovwb512mem_mask): New builtins.
25400 2017-08-08 Julia Koval <julia.koval@intel.com>
25402 PR target/73350,80862
25403 * config/i386/subst.md (round): Fix round pattern.
25404 * config/i386/i386.c (ix86_erase_embedded_rounding):
25405 Fix erasing rounding for the fixed pattern.
25407 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25409 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
25411 2017-06-08 Martin Liska <mliska@suse.cz>
25413 PR gcov-profile/80911
25414 * gcov.c (block_info::block_info): New constructor.
25416 2017-06-07 Carl Love <cel@us.ibm.com>
25418 * config/rs6000/rs6000-c: The return type of the following
25419 built-in functions was implemented as int not long long. Fix sign
25420 of return value for the unsigned version of vec_mulo and vec_mule.
25421 vector unsigned long long vec_bperm (vector unsigned long long,
25422 vector unsigned char)
25423 vector signed long long vec_mule (vector signed int,
25425 vector unsigned long long vec_mule (vector unsigned int,
25426 vector unsigned int)
25427 vector signed long long vec_mulo (vector signed int,
25429 vector unsigned long long vec_mulo (vector unsigned int,
25430 vector unsigned int)
25431 * doc/extend.texi: Fix the documentation for the built-in
25434 2017-06-07 Carl Love <cel@us.ibm.com>
25437 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
25440 2017-06-07 Carl Love <cel@us.ibm.com>
25442 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
25443 support, Generate doublehv for signed int/float for BE case only.
25445 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
25447 * doc/invoke.texi (mcx16): Rewrite.
25449 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25451 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
25452 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
25453 *mov<mode>_softfloat, and an anonymous splitter): Use
25454 nonimmediate_operand instead of rs6000_nonimmediate_operand.
25456 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25458 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
25460 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25461 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
25462 (rs6000_debug_reg_global): Adjust.
25463 (rs6000_init_hard_regno_mode_ok): Adjust.
25464 (rs6000_dbx_register_number): Adjust.
25465 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
25466 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25467 Remove SPE_ACC and SPEFSCR.
25468 (REG_ALLOC_ORDER): Ditto.
25469 (FRAME_POINTER_REGNUM): Change to 111.
25470 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
25471 (REG_CLASS_NAMES): Ditto.
25472 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
25473 (REGISTER_NAMES): Ditto.
25474 (ADDITIONAL_REG_NAMES): Ditto.
25475 (rs6000_reg_names): Ditto.
25476 * config/rs6000/rs6000.md: Renumber some register number
25479 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25481 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
25483 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25484 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
25486 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
25487 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25488 Delete the SPE high registers.
25489 (REG_ALLOC_ORDER): Ditto.
25490 (enum reg_class): Remove SPE_HIGH_REGS.
25491 (REG_CLASS_NAMES): Ditto.
25492 (REG_CLASS_CONTENTS): Delete the SPE high registers.
25493 (REGISTER_NAMES): Ditto.
25494 (rs6000_reg_names): Ditto.
25495 * doc/tm.texi.in: Remove SPE as example.
25496 * doc/tm.texi: Regenerate.
25498 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25500 * config/rs6000/8540.md (ppc8540_brinc): Delete.
25501 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
25502 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
25503 * config/rs6000/rs6000.md (type): Remove "brinc".
25505 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25507 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
25508 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
25509 * config/rs6000/linuxspe.h: Delete file.
25510 * config/rs6000/rs6000.md: Don't include spe.md.
25511 * config/rs6000/spe.h: Delete file.
25512 * config/rs6000/spe.md: Delete file.
25513 * config/rs6000/t-rs6000: Remove spe.md.
25515 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25517 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
25518 (reg_or_none500mem_operand): Delete.
25519 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
25520 instead of reg_or_none500mem_operand.
25522 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25524 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
25525 handling of SPE flags.
25526 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
25528 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25530 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
25532 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
25533 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
25534 paired_divv2sf3): Similar.
25535 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
25536 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
25537 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
25539 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
25540 Rename the paired_* instruction patterns.
25541 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
25543 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
25544 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
25545 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
25546 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
25547 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
25548 PAIRED_VECTOR_MODE.
25549 (struct machine_function): Delete field spe_insn_chain_scanned_p.
25550 (spe_func_has_64bit_regs_p): Delete.
25551 (spe_expand_predicate_builtin): Delete.
25552 (spe_expand_evsel_builtin): Delete.
25553 (TARGET_DWARF_REGISTER_SPAN): Do not define.
25554 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
25555 (invalid_e500_subreg): Delete.
25556 (rs6000_legitimize_address): Always force_reg op2 as well, for
25557 paired single memory accesses.
25558 (rs6000_member_type_forces_blk): Delete.
25559 (rs6000_spe_function_arg): Delete.
25560 (rs6000_expand_unop_builtin): Delete SPE handling.
25561 (rs6000_expand_binop_builtin): Ditto.
25562 (spe_expand_stv_builtin): Delete.
25563 (bdesc_2arg_spe): Delete.
25564 (spe_expand_builtin): Delete.
25565 (spe_expand_predicate_builtin): Delete.
25566 (spe_expand_evsel_builtin): Delete.
25567 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
25568 (spe_init_builtins): Delete.
25569 (spe_func_has_64bit_regs_p): Delete.
25570 (savres_routine_name): Delete "info" parameter. Adjust callers.
25571 (rs6000_emit_stack_reset): Ditto.
25572 (rs6000_dwarf_register_span): Delete.
25573 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
25574 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
25575 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
25577 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
25579 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
25580 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
25581 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
25582 mulv2sf3, divv2sf3): Delete expanders.
25584 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25586 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
25588 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25590 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
25591 * config/rs6000/rs6000.c: Ditto.
25593 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25595 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
25596 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
25597 comparison_operator.
25599 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25601 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
25602 * config/rs6000/rs6000.opt: Ditto.
25603 * config/rs6000/t-rtems: Ditto.
25605 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25607 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
25608 TARGET_E500_SINGLE by 0, simplify.
25609 * config/rs6000/rs6000.c: Ditto.
25610 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
25611 (spe_build_register_parallel): Delete.
25612 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
25613 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
25614 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
25615 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
25616 (E500_CONVERT): Delete.
25617 * config/rs6000/spe.md: Remove many patterns and all define_constants.
25619 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25621 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
25622 * config/rs6000/dfp.md: Ditto.
25623 (negdd2, *negdd2_fpr): Merge.
25624 (absdd2, *absdd2_fpr): Merge.
25625 (negtd2, *negtd2_fpr): Merge.
25626 (abstd2, *abstd2_fpr): Merge.
25627 * config/rs6000/e500.h: Delete file.
25628 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
25629 TARGET_FPRS by 1 and simplify.
25630 * config/rs6000/rs6000-c.c: Ditto.
25631 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
25632 TARGET_DF_SPE by 0.
25633 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
25635 * config/rs6000/rs6000.md: Ditto.
25636 (floatdidf2, *floatdidf2_fpr): Merge.
25637 (move_from_CR_gt_bit): Delete.
25638 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
25639 (E500_CR_IOR_COMPARE): Delete.
25640 (All patterns that require !TARGET_FPRS): Delete.
25641 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
25643 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25645 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
25647 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25649 * graphds.c (add_edge): Intitialize edge's attached data.
25650 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
25651 pointer parameter. Call pointed function on each edge during
25652 graph traversing. Skip traversing the edge when the function
25654 (graphds_dfs, graphds_scc): Ditto.
25655 (for_each_edge): New parameter. Pass the new parameter to callback
25657 * graphds.h (skip_edge_callback): New function pointer type.
25658 (graphds_dfs, graphds_scc): New function pointer parameter.
25659 (graphds_edge_callback, for_each_edge): New parameter.
25661 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25663 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
25664 out code checking if runtime alias check is possible to below ...
25665 Call the new function.
25666 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
25667 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
25669 2017-06-07 Marek Polacek <polacek@redhat.com>
25672 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
25673 TYPE_OVERFLOW_WRAPS check.
25675 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25677 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
25678 if versioning is required.
25679 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
25680 peeling with the check for versioning.
25682 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25684 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
25685 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
25686 Set true to new parameter if new ssa variable is defined.
25687 (vect_gen_vector_loop_niters): Refactor. Set range information
25688 for the new vector loop bound variable.
25689 (vect_do_peeling): Ditto.
25691 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25693 * tree-affine.c (ssa.h): Include header file.
25694 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
25695 has wrapping overflow behavior.
25697 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25699 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
25701 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25703 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
25704 (tree_to_aff_combination): ... here.
25706 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25708 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
25709 reg_pressure model function.
25710 (ivopts_global_cost_for_size): Delete.
25711 (determine_set_costs, iv_ca_recount_cost): Call new model function
25712 ivopts_estimate_reg_pressure.
25714 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25716 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
25717 expensive than udiv. Remove floating point cases from mod.
25719 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25721 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
25722 Increase idiv cost.
25724 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25726 * config/aarch64/aarch64.md
25727 (copysignsf3): Fix mask generation.
25729 2017-06-07 Jakub Jelinek <jakub@redhat.com>
25731 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
25733 (class dump_manager): Add register_dumps method.
25734 * dumpfile.c: Include langhooks.h.
25735 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
25736 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
25737 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
25738 (dump_manager::dump_register): Start with 512 entries instead of 32.
25739 (dump_manager::register_dumps): New method.
25740 * toplev.c (general_init): Instead of invoking register_dumps
25741 langhook, invoke register_dumps method on the dump manager.
25742 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
25745 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
25747 * doc/md.texi: Clarify the restrictions on a define_insn condition.
25748 Say that # requires an associated define_split to exist, and that
25749 the define_split must be suitable for use after register allocation.
25751 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25753 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
25754 (compute_outgoing_frequencies): Also initialize zero counts.
25755 (find_many_sub_basic_blocks): Do not produce uninitialized profile
25756 around loops; preserve more of profile when nothing changes.
25758 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
25760 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
25762 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
25763 * config/arm/arm-cpu-cdata.h: Regenerate.
25764 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
25765 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
25766 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
25767 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
25768 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
25770 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
25772 * config/arm/t-rmprofile: Likewise.
25773 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
25775 2017-06-06 David S. Miller <davem@davemloft.net>
25778 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
25779 blockage if function uses alloca.
25781 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25783 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
25784 New "uid" fields to hold pretty-print IDs of group and ref.
25785 Memory references are now identified as <group_id>:<ref_id>
25786 instead of using [random] addresses.
25787 (dump_mem_details): Simplify, no functional change.
25788 (dump_mem_ref): Simplify and make output more concise.
25789 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
25790 (find_or_create_group): Initialize group uid.
25791 (record_ref): Initialize ref uid. Improve debug output.
25792 (prune_group_by_reuse, should_issue_prefetch_p,)
25793 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
25794 (mark_nontemporal_store, determine_loop_nest_reuse):
25795 Improve debug output.
25797 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25799 * dbgcnt.def (prefetch): New debug counter.
25800 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
25801 (schedule_prefetches): Stop issueing prefetches if debug counter
25804 2017-06-06 Tom de Vries <tom@codesourcery.com>
25806 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
25807 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
25809 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25811 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
25812 Use aarch64_reg_or_zero predicate for operand 4.
25813 (aarch64_compare_and_swap<mode> define_insn_and_split):
25814 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
25815 (aarch64_store_exclusive<mode>): Likewise for operand 2.
25817 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25819 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
25820 (arm_compute_save_core_reg_mask): This.
25821 (thumb1_compute_save_reg_mask): Rename into ...
25822 (thumb1_compute_save_core_reg_mask): This.
25823 (arm_compute_save_reg0_reg12_mask): Adapt comment.
25824 (arm_compute_frame_layout): Likewise.
25826 2017-06-06 Richard Biener <rguenther@suse.de>
25828 PR tree-optimization/80974
25829 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
25830 keep or clear leaders SSA info.
25832 2017-06-06 Tom de Vries <tom@codesourcery.com>
25834 * config/nvptx/nvptx.c (split_mode_p): New function.
25835 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
25837 2017-06-06 Tom de Vries <tom@codesourcery.com>
25839 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
25841 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25844 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
25847 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25849 * shrink-wrap.c (handle_simple_exit): Update profile.
25850 (try_shrink_wrapping): Upate profile.
25852 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25854 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
25855 (tree_guess_outgoing_edge_probabilities): New.
25856 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
25857 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
25859 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25861 * ipa-split.c (split_function): Initialize return bb profile.
25863 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25865 * profile.c (compute_branch_probabilities): Also initialize
25866 EXIT_BLOCK profile.
25868 2017-06-06 Richard Biener <rguenther@suse.de>
25870 PR tree-optimization/80928
25871 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
25872 (vect_analyze_loop_operations): Properly guard analysis for
25874 (vect_transform_loop): Likewise.
25875 (vect_analyze_loop_2): Also reset SLP type on PHIs.
25876 (vect_model_induction_cost): Do not cost for pure SLP.
25877 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
25878 of induction in inner loop vectorization.
25879 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
25880 (vect_get_and_check_slp_defs): Handle vect_induction_def.
25881 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
25883 (vect_analyze_slp_cost_1): Cost induction.
25884 (vect_detect_hybrid_slp_stmts): Handle PHIs.
25885 (vect_get_slp_vect_defs): Likewise.
25886 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
25887 (vect_transform_stmt): Handle SLP reductions.
25888 * tree-vectorizer.h (vectorizable_induction): Adjust.
25890 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25892 * config/rs6000/rs6000.c (make_resolver_func): Update
25893 init_lowered_empty_function call.
25895 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
25897 * doc/invoke.texi: Document the -fprofile-abs-path option.
25898 * common.opt (fprofile-abs-path): New option.
25899 * gcov-io.h (gcov_write_filename): Declare.
25900 * gcov-io.c (gcov_write_filename): New function.
25901 * coverage.c (coverage_begin_function): Use gcov_write_filename.
25902 * profile.c (output_location): Likewise.
25904 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25906 * shring-wrap.c: Revert accidental commit.
25908 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
25910 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
25912 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25914 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
25916 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
25917 profile in callgraph edge.
25918 * profile-count.h (apply_probability): If THIS is 0, then result is 0
25919 (apply_scale): Likewise.
25920 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
25921 Also scale profile when inlining function with zero profile.
25922 (initialize_cfun): Update exit block profile even when it is zero.
25923 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
25924 when profile is read.
25926 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25928 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
25929 (CLONE_*): New constants to define the processors we can generate
25930 code for with the target_clone attribute.
25931 (rs6000_clone_map): New array to identify which clone processors
25932 the current program is running on.
25933 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
25934 target_clone attribute.
25935 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
25936 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
25937 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
25938 (cpu_expand_builtin): Add support for target_clone attribute.
25939 (rs6000_valid_attribute_p): Allow "default" attribute.
25940 (get_decl_name): New debug function to simplify printing the
25941 current function name in debugging statements.
25942 (rs6000_clone_priority): New functions to support the target_clone
25943 attribute, and be able to generate code to switch between ISA 2.05
25944 through ISA 3.0 (power6 through power9).
25945 (rs6000_compare_version_priority): Likewise.
25946 (rs6000_get_function_versions_dispatcher): Likewise.
25947 (make_resolver_func): Likewise.
25948 (add_condition_to_bb): Likewise.
25949 (dispatch_function_versions): Likewise.
25950 (rs6000_generate_version_dispatcher_body): Likewise.
25951 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
25952 (fusion_gpr_load_p): Fix a spacing issue.
25953 * doc/extend.texi (Common Function Attributes): Document that the
25954 PowerPC supports the target_clone attribute.
25956 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25958 * config/arm/arm.h: explain F symbol found in description of ARM
25959 register allocation in its legend.
25961 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25963 * config/mips/frame-header-opt.c: Include profile-count.h.
25964 * config/riscv/riscv.c: Include profile-count.h
25966 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25968 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
25970 (sm_set_flag_if_changed): Add bbs field.
25971 (execute_sm_if_changed_flag_set): Pass BBS.
25972 (execute_sm): Update.
25974 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25976 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
25979 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25981 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
25982 (peephole2): New peephole2 to emit the above.
25983 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
25985 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25987 * config/aarch64/aarch64.c (define_peephole2 above
25988 *sub_<shift>_<mode>): New peephole.
25990 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25992 * config/i386/i386.c (make_resolver_func): Update.
25993 * Makefile.in: Add profile-count.h and profile-count.o
25994 * auto-profile.c (afdo_indirect_call): Update to new API.
25995 (afdo_set_bb_count): Update.
25996 (afdo_propagate_edge): Update.
25997 (afdo_propagate_circuit): Update.
25998 (afdo_calculate_branch_prob): Update.
25999 (afdo_annotate_cfg): Update.
26000 * basic-block.h: Include profile-count.h
26001 (struct edge_def): Turn count to profile_count.
26002 (struct basic_block_def): Likewie.
26003 (REG_BR_PROB_BASE): Move to profile-count.h
26004 (RDIV): Move to profile-count.h
26005 * bb-reorder.c (max_entry_count): Turn to profile_count.
26006 (find_traces): Update.
26007 (rotate_loop):Update.
26008 (connect_traces):Update.
26009 (sanitize_hot_paths):Update.
26010 * bt-load.c (migrate_btr_defs): Update.
26011 * cfg.c (RDIV): Remove.
26012 (init_flow): Use alloc_block.
26013 (alloc_block): Uninitialize count.
26014 (unchecked_make_edge): Uninitialize count.
26015 (check_bb_profile): Update.
26016 (dump_edge_info): Update.
26017 (dump_bb_info): Update.
26018 (update_bb_profile_for_threading): Update.
26019 (scale_bbs_frequencies_int): Update.
26020 (scale_bbs_frequencies_gcov_type): Update.
26021 (scale_bbs_frequencies_profile_count): New.
26022 * cfg.h (update_bb_profile_for_threading): Update.
26023 (scale_bbs_frequencies_profile_count): Declare.
26024 * cfgbuild.c (compute_outgoing_frequencies): Update.
26025 (find_many_sub_basic_blocks): Update.
26026 * cfgcleanup.c (try_forward_edges): Update.
26027 (try_crossjump_to_edge): Update.
26028 * cfgexpand.c (expand_gimple_tailcall): Update.
26029 (construct_exit_block): Update.
26030 * cfghooks.c (verify_flow_info): Update.
26031 (dump_bb_for_graph): Update.
26032 (split_edge): Update.
26033 (make_forwarder_block): Update.
26034 (duplicate_block): Update.
26035 (account_profile_record): Update.
26036 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
26037 (get_estimated_loop_iterations): Update.
26038 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
26039 (single_likely_exit): Update.
26040 * cfgloopmanip.c (scale_loop_profile): Update.
26042 (set_zero_probability): Update.
26043 (lv_adjust_loop_entry_edge): Update.
26044 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
26045 (purge_dead_edges): Update.
26046 (rtl_account_profile_record): Update.
26047 * cgraph.c (cgraph_node::create): Uninitialize count.
26048 (symbol_table::create_edge): Uninitialize count.
26049 (cgraph_update_edges_for_call_stmt_node): Update.
26050 (cgraph_edge::dump_edge_flags): Update.
26051 (cgraph_node::dump): Update.
26052 (cgraph_edge::maybe_hot_p): Update.
26053 * cgraph.h: Include profile-count.h
26054 (create_clone), create_edge, create_indirect_edge): Update.
26055 (cgraph_node): Turn count to profile_count.
26056 (cgraph_edge0: Likewise.
26057 (make_speculative, clone): Update.
26058 (create_edge): Update.
26059 (init_lowered_empty_function): Update.
26060 * cgraphclones.c (cgraph_edge::clone): Update.
26061 (duplicate_thunk_for_node): Update.
26062 (cgraph_node::create_clone): Update.
26063 * cgraphunit.c (cgraph_node::analyze): Update.
26064 (cgraph_node::expand_thunk): Update.
26065 * final.c (dump_basic_block_info): Update.
26066 * gimple-streamer-in.c (input_bb): Update.
26067 * gimple-streamer-out.c (output_bb): Update.
26068 * graphite.c (print_global_statistics): Update.
26069 (print_graphite_scop_statistics): Update.
26070 * hsa-brig.c: Include basic-block.h.
26071 * hsa-dump.c: Include basic-block.h.
26072 * hsa-gen.c (T sum_slice): Update.
26073 (convert_switch_statements):Update.
26074 * hsa-regalloc.c: Include basic-block.h.
26075 * ipa-chkp.c (chkp_produce_thunks): Update.
26076 * ipa-cp.c (struct caller_statistics): Update.
26077 (init_caller_stats): Update.
26078 (gather_caller_stats): Update.
26079 (ipcp_cloning_candidate_p): Update.
26080 (good_cloning_opportunity_p): Update.
26081 (get_info_about_necessary_edges): Update.
26082 (dump_profile_updates): Update.
26083 (update_profiling_info): Update.
26084 (update_specialized_profile): Update.
26085 (perhaps_add_new_callers): Update.
26086 (decide_about_value): Update.
26087 (ipa_cp_c_finalize): Update.
26088 * ipa-devirt.c (struct odr_type_warn_count): Update.
26089 (struct decl_warn_count): Update.
26090 (struct final_warning_record): Update.
26091 (possible_polymorphic_call_targets): Update.
26092 (ipa_devirt): Update.
26093 * ipa-fnsummary.c (redirect_to_unreachable): Update.
26094 * ipa-icf.c (sem_function::merge): Update.
26095 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
26096 * ipa-inline.c (compute_uninlined_call_time): Update.
26097 (compute_inlined_call_time): Update.
26098 (want_inline_small_function_p): Update.
26099 (want_inline_self_recursive_call_p): Update.
26100 (edge_badness): Update.
26101 (lookup_recursive_calls): Update.
26102 (recursive_inlining): Update.
26103 (inline_small_functions): Update.
26104 (dump_overall_stats): Update.
26105 (dump_inline_stats): Update.
26106 * ipa-profile.c (ipa_profile_generate_summary): Update.
26107 (ipa_propagate_frequency): Update.
26108 (ipa_profile): Update.
26109 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26110 * ipa-utils.c (ipa_merge_profiles): Update.
26111 * loop-doloop.c (doloop_modify): Update.
26112 * loop-unroll.c (report_unroll): Update.
26113 (unroll_loop_runtime_iterations): Update.
26114 * lto-cgraph.c (lto_output_edge): Update.
26115 (lto_output_node): Update.
26116 (input_node): Update.
26117 (input_edge): Update.
26118 (merge_profile_summaries): Update.
26119 * lto-streamer-in.c (input_cfg): Update.
26120 * lto-streamer-out.c (output_cfg): Update.
26121 * mcf.c (create_fixup_graph): Update.
26122 (adjust_cfg_counts): Update.
26123 (sum_edge_counts): Update.
26124 * modulo-sched.c (sms_schedule): Update.
26125 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
26126 * predict.c (maybe_hot_count_p): Update.
26127 (probably_never_executed): Update.
26128 (dump_prediction): Update.
26129 (combine_predictions_for_bb): Update.
26130 (propagate_freq): Update.
26131 (handle_missing_profiles): Update.
26132 (counts_to_freqs): Update.
26133 (rebuild_frequencies): Update.
26134 (force_edge_cold): Update.
26135 * predict.h: Include profile-count.h
26136 (maybe_hot_count_p, counts_to_freqs): UPdate.
26137 * print-rtl-function.c: Do not include cfg.h
26138 * print-rtl.c: Include basic-block.h
26139 * profile-count.c: New file.
26140 * profile-count.h: New file.
26141 * profile.c (is_edge_inconsistent): Update.
26142 (correct_negative_edge_counts): Update.
26143 (is_inconsistent): Update.
26144 (set_bb_counts): Update.
26145 (read_profile_edge_counts): Update.
26146 (compute_frequency_overlap): Update.
26147 (compute_branch_probabilities): Update; Initialize and deinitialize
26149 (branch_prob): Update.
26150 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
26151 (edge_gcov_count): New.
26152 (bb_gcov_count): New.
26153 * shrink-wrap.c (try_shrink_wrapping): Update.
26154 * tracer.c (better_p): Update.
26155 * trans-mem.c (expand_transaction): Update.
26156 (ipa_tm_insert_irr_call): Update.
26157 (ipa_tm_insert_gettmclone_call): Update.
26158 * tree-call-cdce.c: Update.
26159 * tree-cfg.c (gimple_duplicate_sese_region): Update.
26160 (gimple_duplicate_sese_tail): Update.
26161 (gimple_account_profile_record): Update.
26162 (execute_fixup_cfg): Update.
26163 * tree-inline.c (copy_bb): Update.
26164 (copy_edges_for_bb): Update.
26165 (initialize_cfun): Update.
26166 (freqs_to_counts): Update.
26167 (copy_cfg_body): Update.
26168 (expand_call_inline): Update.
26169 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26170 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
26171 (try_unroll_loop_completely): Update.
26172 (try_peel_loop): Update.
26173 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26174 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
26175 * tree-ssa-loop-split.c (connect_loops): Update.
26176 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
26177 * tree-ssa-reassoc.c (branch_fixup): Update.
26178 * tree-ssa-tail-merge.c (replace_block_by): Update.
26179 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
26180 (compute_path_counts): Update.
26181 (update_profile): Update.
26182 (recompute_probabilities): Update.
26183 (update_joiner_offpath_counts): Update.
26184 (estimated_freqs_path): Update.
26185 (freqs_to_counts_path): Update.
26186 (clear_counts_path): Update.
26187 (ssa_fix_duplicate_block_edges): Update.
26188 (duplicate_thread_path): Update.
26189 * tree-switch-conversion.c (case_bit_test_cmp): Update.
26190 (struct switch_conv_info): Update.
26191 * tree-tailcall.c (decrease_profile): Update.
26192 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26193 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26194 * value-prof.c (check_counter): Update.
26195 (gimple_divmod_fixed_value): Update.
26196 (gimple_mod_pow2): Update.
26197 (gimple_mod_subtract): Update.
26198 (gimple_ic_transform): Update.
26199 (gimple_stringop_fixed_value): Update.
26200 * value-prof.h (gimple_ic): Update.
26202 2017-06-02 Carl Love <cel@us.ibm.com>
26204 * config/rs6000/rs6000-c: Add support for built-in functions
26205 vector double vec_doublee (vector signed int);
26206 vector double vec_doublee (vector unsigned int);
26207 vector double vec_doublee (vector float);
26208 vector double vec_doubleh (vector signed int);
26209 vector double vec_doubleh (vector unsigned int);
26210 vector double vec_doubleh (vector float);
26211 vector double vec_doublel (vector signed int);
26212 vector double vec_doublel (vector unsigned int);
26213 vector double vec_doublel (vector float);
26214 vector double vec_doubleo (vector signed int);
26215 vector double vec_doubleo (vector unsigned int);
26216 vector double vec_doubleo (vector float);.
26217 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
26218 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
26220 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
26221 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
26222 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
26224 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
26225 vec_doublel, vec_doubleh.
26226 * doc/extend.texi: Update the built-in documentation file for the
26227 new built-in functions.
26229 2017-06-02 David Malcolm <dmalcolm@redhat.com>
26232 * ipa-inline-analysis.c (free_growth_caches): Set
26233 edge_removal_hook_holder to NULL after removing it.
26235 2017-06-02 Sudakshina Das <sudi.das@arm.com>
26237 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
26238 comparision with zero.
26240 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26241 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26242 for early expansion of vec_min and vec_max builtins.
26243 (builtin_function_type): Add min/max unsigned variants to those
26244 identified as having unsigned arguments.
26246 2017-06-02 Olivier Hainque <hainque@adacore.com>
26248 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
26250 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26252 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
26253 Use VALL_F16 iterator rather than VALL.
26255 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26257 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
26258 Emit CBNZ inside loop when doing a strong exchange and comparing
26259 against zero. Generate the CC flags after the loop.
26261 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
26263 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
26264 (dl_section_ref): New.
26265 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
26266 On AIX, append an expression to subtract the size of the
26267 section length to dl_section_ref.
26269 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26271 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26272 for early expansion of vector absolute builtins.
26274 2017-06-02 Richard Biener <rguenther@suse.de>
26276 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
26277 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
26279 2017-06-02 Richard Biener <rguenther@suse.de>
26281 PR tree-optimization/80948
26282 * tree-tailcall.c (find_tail_calls): Track stmts to move in
26283 stmt order as well.
26285 2017-06-02 Richard Biener <rguenther@suse.de>
26287 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
26289 * tree-vect-stmts.c (process_use): Do not mark backedge defs
26290 for inductions as relevant.
26292 2017-06-02 Richard Biener <rguenther@suse.de>
26294 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
26295 (vectorizable_induction): ... this. Remove dead code.
26297 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
26299 * builtins. (expand_builtin_alloca): Remove second parameter and
26300 infer its value from the first parameter instead.
26301 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
26303 2017-06-02 Jakub Jelinek <jakub@redhat.com>
26305 PR rtl-optimization/80903
26306 * loop-doloop.c (add_test): Unshare sequence.
26308 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
26310 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
26312 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
26314 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
26316 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
26317 xlogue_layout::get_instance, logue_layout::xlogue_layout,
26318 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
26319 (xlogue_layout::get_stub_rtx): Make static.
26320 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
26321 (xlogue_layout::compute_stub_managed_regs): Rename to...
26322 (xlogue_layout::count_stub_managed_regs): ...this.
26323 (xlogue_layout::is_stub_managed_reg): New function.
26324 (xlogue_layout::m_stub_names): Rename to...
26325 (xlogue_layout::s_stub_names): ...this, make static.
26326 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
26327 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
26328 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
26329 xlogue_layout::s_stub_names): Instantiate statics.
26330 (stub_managed_regs): Remove.
26331 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
26332 (disable_call_ms2sysv_xlogues): Rename to...
26333 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
26334 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
26336 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
26337 change after reload_completed.
26338 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
26340 (ix86_expand_prologue): Likewise.
26341 (ix86_expand_epilogue): Likewise.
26342 (ix86_expand_split_stack_prologue): Likewise.
26343 (ix86_compute_frame_layout): Remove frame parameter ...
26344 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
26345 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
26347 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
26348 (ix86_frame): Move from here ...
26349 * config/i386/i386.h (ix86_frame): ... to here.
26350 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
26351 complete ix86_frame data structure instead. Remove some_ld_name.
26353 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26355 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
26356 symbols that hold a DECL_VALUE_EXPR.
26358 2017-06-01 Martin Jambor <mjambor@suse.cz>
26360 PR tree-optimization/80898
26361 * tree-sra.c (process_subtree_disqualification): Removed.
26362 (disqualify_candidate): Do not acll
26363 process_subtree_disqualification.
26364 (subtree_mark_written_and_enqueue): New function.
26365 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
26366 RHS has been disqualified and re-queue LHS if necessary. Apart
26367 from that, ignore disqualified RHS.
26369 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26371 * config/s390/s390.c (s390_emit_epilogue): Disable early return
26372 address fetch for z10 or later.
26374 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26376 * config/arc/arc.md (tst_movb): Add guard when splitting.
26378 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26380 * config/arc/arc.c (arc_can_eliminate): Test against
26381 arc_frame_pointer_needed.
26383 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26385 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
26386 to prevent store reordering.
26387 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
26388 (type): Add block type.
26389 (stack_tie): Define special instruction to be used in
26392 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26394 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
26395 constraint. It is not valid for the pattern.
26396 (noncommutative_binary_comparison): Likewise.
26398 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26400 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
26403 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26405 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
26406 be used by the reg-alloc.
26408 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26410 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
26411 reg-alloc when having mul64 or mul32x16 instructions.
26412 (mulsidi3): Likewise.
26413 (umulsidi3): Likewise.
26414 (mulsi32x16): New pattern.
26415 (mulsi64): Likewise.
26416 (mulsidi64): Likewise.
26417 (umulsidi64): Likewise.
26418 (MUL32x16_REG): Define.
26419 (mul64_600): Use MUL32x16_REG.
26420 (mac64_600): Likewise.
26421 (umul64_600): Likewise.
26422 (umac64_600): Likewise.
26424 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26426 * config/arc/arc.md (mulsi3_700): Make it commutative.
26428 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
26430 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
26432 (*sign_extendsidi2_insn): Likewise for movstosw.
26434 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26436 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
26437 the type of the input discriminant value. Convert the
26438 discriminant value of signedness vary.
26440 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
26442 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
26443 Add to -Wall section.
26445 2017-06-01 Richard Biener <rguenther@suse.de>
26447 PR middle-end/66313
26448 * fold-const.c (fold_plusminus_mult_expr): If the factored
26449 factor may be zero use a wrapping type for the inner operation.
26450 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
26451 and handle moved defs.
26452 (process_assignment): Properly guard the unary op case. Return a
26453 tri-state indicating that moving the stmt before the call may allow
26454 to continue. Pass through to_move.
26455 (find_tail_calls): Handle moving unrelated defs before
26458 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
26461 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
26462 splitter result in the canonical way.
26464 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
26466 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
26467 also for 32bit target. Update insn attributes.
26468 (zero-extendsidi2 splitter): Allow all registers for operand 1.
26470 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
26472 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
26473 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
26474 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
26475 (_mm_maskz_min_ss): New intrinsics.
26477 2017-05-31 Martin Liska <mliska@suse.cz>
26479 * tree-vect-loop.c (vect_create_epilog_for_reduction):
26480 Change comment style to one we normally use.
26481 (vectorizable_reduction): Likewise.
26482 (vectorizable_induction): Likewise.
26483 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
26484 (vectorizable_call): Likewise.
26485 (vectorizable_simd_clone_call): Likewise.
26486 (vectorizable_conversion): Likewise.
26487 (vectorizable_assignment): Likewise.
26488 (vectorizable_shift): Likewise.
26489 (vectorizable_operation): Likewise.
26490 (vectorizable_store): Likewise.
26491 (vectorizable_load): Likewise.
26492 * tree-vectorizer.h: Likewise.
26494 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
26496 * passes.c (emergency_dump_function): New.
26497 * tree-pass.h (emergency_dump_function): Declare.
26498 * plugin.c (plugins_internal_error_function): Remove.
26499 * plugin.h (plugins_internal_error_function): Remove declaration.
26500 * toplev.c (internal_error_function): New static function. Use it...
26501 (general_init): ...here.
26503 2017-05-31 Graham Markall <graham.markall@embecosm.com>
26505 * config/arc/arc.c (arc_print_operand): Handle constant operands.
26506 (arc_rtx_costs): Add costs for new patterns.
26507 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
26508 * config/arc/predicates.md: Add _1_2_3_operand predicate.
26510 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26512 * tree-ssa-strlen.c (get_next_strinfo): New function.
26513 (get_stridx_plus_constant): Use it.
26514 (zero_length_string): Likewise.
26515 (adjust_related_strinfos): Likewise.
26516 (adjust_last_stmt): Likewise.
26518 2017-05-31 Richard Biener <rguenther@suse.de>
26521 * config/i386/i386.c (ix86_expand_builtin): Remove assert
26522 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
26524 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26526 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
26527 loop_vinfo argument and use of dependence distance vectors.
26528 Check instead whether the two references differ only in their
26529 initial value and assume that they have the same alignment if the
26530 difference is a multiple of the vector alignment.
26531 (vect_analyze_data_refs_alignment): Update call accordingly.
26533 2017-05-31 Martin Liska <mliska@suse.cz>
26536 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
26538 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26540 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
26541 (create_intersect_range_checks): Move from ...
26542 * tree-data-ref.c (create_intersect_range_checks_index)
26543 (create_intersect_range_checks): ... to here.
26544 (create_runtime_alias_checks): New function factored from ...
26545 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
26546 here. Call above function.
26547 * tree-data-ref.h (create_runtime_alias_checks): New function.
26549 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26551 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
26552 segment length for dr_b and compute it in wide_int.
26554 2017-05-31 Richard Biener <rguenther@suse.de>
26556 PR tree-optimization/80906
26557 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
26558 and pass through iv_map.
26559 (copy_bb_and_scalar_dependences): Adjust.
26560 (translate_pending_phi_nodes): Likewise.
26561 (copy_loop_close_phi_args): Handle code-generating IVs instead
26564 2017-05-30 David Malcolm <dmalcolm@redhat.com>
26566 * diagnostic-color.c (color_dict): Add "type-diff".
26567 (parse_gcc_colors): Update comment.
26568 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
26569 -fdiagnostics-show-template-tree and -fno-elide-type.
26570 (GCC_COLORS): Add type-diff to example.
26572 (-fdiagnostics-show-template-tree): New.
26573 (-fno-elide-type): New.
26574 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
26575 the pp_format_decoder callback. Call any m_format_postprocessor's
26577 (pretty_printer::pretty_printer): Initialize
26578 m_format_postprocessor.
26579 (pretty_printer::~pretty_printer): Delete any
26580 m_format_postprocessor.
26581 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
26582 (class format_postprocessor): New class.
26583 (struct pretty_printer::format_decoder): Document the new parameters.
26584 (struct pretty_printer::m_format_postprocessor): New field.
26585 * tree-diagnostic.c (default_tree_printer): Update for new
26586 bool and const char ** params.
26587 * tree-diagnostic.h (default_tree_printer): Likewise.
26589 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
26591 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
26592 (lwa_operand): Delete rs6000_gen_cell_microcode test.
26593 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26594 rs6000_gen_cell_microcode code.
26595 (rs6000_final_prescan_insn): Delete.
26596 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
26597 "warn-cell-microcode" entries.
26598 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
26599 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
26600 throughout. Change cc_reg_not_micro_cr0_operand to
26601 cc_reg_not_cr0_operand throughout.
26602 (*extendhi<mode>2_noload): Delete.
26603 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
26604 (mwarn-cell-microcode): Delete.
26605 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
26606 -mgen-cell-microcode and -mwarn-cell-microcode.
26608 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
26611 * config/i386/constraints.md (Yd): New constraint.
26613 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
26614 and (?Yd, r) alternatives. Update insn attributes.
26615 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
26616 and (?*Yd, r) alternatives. Update insn attributes.
26617 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
26619 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
26621 * gimplify.c (gimplify_modify_expr): Don't create a
26622 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
26625 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
26627 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
26629 2017-05-30 Richard Biener <rguenther@suse.de>
26631 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
26632 and reduc_def fields.
26633 (STMT_VINFO_REDUC_TYPE): New define.
26634 (STMT_VINFO_REDUC_DEF): Likewise.
26635 (vect_force_simple_reduction): Adjust prototype.
26636 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
26637 (vect_is_simple_reduction): Remove check_reduction argument.
26638 (vect_force_simple_reduction): Adjust and set
26639 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
26640 (vectorizable_reduction): Do not re-do reduction analysis
26641 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
26642 * tree-parloops.c (gather_scalar_reductions): Adjust.
26644 2017-05-30 Richard Biener <rguenther@suse.de>
26646 PR middle-end/80901
26647 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
26650 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
26652 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
26653 Introduce unknown_misalignment parameter and remove vf.
26654 (vect_peeling_hash_get_lowest_cost):
26655 Pass unknown_misalignment parameter.
26656 (vect_enhance_data_refs_alignment):
26657 Fix unsupportable data ref treatment.
26659 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26661 * tree-vect-data-refs.c (vect_get_data_access_cost):
26662 Workaround for SLP handling.
26663 (vect_enhance_data_refs_alignment):
26664 Compute costs for doing no peeling at all, compare to the best
26665 peeling costs so far and avoid peeling if cheaper.
26667 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26669 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
26670 Return peeling info and set costs to zero for unlimited cost
26672 (vect_enhance_data_refs_alignment): Also inspect all datarefs
26673 with unknown misalignment. Compute and costs for unknown
26674 misalignment, compare them to the costs for known misalignment
26675 and choose the cheapest for peeling.
26677 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26679 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
26680 (vect_get_peeling_costs_all_drs): Create function.
26681 (vect_peeling_hash_get_lowest_cost):
26682 Use vect_get_peeling_costs_all_drs.
26683 (vect_peeling_supportable): Create function.
26684 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
26686 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26688 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
26689 DR_HAS_NEGATIVE_STEP.
26690 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
26691 (vect_enhance_data_refs_alignment): Use.
26692 (vect_duplicate_ssa_name_ptr_info): Use.
26693 * tree-vectorizer.h (dr_misalignment): Use.
26694 (known_alignment_for_access_p): Use.
26696 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
26699 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
26701 (has_section_name): New function.
26703 2017-05-30 Martin Liska <mliska@suse.cz>
26706 * auto-profile.c (get_function_decl_from_block): Fix
26709 2017-05-30 Richard Biener <rguenther@suse.de>
26711 PR middle-end/80876
26712 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
26714 2017-05-30 Martin Liska <mliska@suse.cz>
26716 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
26717 * dumpfile.h (struct dump_file_info): Remove ctors.
26719 2017-05-30 Martin Liska <mliska@suse.cz>
26721 * predict.def: Fix GNU coding style.
26723 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
26725 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
26726 Mark 'to' argument with ATTRIBUTE_UNUSED.
26728 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
26730 * config/xtensa/xtensa.c (xtensa_emit_call): Use
26731 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
26732 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
26735 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
26737 * doc/install.texi (Options specification): Restore entry of
26738 --enable-sjlj-exceptions.
26740 2017-05-27 Michael Eager <eager@eagercon.com>
26743 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
26745 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
26747 * config/microblaze/microblaze.h
26748 (FIXED_REGISTERS): Update in macro.
26749 (CALL_USED_REGISTERS): Update in macro.
26751 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
26753 * doc/install.texi: Add links to macOS binary distributions.
26755 2017-05-27 Jakub Jelinek <jakub@redhat.com>
26759 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26761 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
26763 2017-05-26 Martin Liska <mliska@suse.cz>
26765 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
26767 2017-05-26 Martin Liska <mliska@suse.cz>
26769 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
26770 always leading ';; '.
26771 (dump_bb_info): Likewise.
26772 (brief_dump_cfg): Likewise.
26773 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
26774 * dumpfile.c: Remove usage of TDF_VERBOSE.
26775 * dumpfile.h (enum dump_kind): Likewise.
26776 (dump_gimple_bb_header): Do not use TDF_COMMENT.
26777 * print-tree.c (debug_verbose): Remove.
26778 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
26779 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
26780 * tree-diagnostic.c (default_tree_printer): Replace
26781 TDF_DIAGNOSTIC with TDF_SLIM.
26783 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26785 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
26786 in parameter loop, rather than loop_vinfo.
26787 (create_intersect_range_checks): Ditto.
26788 (vect_create_cond_for_alias_checks): Update call to above functions.
26790 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26792 PR tree-optimization/80815
26793 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
26794 for merging runtime alias checks. Handle negative DR_STEPs.
26796 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26798 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
26800 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
26801 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
26802 out code pruning runtime alias checks.
26803 * tree-data-ref.c (prune_runtime_alias_test_list): New function
26804 factored out from above.
26805 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
26807 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
26809 (prune_runtime_alias_test_list): New decalaration.
26811 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26813 * tree-vect-data-refs.c (compare_tree): Rename and move ...
26814 * tree-data-ref.c (data_ref_compare_tree): ... to here.
26815 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
26816 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
26817 (operator==, comp_dr_with_seg_len_pair): Ditto.
26818 (vect_prune_runtime_alias_test_list): Ditto.
26820 2017-05-26 Martin Liska <mliska@suse.cz>
26823 * params.def: Bound partial-inlining-entry-probability param.
26825 2017-05-26 Marek Polacek <polacek@redhat.com>
26828 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
26831 2017-05-26 Richard Biener <rguenther@suse.de>
26833 PR tree-optimization/80842
26834 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
26837 2017-05-26 Richard Biener <rguenther@suse.de>
26839 PR tree-optimization/80844
26840 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
26842 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
26844 * doc/md.texi (Machine Constraints): Update x86 family
26845 machine constraints section to match 'config/i386/constraints.md'.
26847 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
26849 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
26851 2017-05-25 Nathan Sidwell <nathan@acm.org>
26853 * doc/invoke.texi (--enable-languages): Update documentation.
26855 2017-05-25 Martin Liska <mliska@suse.cz>
26857 * dumpfile.c: Add TDF_FOLDING.
26858 * dumpfile.h (enum dump_kind): Likewise.
26859 * genmatch.c (dt_simplify::gen_1): Use it.
26861 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26863 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
26865 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26867 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
26868 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
26870 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26872 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
26873 * match.pd (X == C): Rewrite it here.
26874 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
26875 with_certain_nonzero_bits2): New predicates.
26876 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
26878 2017-05-24 Nathan Sidwell <nathan@acm.org>
26880 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
26883 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
26886 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26888 * config/powerpcspe: New port. Files are copied from the rs6000
26889 port, with "rs6000" in filenames replaced by "powerpcspe".
26891 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
26893 PR rtl-optimization/80754
26894 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
26896 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
26898 * config/sparc/sparc.md (length): Return the correct value for -mflat
26899 sibcalls to match output_sibcall.
26901 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26905 * config/rs6000/rs6000.c (struct machine_function): Add new field
26907 (rs6000_get_separate_components): Init that field, use it.
26908 (rs6000_components_for_bb): Use the field.
26910 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
26912 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
26914 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
26916 PR middle-end/80823
26917 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
26919 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26922 * config/s390/s390.c (s390_check_qrst_address): Check incoming
26923 address against address_operand predicate.
26924 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
26926 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
26928 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
26929 parameters passed indirectly.
26931 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
26933 * config/i386/i386.md (*movdi_internal): Remove SSE4
26934 alternative 18 (?r, *v). Update insn attributes.
26935 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
26936 Update insn attributes.
26937 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
26938 Update insn attributes.
26939 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
26940 alternative 1 (r, v). Remove isa attribute.
26941 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
26942 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
26943 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
26945 2017-05-23 Tom de Vries <tom@codesourcery.com>
26947 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
26950 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26952 * cgraphunit.c (symbol_table::process_new_functions): Update.
26953 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
26954 (inline_generate_summary): Rename to ...
26955 (ipa_fn_summary_generate): ... this one.
26956 (inline_read_summary): Rename to ...
26957 (ipa_fn_summary_read): ... this one.
26958 (inline_write_summary): Rename to ...
26959 (ipa_fn_summary_write): ... this one.
26960 (inline_free_summary): Rename to ...
26961 (ipa_free_fn_summary): ... this one.
26962 (pass_data_local_fn_summary, pass_local_fn_summary,
26963 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
26964 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
26965 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
26966 make_pass_ipa_fn_summary): New.
26967 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
26968 inline_write_summary, inline_free_summary): Remove.
26969 (ipa_free_fn_summary) : New.
26970 * ipa-inline.c (ipa_inline): Update.
26971 (pass_ipa_inline): Do not generate summaries.
26972 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
26974 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
26975 and add pass_ipa_fn_summary.
26976 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
26978 (make_pass_inline_parameters): Remove.
26980 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
26982 * omp-low.c (struct omp_context): Remove "default_kind" member.
26985 * omp-offload.c (execute_oacc_device_lower): Remove the
26986 parallelism dimensions function attributes for unparallelized
26987 OpenACC kernels constructs.
26989 2017-05-23 Martin Liska <mliska@suse.cz>
26991 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
26993 (cgraph_edge::make_speculative): Likewise.
26994 (cgraph_edge::resolve_speculation): Likewise.
26995 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
26996 (cgraph_node::dump): Likewise.
26997 * cgraph.h: Likewise.
26998 * cgraphunit.c (analyze_functions): Likewise.
26999 (symbol_table::compile): Likewise.
27000 * ipa-cp.c (print_all_lattices): Likewise.
27001 (determine_versionability): Likewise.
27002 (initialize_node_lattices): Likewise.
27003 (ipcp_verify_propagated_values): Likewise.
27004 (estimate_local_effects): Likewise.
27005 (update_profiling_info): Likewise.
27006 (create_specialized_node): Likewise.
27007 (perhaps_add_new_callers): Likewise.
27008 (decide_about_value): Likewise.
27009 (decide_whether_version_node): Likewise.
27010 (identify_dead_nodes): Likewise.
27011 (ipcp_store_bits_results): Likewise.
27012 * ipa-devirt.c (dump_targets): Likewise.
27013 (ipa_devirt): Likewise.
27014 * ipa-icf.c (sem_item::dump): Likewise.
27015 (sem_function::equals): Likewise.
27016 (sem_variable::equals): Likewise.
27017 (sem_item_optimizer::read_section): Likewise.
27018 (sem_item_optimizer::execute): Likewise.
27019 (congruence_class::dump): Likewise.
27020 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
27021 (dump_inline_summary): Likewise.
27022 (estimate_node_size_and_time): Likewise.
27023 (inline_analyze_function): Likewise.
27024 * ipa-inline-transform.c (inline_call): Likewise.
27025 * ipa-inline.c (report_inline_failed_reason): Likewise.
27026 (want_early_inline_function_p): Likewise.
27027 (edge_badness): Likewise.
27028 (update_edge_key): Likewise.
27029 (inline_small_functions): Likewise.
27030 * ipa-profile.c (ipa_profile): Likewise.
27031 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
27032 (ipa_make_edge_direct_to_target): Likewise.
27033 (remove_described_reference): Likewise.
27034 (ipa_impossible_devirt_target): Likewise.
27035 (propagate_controlled_uses): Likewise.
27036 (ipa_print_node_params): Likewise.
27037 (ipcp_transform_function): Likewise.
27038 * ipa-pure-const.c (pure_const_read_summary): Likewise.
27039 (propagate_pure_const): Likewise.
27040 * ipa-reference.c (generate_summary): Likewise.
27041 (read_write_all_from_decl): Likewise.
27042 (propagate): Likewise.
27043 (ipa_reference_read_optimization_summary): Likewise.
27044 * ipa-utils.c (ipa_merge_profiles): Likewise.
27045 * ipa.c (walk_polymorphic_call_targets): Likewise.
27046 (symbol_table::remove_unreachable_nodes): Likewise.
27047 (ipa_single_use): Likewise.
27048 * passes.c (execute_todo): Likewise.
27049 * predict.c (drop_profile): Likewise.
27050 * symtab.c (symtab_node::get_dump_name): New function.
27051 (symtab_node::dump_name): Likewise.
27052 (symtab_node::dump_asm_name): Likewise.
27053 (symtab_node::dump_references): Likewise.
27054 (symtab_node::dump_referring): Likewise.
27055 (symtab_node::dump_base): Likewise.
27056 (symtab_node::debug_symtab): Likewise.
27057 * tree-sra.c (convert_callers_for_node): Likewise.
27058 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
27059 * value-prof.c (init_node_map): Likewise.
27061 2017-05-23 Martin Liska <mliska@suse.cz>
27063 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
27064 and symtab_node::debug_symtab to symbol_table::debug.
27065 * cgraphunit.c (analyze_functions): Use the renamed function.
27066 (symbol_table::compile): Likewise.
27067 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
27068 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
27069 * passes.c (execute_todo): Likewise.
27070 * symtab.c (symbol_table::dump): New function.
27071 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
27073 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27075 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
27076 that nonconst implies exec.
27078 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27080 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
27081 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27082 (inline_edge_summary_vec): Turn into ...
27083 (ipa_call_summaries): ... this one.
27084 (redirect_to_unreachable, edge_set_predicate,
27085 evaluate_properties_for_edge, inline_summary_alloc,
27086 reset_ipa_call_summary, reset_inline_summary,
27087 inline_summary_t::duplicate): Update.
27088 (inline_edge_duplication_hook): Turn to ...
27089 (ipa_call_summary_t::duplicate): ... this one.
27090 (inline_edge_removal_hook): Turn to ...
27091 (ipa_call_summary_t::remove): ... this one.
27092 (dump_inline_edge_summary): Turn to ...
27093 (dump_ipa_call_summary): ... this one.
27094 (estimate_function_body_sizes): Update.
27095 (inline_update_callee_summaries): Update.
27096 (remap_edge_change_prob): Update.
27097 (remap_edge_summaries): Update.
27098 (inline_merge_summary): Update.
27099 (do_estimate_edge_time): Update.
27100 (inline_generate_summary): Update.
27101 (inline_read_section): Update.
27102 (inline_read_summary): Update.
27103 (inline_free_summary): Update.
27104 * ipa-inline.c (can_inline_edge_p): Update.
27105 (compute_inlined_call_time): Update.
27106 (want_inline_small_function_p): Update.
27107 (edge_badness): Update.
27108 (early_inliner): Update.
27109 * ipa-inline.h (inline_edge_summary): Turn to ...
27110 (ipa_call_summary): ... this one.
27111 (ipa_call_summary_t): New class.
27112 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27113 (ipa_call_summaries): New.
27114 (inline_edge_summary): Remove.
27115 (estimate_edge_growth): Update.
27116 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27117 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27118 * ipa-split.c (execute_split_functions): Update.
27119 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27121 2017-05-23 Tom de Vries <tom@codesourcery.com>
27123 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
27124 attributes): Document rdrand effective target.
27126 2017-05-23 Tom de Vries <tom@codesourcery.com>
27128 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
27129 attributes): Sort alphabetically.
27131 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
27133 * config/avr/genmultilib.awk: Use gsub instead of gensub.
27135 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
27138 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
27139 V2DF/V2DI splat into two separate patterns, one that handles
27140 registers, and the other that only handles memory. Drop support
27141 for splatting from a GPR on ISA 2.07 and then splitting the
27142 splat into direct move and splat.
27143 (vsx_splat_<mode>_reg): Likewise.
27144 (vsx_splat_<mode>_mem): Likewise.
27146 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
27148 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
27150 2017-05-22 Jakub Jelinek <jakub@redhat.com>
27152 PR middle-end/80809
27153 * omp-low.c (finish_taskreg_remap): New function.
27154 (finish_taskreg_scan): If unit size of ctx->record_type
27155 is non-constant, unshare the size expression and replace
27156 decls in it with possible outer var refs.
27158 PR middle-end/80809
27159 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
27160 GOVD_SHARED rather than GOVD_PRIVATE with it.
27161 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
27162 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
27164 PR middle-end/80853
27165 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
27166 as last argument to build_outer_var_ref for pointer bases of array
27167 section reductions.
27169 2017-05-19 Martin Sebor <msebor@redhat.com>
27171 * print-tree.c (print_node): Print DECL_READ_P flag.
27173 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27175 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
27176 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
27177 * cgraph.c: Likewise.
27178 * cgraphunit.c: Likewise.
27179 * gengtype.c: Likewise.
27180 * ipa-cp.c: Likewise.
27181 * ipa-devirt.c: Likewise.
27182 * ipa-icf.c: Likewise.
27183 * ipa-predicate.c: Likewise.
27184 * ipa-profile.c: Likewise.
27185 * ipa-prop.c: Likewise.
27186 * ipa-split.c: Likewise.
27188 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
27189 edge_predicate_pool, dump_inline_hints,
27190 inline_summary::account_size_time, redirect_to_unreachable,
27191 edge_set_predicate, set_hint_predicate,
27192 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
27193 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
27194 inline_summary_t::remove, remap_hint_predicate_after_duplication,
27195 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
27196 ipa_call_summary_t::remove, initialize_growth_caches,
27197 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
27198 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
27199 mark_modified, unmodified_parm_1, unmodified_parm,
27200 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
27201 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
27202 compute_bb_predicates, will_be_nonconstant_expr_predicate,
27203 will_be_nonconstant_predicate, record_modified_bb_info,
27204 get_minimal_bb, record_modified, param_change_prob,
27205 phi_result_unknown_predicate, predicate_for_phi_result,
27206 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
27207 estimate_function_body_sizes, compute_inline_parameters,
27208 compute_inline_parameters_for_curren, pass_data_inline_parameters,
27209 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
27210 inline_update_callee_summaries, remap_edge_change_prob,
27211 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
27212 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
27213 inline_analyze_function, inline_summary_t::insert,
27214 inline_generate_summary, read_ipa_call_summary, inline_read_section,
27215 inline_read_summary, write_ipa_call_summary, inline_write_summary,
27216 inline_free_summary): Move to ipa-fnsummary.h
27217 (predicate_t): Remove.
27218 * ipa-fnsummary.c: New file.
27219 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
27220 (enum inline_hints_vals, inline_hints, agg_position_info,
27221 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
27222 inline_summaries, ipa_call_summary, ipa_call_summary_t,
27223 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
27224 dump_inline_summary, dump_inline_hints, inline_generate_summary,
27225 inline_read_summary, inline_write_summary, inline_free_summary,
27226 inline_analyze_function, initialize_inline_failed,
27227 inline_merge_summary, inline_update_overall_summary,
27228 compute_inline_parameters): Move to ipa-fnsummary.h
27229 * ipa-fnsummary.h: New file.
27230 * ipa-inline-transform.h: Include ipa-inline.h.
27231 * ipa-inline.c: LIkewise.
27233 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27235 * ipa-inline.c (edge_badness): Use inlined_time instead of
27236 inline_summaries->get.
27238 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27240 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
27242 2017-05-22 Nathan Sidwell <nathan@acm.org>
27244 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
27245 (fdump-lang): Document 'raw' option.
27246 * dumpfile.h (TDI_tu): Delete.
27247 * dumpfile.c (dump_files): Remove translation-unit.
27248 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
27250 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
27252 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
27253 command option from $(AWK) call.
27254 * config/avr/genmultilib.awk: Simplify and rewrite so that it
27255 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
27256 [FORMAT]: Remove handling of variable.
27257 * config/avr/t-multilib: Regenerate.
27259 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27261 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
27263 (dump_inline_summary): Do not print self_time.
27264 (estimate_function_body_sizes): Do not set self_time.
27265 (compute_inline_parameters): Likewise.
27266 (inline_read_section, inline_write_summary): Do not stream self_time.
27267 * ipa-inline.h (inline_summary): Drop self_time.
27269 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27271 * ipa-inline-analysis.c (account_size_time): Rename to ...
27272 (inline_summary::account_size_time): ... this one.
27273 (reset_ipa_call_summary): Turn to ...
27274 (ipa_call_summary::reset): ... this one.
27275 (reset_inline_summary): Turn to ...
27276 (inline_summary::reset): ... this one.
27277 (inline_summary_t::remove): Update.
27278 (inline_summary_t::duplicate): Update.
27279 (ipa_call_summary_t::remove): Update.
27280 (dump_inline_summary): Update.
27281 (estimate_function_body_sizes): Update.
27282 (compute_inline_parameters): Update.
27283 (estimate_node_size_and_time): Update.
27284 (inline_merge_summary): Update.
27285 (inline_update_overall_summary): Update.
27286 (inline_read_section): Update.
27287 (inline_write_summary): Update.
27288 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
27289 add account_size_time and reset member functions.
27290 (ipa_call_summary): Add reset function.
27291 * ipa-predicate.h (predicate::operator &): Constify.
27293 2017-05-22 Richard Biener <rguenther@suse.de>
27295 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
27297 2017-05-19 Jason Merrill <jason@redhat.com>
27299 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
27301 2017-05-19 Marek Polacek <polacek@redhat.com>
27304 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
27305 TYPE_OVERFLOW_WRAPS checks.
27307 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
27309 * tree-core.h (enum omp_clause_default_kind): Add
27310 "OMP_CLAUSE_DEFAULT_PRESENT".
27311 * tree-pretty-print.c (dump_omp_clause): Handle it.
27312 * gimplify.c (enum gimplify_omp_var_data): Add
27313 "GOVD_MAP_FORCE_PRESENT".
27314 (gimplify_adjust_omp_clauses_1): Map it to
27315 "GOMP_MAP_FORCE_PRESENT".
27316 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
27318 * gimplify.c (oacc_default_clause): Clarify.
27320 2017-05-19 Nathan Sidwell <nathan@acm.org>
27322 LANG_HOOK_REGISTER_DUMPS
27323 * toplev.c (general_init): Call register dump lang hook.
27324 * doc/invoke.texi: Document -fdump-lang option family.
27325 * dumpfile.c (dump_files): Remove class dump here.
27326 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
27327 * dumpfile.h (tree_dump_index): Remove TDI_class.
27328 * langhooks-def.h (lhd_register_dumps): Declare.
27329 (LANG_HOOKS_REGISTER_DUMPS): Define.
27330 (LANG_HOOKS_INITIALIZER): Add it.
27331 * langhooks.c (lhd_register_dumps): Define.
27332 * langhooks.h (struct lang_hooks): Add register_dumps.
27334 2017-05-19 Nathan Sidwell <nathan@acm.org>
27336 * context.h (context::set_passes): New.
27337 * context.c (context::context): Do not create pass manager.
27338 * toplev.c (general_init): Create pass manager here.
27340 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
27342 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
27343 use this splitter if two add or or instructions would also work for
27344 the constant we want to generate.
27346 2017-05-19 Richard Biener <rguenther@suse.de>
27349 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
27350 predicate evaluation.
27352 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27354 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
27356 * ipa-inline.c (want_inline_small_function_p): Do not cast to
27359 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27361 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
27362 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27363 (inline_edge_summary_vec): Turn into ...
27364 (ipa_call_summaries): ... this one.
27365 (redirect_to_unreachable, edge_set_predicate,
27366 evaluate_properties_for_edge, inline_summary_alloc,
27367 reset_ipa_call_summary, reset_inline_summary,
27368 inline_summary_t::duplicate): Update.
27369 (inline_edge_duplication_hook): Turn to ...
27370 (ipa_call_summary_t::duplicate): ... this one.
27371 (inline_edge_removal_hook): Turn to ...
27372 (ipa_call_summary_t::remove): ... this one.
27373 (dump_inline_edge_summary): Turn to ...
27374 (dump_ipa_call_summary): ... this one.
27375 (estimate_function_body_sizes): Update.
27376 (inline_update_callee_summaries): Update.
27377 (remap_edge_change_prob): Update.
27378 (remap_edge_summaries): Update.
27379 (inline_merge_summary): Update.
27380 (do_estimate_edge_time): Update.
27381 (inline_generate_summary): Update.
27382 (inline_read_section): Update.
27383 (inline_read_summary): Update.
27384 (inline_free_summary): Update.
27385 * ipa-inline.c (can_inline_edge_p): Update.
27386 (compute_inlined_call_time): Update.
27387 (want_inline_small_function_p): Update.
27388 (edge_badness): Update.
27389 (early_inliner): Update.
27390 * ipa-inline.h (inline_edge_summary): Turn to ...
27391 (ipa_call_summary): ... this one.
27392 (ipa_call_summary_t): New class.
27393 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27394 (ipa_call_summaries): New.
27395 (inline_edge_summary): Remove.
27396 (estimate_edge_growth): Update.
27397 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27398 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27399 * ipa-split.c (execute_split_functions): Update.
27400 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27402 2017-05-19 Richard Biener <rguenther@suse.de>
27404 PR middle-end/80764
27405 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
27407 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
27409 * config/rs6000/rs6000.c (struct machine_function): Add field
27410 fpr_is_wrapped_separately.
27411 (rs6000_get_separate_components): Use 64 components. Handle the
27412 new FPR components.
27413 (rs6000_components_for_bb): Handle the FPR components.
27414 (rs6000_emit_prologue_components): Handle the FPR components.
27415 (rs6000_emit_epilogue_components): Handle the FPR components.
27416 (rs6000_set_handled_components): Handle the FPR components.
27417 (rs6000_emit_prologue): Don't output prologue code for those FPRs
27418 that are already separately shrink-wrapped.
27419 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
27420 that are already separately shrink-wrapped.
27422 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27425 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
27428 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
27429 (define_peephole2 for Altivec d-form load): Add peepholes to catch
27430 cases where the register allocator uses a move and an offsettable
27431 memory operation to/from a FPR register on ISA 2.06/2.07.
27432 (define_peephole2 for Altivec d-form store): Likewise.
27434 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
27437 * config/i386/mmx.md (*mov<mode>_internal): Enable
27438 alternatives 11, 12, 13 and 14 also for 32bit targets.
27439 Remove alternatives 15, 16, 17 and 18.
27440 * config/i386/sse.md (vec_concatv2di): Change
27441 alternative (!x, *y) to (x, ?!*Yn).
27443 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
27445 * dumpfile.h (enum dump_kind): Remove stray comma.
27447 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27449 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
27450 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
27451 predicate::num_conditions
27452 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
27453 (CHANGED): turn into predicate::changed.
27454 (agg_position_info): Move to ipa-predicate.h
27455 (add_condition, predicate::add_clause, predicate::operator &=,
27456 predicate::or_with, predicate::evaluate, predicate::probability,
27457 dump_condition, dump_clause, predicate::dump,
27458 predicate::remap_after_duplication, predicate::remap_after_inlining,
27459 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
27460 (evaluate_conditions_for_known_args): Update.
27461 (set_cond_stmt_execution_predicate): Update.
27462 * ipa-inline.h: Include ipa-predicate.h
27463 (condition, inline_param_summary, conditions, agg_position_info,
27464 predicate): Move to ipa-predicate.h
27465 * ipa-predicate.c: New file.
27466 * ipa-predicate.h: New file.
27468 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
27470 * final.c (leaf_function_p): Check we are not in a sequence.
27472 2017-05-18 Martin Liska <mliska@suse.cz>
27474 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
27475 * dumpfile.c (dump_register): Use new enum dump_kind.
27476 (get_dump_file_name): Likewise.
27477 (dump_enable_all): Likewise.
27478 (dump_switch_p_1): Likewise.
27479 (enable_rtl_dump_file): Remove usage of TDF_RTL.
27480 * dumpfile.h (enum dump_kind): New enum type.
27481 (struct dump_file_info): Create constructor and
27482 format fields and comments.
27483 * passes.c (pass_manager::register_one_dump_file):
27485 * statistics.c (statistics_early_init): Likewise.
27486 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
27487 TDF_TREE with TDF_SLIM.
27488 (gather_memory_references_ref): Likewise.
27490 2017-05-18 Martin Liska <mliska@suse.cz>
27492 * vec.h (struct vnull): Use it.
27494 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27496 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
27497 (true_predicate, false_predicate, true_predicate_p,
27498 false_predicate_p): Remove.
27499 (single_cond_predicate, not_inlined_predicate): Turn to member function
27501 (add_condition): Update.
27502 (add_clause): Turn to...
27503 (predicate::add_clause): ... this one; update; allow passing NULL
27505 (and_predicates): Turn to ...
27506 (predicate::operator &=): ... this one.
27507 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
27508 (or_predicates): Turn to ...
27509 (predicate::or_with): ... this one.
27510 (evaluate_predicate): Turn to ...
27511 (predicate::evaluate): ... this one.
27512 (predicate_probability): Turn to ...
27513 (predicate::probability): ... this one.
27514 (dump_condition): Update.
27515 (dump_predicate): Turn to ...
27516 (predicate::dump): ... this one.
27517 (account_size_time): Update.
27518 (edge_set_predicate): Update.
27519 (set_hint_predicate): UPdate.
27520 (evaluate_conditions_for_known_args): Update.
27521 (evaluate_properties_for_edge): Update.
27522 (remap_predicate_after_duplication): Turn to...
27523 (predicate::remap_after_duplication): ... this one.
27524 (remap_hint_predicate_after_duplication): Update.
27525 (inline_summary_t::duplicate): UPdate.
27526 (dump_inline_edge_summary): Update.
27527 (dump_inline_summary): Update.
27528 (set_cond_stmt_execution_predicate): Update.
27529 (set_switch_stmt_execution_predicate): Update.
27530 (compute_bb_predicates): Update.
27531 (will_be_nonconstant_expr_predicate): Update.
27532 (will_be_nonconstant_predicate): Update.
27533 (phi_result_unknown_predicate): Update.
27534 (predicate_for_phi_result): Update.
27535 (array_index_predicate): Update.
27536 (estimate_function_body_sizes): Update.
27537 (estimate_node_size_and_time): Update.
27538 (estimate_ipcp_clone_size_and_time): Update.
27539 (remap_predicate): Rename to ...
27540 (predicate::remap_after_inlining): ... this one.
27541 (remap_hint_predicate): Update.
27542 (inline_merge_summary): Update.
27543 (inline_update_overall_summary): Update.
27544 (estimate_size_after_inlining): Update.
27545 (read_predicate): Rename to ...
27546 (predicate::stream_in): ... this one.
27547 (read_inline_edge_summary): Update.
27548 (write_predicate): Rename to ...
27549 (predicate::stream_out): ... this one.
27550 (write_inline_edge_summary): Update.
27551 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
27552 (clause_t): Turn to uint32_t
27553 (predicate): Turn to class; implement constructor and operators
27555 (size_time_entry): Update.
27556 (inline_summary): Update.
27557 (inline_edge_summary): Update.
27559 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
27561 * fold-const.c (fold_binary_loc): Move transformation...
27562 * match.pd (C - X CMP X): ... here.
27564 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
27566 * config/sparc/sparc.c (sparc_option_override): Set function
27567 alignment for -mcpu=niagara7 to 64 to match the I$ line.
27568 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
27570 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
27572 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
27574 2017-05-18 Marek Polacek <polacek@redhat.com>
27577 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
27578 (pass_ubsan::execute): Call gimple_assign_single_p instead of
27579 gimple_assign_load_p.
27581 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
27583 PR middle-end/80692
27584 * real.c (do_compare): Give decimal_do_compare preference over
27585 comparing just the signs.
27587 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
27589 * doc/md.texi (Canonicalization of Instructions): Describe the
27590 canonical form of instructions that inherently set a condition
27593 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
27595 PR middle-end/80775
27596 * tree-cfg.c: Move deletion of unreachable case statements to after
27597 the merging of consecutive case labels.
27599 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
27601 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
27602 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
27603 restoring of callee-saved registers.
27605 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
27607 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
27608 * config/visium/visium.c (single_set_and_flags): Likewise.
27609 * config/visium/visium.md (Substitutions): Likewise.
27611 2017-05-17 Martin Liska <mliska@suse.cz>
27613 * cfg.c: Introduce dump_flags_t type and
27614 use it instead of int type.
27616 * cfghooks.c: Likewise.
27617 * cfghooks.h (struct cfg_hooks): Likewise.
27618 * cfgrtl.c: Likewise.
27619 * cfgrtl.h: Likewise.
27620 * cgraph.c (cgraph_node::get_body): Likewise.
27621 * coretypes.h: Likewise.
27622 * domwalk.c: Likewise.
27623 * domwalk.h: Likewise.
27624 * dumpfile.c (struct dump_option_value_info): Likewise.
27625 (dump_enable_all): Likewise.
27626 (dump_switch_p_1): Likewise.
27627 (opt_info_switch_p): Likewise.
27628 * dumpfile.h (enum tree_dump_index): Likewise.
27629 (struct dump_file_info): Likewise.
27630 * genemit.c: Likewise.
27631 * generic-match-head.c: Likewise.
27632 * gengtype.c (open_base_files): Likewise.
27633 * gimple-pretty-print.c: Likewise.
27634 * gimple-pretty-print.h: Likewise.
27635 * graph.c (print_graph_cfg): Likewise.
27636 * graphite-scop-detection.c (dot_all_sese): Likewise.
27637 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
27638 * loop-unroll.c (report_unroll): Likewise.
27639 * passes.c (pass_manager::register_one_dump_file): Likewise.
27640 * print-tree.c: Likewise.
27641 * statistics.c: Likewise.
27642 * tree-cfg.c: Likewise.
27643 * tree-cfg.h: Likewise.
27644 * tree-dfa.c: Likewise.
27645 * tree-dfa.h: Likewise.
27646 * tree-dump.c (dump_function): Likewise.
27647 * tree-dump.h (struct dump_info): Likewise.
27648 * tree-pretty-print.c: Likewise.
27649 * tree-pretty-print.h: Likewise.
27650 * tree-ssa-live.c: Likewise.
27651 * tree-ssa-live.h: Likewise.
27652 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
27653 * tree-vect-loop.c: Likewise.
27654 * tree-vect-slp.c: Likewise.
27656 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
27657 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27659 PR tree-optimization/80457
27660 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
27661 of all arguments to a statement as scalar_to_vec operations.
27662 (vectorizable_call): Adjust call to vect_model_simple_cost for
27664 (vectorizable_conversion): Likewise.
27665 (vectorizable_assignment): Likewise.
27666 (vectorizable_shift): Likewise.
27667 (vectorizable_operation): Likewise.
27668 (vectorizable_comparison): Likewise.
27669 (vect_is_simple_cond): Record the def types for operands.
27670 (vectorizable_condition): Likewise, call vect_model_simple_cost.
27671 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
27672 for statement argument count.
27674 2017-05-16 Carl Love <cel@us.ibm.com>
27676 * config/rs6000/rs6000-c: Add support for built-in functions
27677 vector unsigned long long vec_bperm (vector unsigned long long,
27678 vector unsigned char)
27679 vector signed long long vec_mule (vector signed int,
27681 vector unsigned long long vec_mule (vector unsigned int,
27682 vector unsigned int)
27683 vector signed long long vec_mulo (vector signed int,
27685 vector unsigned long long vec_mulo (vector unsigned int,
27686 vector unsigned int)
27687 vector signed char vec_sldw (vector signed char,
27688 vector signed char,
27690 vector unsigned char vec_sldw (vector unsigned char,
27691 vector unsigned char,
27693 vector signed short vec_sldw (vector signed short,
27694 vector signed short,
27696 vector unsigned short vec_sldw (vector unsigned short,
27697 vector unsigned short,
27699 vector signed int vec_sldw (vector signed int,
27702 vector unsigned int vec_sldw (vector unsigned int,
27703 vector unsigned int,
27705 vector signed long long vec_sldw (vector signed long long,
27706 vector signed long long,
27708 vector unsigned long long vec_sldw (vector unsigned long long,
27709 vector unsigned long long,
27711 * config/rs6000/rs6000-c: Add support for built-in functions
27712 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
27713 * config/rs6000/altivec.h: Add defintion for vec_sldw.
27714 * doc/extend.texi: Update the built-in documentation for the
27715 new built-in functions.
27717 2017-05-16 Marek Polacek <polacek@redhat.com>
27721 * tree.c (save_expr): Don't fold the expression.
27723 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
27725 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
27726 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
27727 and (?*y,m). Update insn attributes.
27729 2017-05-16 Martin Liska <mliska@suse.cz>
27731 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
27732 flags argument of print_gimple_stmt, print_gimple_expr,
27733 print_generic_stmt and print_generic_expr.
27734 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
27735 * coretypes.h: Likewise.
27736 * except.c (dump_eh_tree): Likewise.
27737 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
27738 * gimple-pretty-print.h: Likewise.
27739 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
27740 (backprop::push_to_worklist): Likewise.
27741 (backprop::pop_from_worklist): Likewise.
27742 (backprop::process_use): Likewise.
27743 (backprop::intersect_uses): Likewise.
27744 (note_replacement): Likewise.
27745 * gimple-ssa-store-merging.c
27746 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
27747 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
27748 (pass_store_merging::execute): Likewise.
27749 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
27750 (ssa_base_cand_dump_callback): Likewise.
27751 (dump_incr_vec): Likewise.
27752 (replace_refs): Likewise.
27753 (replace_mult_candidate): Likewise.
27754 (create_add_on_incoming_edge): Likewise.
27755 (create_phi_basis): Likewise.
27756 (insert_initializers): Likewise.
27757 (all_phi_incrs_profitable): Likewise.
27758 (introduce_cast_before_cand): Likewise.
27759 (replace_one_candidate): Likewise.
27760 * gimplify.c (gimplify_expr): Likewise.
27761 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
27762 (set_rename): Likewise.
27763 (rename_uses): Likewise.
27764 (copy_loop_phi_nodes): Likewise.
27765 (add_close_phis_to_merge_points): Likewise.
27766 (copy_loop_close_phi_args): Likewise.
27767 (copy_cond_phi_args): Likewise.
27768 (graphite_copy_stmts_from_block): Likewise.
27769 (translate_pending_phi_nodes): Likewise.
27770 * graphite-poly.c (print_pdr): Likewise.
27771 (dump_gbb_cases): Likewise.
27772 (dump_gbb_conditions): Likewise.
27773 (print_scop_params): Likewise.
27774 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
27775 (build_cross_bb_scalars_use): Likewise.
27776 (gather_bbs::before_dom_children): Likewise.
27777 * hsa-dump.c (dump_hsa_immed): Likewise.
27778 * ipa-cp.c (print_ipcp_constant_value): Likewise.
27779 (get_replacement_map): Likewise.
27780 * ipa-inline-analysis.c (dump_condition): Likewise.
27781 (estimate_function_body_sizes): Likewise.
27782 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
27783 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
27784 * ipa-prop.c (ipa_dump_param): Likewise.
27785 (ipa_print_node_jump_functions_for_edge): Likewise.
27786 (ipa_modify_call_arguments): Likewise.
27787 (ipa_modify_expr): Likewise.
27788 (ipa_dump_param_adjustments): Likewise.
27789 (ipa_dump_agg_replacement_values): Likewise.
27790 (ipcp_modif_dom_walker::before_dom_children): Likewise.
27791 * ipa-pure-const.c (check_stmt): Likewise.
27792 (pass_nothrow::execute): Likewise.
27793 * ipa-split.c (execute_split_functions): Likewise.
27794 * omp-offload.c (dump_oacc_loop_part): Likewise.
27795 (dump_oacc_loop): Likewise.
27796 * trans-mem.c (tm_log_emit): Likewise.
27797 (tm_memopt_accumulate_memops): Likewise.
27798 (dump_tm_memopt_set): Likewise.
27799 (dump_tm_memopt_transform): Likewise.
27800 * tree-cfg.c (gimple_verify_flow_info): Likewise.
27801 (print_loop): Likewise.
27802 * tree-chkp-opt.c (chkp_print_addr): Likewise.
27803 (chkp_gather_checks_info): Likewise.
27804 (chkp_get_check_result): Likewise.
27805 (chkp_remove_check_if_pass): Likewise.
27806 (chkp_use_outer_bounds_if_possible): Likewise.
27807 (chkp_reduce_bounds_lifetime): Likewise.
27808 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
27809 (chkp_mark_completed_bounds): Likewise.
27810 (chkp_register_incomplete_bounds): Likewise.
27811 (chkp_mark_invalid_bounds): Likewise.
27812 (chkp_maybe_copy_and_register_bounds): Likewise.
27813 (chkp_build_returned_bound): Likewise.
27814 (chkp_get_bound_for_parm): Likewise.
27815 (chkp_build_bndldx): Likewise.
27816 (chkp_get_bounds_by_definition): Likewise.
27817 (chkp_generate_extern_var_bounds): Likewise.
27818 (chkp_get_bounds_for_decl_addr): Likewise.
27819 * tree-chrec.c (chrec_apply): Likewise.
27820 * tree-data-ref.c (dump_data_reference): Likewise.
27821 (dump_subscript): Likewise.
27822 (dump_data_dependence_relation): Likewise.
27823 (analyze_overlapping_iterations): Likewise.
27824 * tree-inline.c (expand_call_inline): Likewise.
27825 (tree_function_versioning): Likewise.
27826 * tree-into-ssa.c (dump_defs_stack): Likewise.
27827 (dump_currdefs): Likewise.
27828 (dump_names_replaced_by): Likewise.
27829 (dump_update_ssa): Likewise.
27830 (update_ssa): Likewise.
27831 * tree-object-size.c (pass_object_sizes::execute): Likewise.
27832 * tree-parloops.c (build_new_reduction): Likewise.
27833 (try_create_reduction_list): Likewise.
27834 (ref_conflicts_with_region): Likewise.
27835 (oacc_entry_exit_ok_1): Likewise.
27836 (oacc_entry_exit_single_gang): Likewise.
27837 * tree-pretty-print.h: Likewise.
27838 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
27839 (get_scalar_evolution): Likewise.
27840 (add_to_evolution): Likewise.
27841 (get_loop_exit_condition): Likewise.
27842 (analyze_evolution_in_loop): Likewise.
27843 (analyze_initial_condition): Likewise.
27844 (analyze_scalar_evolution): Likewise.
27845 (instantiate_scev): Likewise.
27846 (number_of_latch_executions): Likewise.
27847 (gather_chrec_stats): Likewise.
27848 (final_value_replacement_loop): Likewise.
27849 (scev_const_prop): Likewise.
27850 * tree-sra.c (dump_access): Likewise.
27851 (disqualify_candidate): Likewise.
27852 (create_access): Likewise.
27853 (reject): Likewise.
27854 (maybe_add_sra_candidate): Likewise.
27855 (create_access_replacement): Likewise.
27856 (analyze_access_subtree): Likewise.
27857 (analyze_all_variable_accesses): Likewise.
27858 (sra_modify_assign): Likewise.
27859 (initialize_constant_pool_replacements): Likewise.
27860 (find_param_candidates): Likewise.
27861 (decide_one_param_reduction): Likewise.
27862 (replace_removed_params_ssa_names): Likewise.
27863 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
27864 * tree-ssa-copy.c (dump_copy_of): Likewise.
27865 (copy_prop_visit_cond_stmt): Likewise.
27866 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
27867 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
27868 (record_equivalences_from_stmt): Likewise.
27869 * tree-ssa-dse.c (compute_trims): Likewise.
27870 (delete_dead_call): Likewise.
27871 (delete_dead_assignment): Likewise.
27872 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
27873 (forward_propagate_into_cond): Likewise.
27874 (pass_forwprop::execute): Likewise.
27875 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
27876 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
27878 (move_computations_worker): Likewise.
27879 (execute_sm): Likewise.
27880 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
27881 (remove_exits_and_undefined_stmts): Likewise.
27882 (remove_redundant_iv_tests): Likewise.
27883 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
27884 (adjust_iv_update_pos): Likewise.
27885 * tree-ssa-math-opts.c (bswap_replace): Likewise.
27886 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
27887 (value_replacement): Likewise.
27888 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
27889 * tree-ssa-pre.c (print_pre_expr): Likewise.
27890 (get_representative_for): Likewise.
27891 (create_expression_by_pieces): Likewise.
27892 (insert_into_preds_of_block): Likewise.
27893 (eliminate_insert): Likewise.
27894 (eliminate_dom_walker::before_dom_children): Likewise.
27895 (eliminate): Likewise.
27896 (remove_dead_inserted_code): Likewise.
27897 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
27898 * tree-ssa-reassoc.c (get_rank): Likewise.
27899 (eliminate_duplicate_pair): Likewise.
27900 (eliminate_plus_minus_pair): Likewise.
27901 (eliminate_not_pairs): Likewise.
27902 (undistribute_ops_list): Likewise.
27903 (eliminate_redundant_comparison): Likewise.
27904 (update_range_test): Likewise.
27905 (optimize_range_tests_var_bound): Likewise.
27906 (optimize_vec_cond_expr): Likewise.
27907 (rewrite_expr_tree): Likewise.
27908 (rewrite_expr_tree_parallel): Likewise.
27909 (linearize_expr): Likewise.
27910 (break_up_subtract): Likewise.
27911 (linearize_expr_tree): Likewise.
27912 (attempt_builtin_powi): Likewise.
27913 (attempt_builtin_copysign): Likewise.
27914 (transform_stmt_to_copy): Likewise.
27915 (transform_stmt_to_multiply): Likewise.
27916 (dump_ops_vector): Likewise.
27917 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
27918 (print_scc): Likewise.
27919 (set_ssa_val_to): Likewise.
27920 (visit_reference_op_store): Likewise.
27921 (visit_use): Likewise.
27922 (sccvn_dom_walker::before_dom_children): Likewise.
27923 (run_scc_vn): Likewise.
27924 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
27926 (expr_hash_elt::print): Likewise.
27927 (const_and_copies::pop_to_marker): Likewise.
27928 (const_and_copies::record_const_or_copy_raw): Likewise.
27929 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
27930 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
27931 (dump_predicates): Likewise.
27932 (find_uninit_use): Likewise.
27933 (warn_uninitialized_phi): Likewise.
27934 (pass_late_warn_uninitialized::execute): Likewise.
27935 * tree-ssa.c (verify_vssa): Likewise.
27936 (verify_ssa): Likewise.
27937 (maybe_optimize_var): Likewise.
27938 * tree-vrp.c (dump_value_range): Likewise.
27939 (dump_all_value_ranges): Likewise.
27940 (dump_asserts_for): Likewise.
27941 (register_edge_assert_for_2): Likewise.
27942 (vrp_visit_cond_stmt): Likewise.
27943 (vrp_visit_switch_stmt): Likewise.
27944 (vrp_visit_stmt): Likewise.
27945 (vrp_visit_phi_node): Likewise.
27946 (simplify_cond_using_ranges_1): Likewise.
27947 (fold_predicate_in): Likewise.
27948 (evrp_dom_walker::before_dom_children): Likewise.
27949 (evrp_dom_walker::push_value_range): Likewise.
27950 (evrp_dom_walker::pop_value_range): Likewise.
27951 (execute_early_vrp): Likewise.
27953 2017-05-16 Richard Biener <rguenther@suse.de>
27955 * dwarf2out.c (loc_list_from_tree_1): Do not create
27956 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
27958 2017-05-16 Richard Biener <rguenther@suse.de>
27960 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
27962 (note_variable_value_in_expr): If we resolved the decl ref
27963 do not push to the stack.
27965 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
27967 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
27968 operations in fast-math mode.
27969 (vaddq_f16): Likewise.
27970 (vmul_f16): Likewise.
27971 (vmulq_f16): Likewise.
27972 (vsub_f16): Likewise.
27973 (vsubq_f16): Likewise.
27974 * config/arm/neon.md (add<mode>3): New.
27976 (fma:<VH:mode>3): New. Also remove outdated comment.
27979 2017-05-16 Martin Liska <mliska@suse.cz>
27983 * ipa-devirt.c (warn_types_mismatch): Fix typo.
27984 (odr_types_equivalent_p): Likewise.
27986 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
27988 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
27990 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
27993 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
27994 non-interunit SSE move alternatives with '?'.
27995 (zero-extendsidi peephole2): New peephole to skip intermediate
27996 general register in SSE zero-extend sequence.
27998 2017-05-15 Jeff Law <law@redhat.com>
28000 * reorg.c (relax_delay_slots): Create a new variable to hold
28001 the temporary target rather than clobbering TARGET_LABEL.
28003 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
28004 missing argument to extract_bit_field call.
28005 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
28007 2017-05-15 Martin Liska <mliska@suse.cz>
28010 * gcc.c (process_command): Do not allow empty argument of -o option.
28012 2017-05-15 Renlin Li <renlin.li@arm.com>
28014 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
28015 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
28016 * config/aarch64/constraints.md (Usf): Add long call check.
28017 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
28018 (call_value): Likewise.
28019 (sibcall): Likewise.
28020 (sibcall_value): Likewise.
28022 (call_value_insn): New.
28023 (sibcall_insn): Update rtx pattern.
28024 (sibcall_value_insn): Likewise.
28025 (call_internal): Remove.
28026 (call_value_internal): Likewise.
28027 (sibcall_internal): Likewise.
28028 (sibcall_value_internal): Likewise.
28029 (call_reg): Likewise.
28030 (call_symbol): Likewise.
28031 (call_value_reg): Likewise.
28032 (call_value_symbol): Likewise.
28034 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
28037 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
28039 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
28041 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
28042 compatible with CCGOCmode and with CCZmode.
28044 2017-05-14 Martin Sebor <msebor@redhat.com>
28046 PR middle-end/77671
28047 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
28048 (gimple_fold_builtin_snprintf): Same.
28049 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
28050 (gimple_fold_builtin_snprintf): Same.
28051 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
28052 of character types.
28053 (is_call_safe): New function.
28054 (try_substitute_return_value): Call it.
28055 (try_simplify_call): New function.
28056 (pass_sprintf_length::handle_gimple_call): Call it.
28058 2017-05-14 Martin Sebor <msebor@redhat.com>
28060 PR middle-end/80669
28061 * builtins.c (expand_builtin_stpncpy): Simplify.
28063 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
28065 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
28066 * config/i386/i386.h
28067 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
28068 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
28069 (struct machine_function): Add new members call_ms2sysv,
28070 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
28071 (struct machine_frame_state): New fields sp_realigned and
28072 sp_realigned_offset.
28073 * config/i386/i386.c
28074 (enum xlogue_stub): New enum.
28075 (enum xlogue_stub_sets): New enum.
28076 (class xlogue_layout): New class.
28077 (struct ix86_frame): New fields stack_realign_allocate_offset,
28078 stack_realign_offset and outlined_save_offset. Modify comments to
28079 detail stack layout when using out-of-line stubs.
28080 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
28081 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
28082 -mcall-ms2sysv-xlogues.
28083 (stub_managed_regs): New static variable.
28084 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
28085 registers managed by out-of-line stub.
28086 (disable_call_ms2sysv_xlogues): New function.
28087 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
28088 m->call_ms2sysv when appropriate and compute frame layout for
28090 (sp_valid_at, fp_valid_at): New inline functions.
28091 (choose_basereg): New function.
28092 (choose_baseaddr): Add align parameter, use choose_basereg and modify
28094 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
28095 Use align parameter of choose_baseaddr to generated aligned SSE movs
28097 (pro_epilogue_adjust_stack): Modify to track
28098 machine_frame_state::sp_realigned.
28099 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
28100 (ix86_nsaved_sseregs): Likewise.
28101 (ix86_emit_save_regs): Likewise.
28102 (ix86_emit_save_regs_using_mov): Likewise.
28103 (ix86_emit_save_sse_regs_using_mov): Likewise.
28104 (get_scratch_register_on_entry): Likewise.
28105 (gen_frame_set): New function.
28106 (gen_frame_load): Likewise.
28107 (gen_frame_store): Likewise.
28108 (emit_outlined_ms2sysv_save): Likewise.
28109 (emit_outlined_ms2sysv_restore): Likewise.
28110 (ix86_expand_prologue): Modify stack re-alignment code and call
28111 emit_outlined_ms2sysv_save when appropriate.
28112 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
28113 parameter rtx_insn *insn, which allows the function to be used to only
28114 generate the notes.
28115 (ix86_expand_epilogue): Modify validity checks of frame and stack
28116 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
28117 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
28118 * config/i386/predicates.md
28119 (save_multiple): New predicate.
28120 (restore_multiple): Likewise.
28121 * config/i386/sse.md
28122 (save_multiple<mode>): New pattern.
28123 (save_multiple_realign<mode>): Likewise.
28124 (restore_multiple<mode>): Likewise.
28125 (restore_multiple_and_return<mode>): Likewise.
28126 (restore_multiple_leave_return<mode>): Likewise.
28127 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
28129 2017-05-14 Julia Koval <julia.koval@intel.com>
28131 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
28132 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
28133 (__builtin_ia32_xsetbv): New builtins.
28134 * config/i386/i386.c (ix86_expand_special_args_builtin):
28136 (ix86_expand_builtin): Special expand for new intrinsics.
28137 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
28138 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
28139 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
28141 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28143 * cfganal.c (inverted_post_order_compute): Change argument type
28145 * cfganal.h (inverted_post_order_compute): Adjust prototype.
28146 * df-core.c (rest_of_handle_df_initialize): Adjust.
28147 (rest_of_handle_df_finish): Likewise.
28148 (df_analyze_1): Likewise.
28149 (df_analyze): Likewise.
28150 (loop_inverted_post_order_compute): Change argument to be a vec *.
28151 (df_analyze_loop): Adjust.
28152 (df_get_n_blocks): Likewise.
28153 (df_get_postorder): Likewise.
28154 * df.h (struct df_d): Change field to be a vec.
28155 * lcm.c (compute_laterin): Adjust.
28156 (compute_available): Likewise.
28157 * lra-lives.c (lra_create_live_ranges_1): Likewise.
28158 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
28159 * tree-ssa-pre.c (compute_antic): Likewise.
28161 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28163 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
28164 (depth_first_search::depth_first_search): Change structure init
28165 function to this constructor.
28166 (depth_first_search::add_bb): Rename function to this member.
28167 (depth_first_search::execute): Likewise.
28168 (flow_dfs_compute_reverse_finish): Adjust.
28170 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28172 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
28173 (longest_simple_path): Likewise.
28174 * shrink-wrap.c (spread_components): Likewise.
28175 (disqualify_problematic_components): Likewise.
28176 (emit_common_heads_for_components): Likewise.
28177 (emit_common_tails_for_components): Likewise.
28178 (insert_prologue_epilogue_for_components): Likewise.
28180 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28182 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
28185 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28187 * df-core.c (df_set_blocks): Start using auto_bitmap.
28188 (df_compact_blocks): Likewise.
28189 * df-problems.c (df_rd_confluence_n): Likewise.
28190 * df-scan.c (df_insn_rescan_all): Likewise.
28191 (df_process_deferred_rescans): Likewise.
28192 (df_update_entry_block_defs): Likewise.
28193 (df_update_exit_block_uses): Likewise.
28194 (df_entry_block_bitmap_verify): Likewise.
28195 (df_exit_block_bitmap_verify): Likewise.
28196 (df_scan_verify): Likewise.
28197 * lra-constraints.c (lra_constraints): Likewise.
28198 (undo_optional_reloads): Likewise.
28199 (lra_undo_inheritance): Likewise.
28200 * lra-remat.c (calculate_gen_cands): Likewise.
28201 (do_remat): Likewise.
28202 * lra-spills.c (assign_spill_hard_regs): Likewise.
28203 (spill_pseudos): Likewise.
28204 * tree-ssa-pre.c (bitmap_set_and): Likewise.
28205 (bitmap_set_subtract_values): Likewise.
28207 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28209 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
28210 management with auto_bitmap.
28211 (fix_inter_tick): Likewise.
28212 (fix_recovery_deps): Likewise.
28213 * ira.c (add_store_equivs): Likewise.
28214 (find_moveable_pseudos): Likewise.
28215 (split_live_ranges_for_shrink_wrap): Likewise.
28216 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
28217 (rtx_reuse_manager::seen_def_p): Likewise.
28218 (rtx_reuse_manager::set_seen_def): Likewise.
28219 * print-rtl.h (class rtx_reuse_manager): Likewise.
28221 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28223 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
28225 (migrate_btr_def): Likewise.
28226 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
28227 * df-core.c (loop_post_order_compute): Likewise.
28228 (loop_inverted_post_order_compute): Likewise.
28229 * hsa-common.h: Likewise.
28230 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
28231 * init-regs.c (initialize_uninitialized_regs): Likewise.
28232 * ipa-inline.c (resolve_noninline_speculation): Likewise.
28233 (inline_small_functions): Likewise.
28234 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
28235 * ira.c (combine_and_move_insns): Likewise.
28236 (build_insn_chain): Likewise.
28237 * loop-invariant.c (find_invariants): Likewise.
28238 * lower-subreg.c (propagate_pseudo_copies): Likewise.
28239 * predict.c (tree_predict_by_opcode): Likewise.
28240 (predict_paths_leading_to): Likewise.
28241 (predict_paths_leading_to_edge): Likewise.
28242 (estimate_loops_at_level): Likewise.
28243 (estimate_loops): Likewise.
28244 * shrink-wrap.c (try_shrink_wrapping): Likewise.
28245 (spread_components): Likewise.
28246 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
28247 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
28248 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
28249 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
28250 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
28251 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
28252 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
28253 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
28254 (mark_threaded_blocks): Likewise.
28255 (thread_through_all_blocks): Likewise.
28256 * tree-ssa.c (verify_ssa): Likewise.
28257 (execute_update_addresses_taken): Likewise.
28258 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
28260 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28262 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
28264 (post_order_compute): Likewise.
28265 (inverted_post_order_compute): Likewise.
28266 (pre_and_rev_post_order_compute_fn): Likewise.
28268 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28270 * genrecog.c (int_set::int_set): Explicitly construct our
28271 auto_vec base class.
28272 * vec.h (auto_vec::auto_vec): New constructor.
28274 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28276 * bitmap.h (class auto_bitmap): New constructor taking
28277 bitmap_obstack * argument.
28279 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28281 * bitmap.h (class auto_bitmap): Change type of m_bits to
28282 bitmap_head, and adjust ctor / dtor and member operators.
28284 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
28286 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
28287 when returned register mode doesn't match original mode.
28289 2017-05-12 Jeff Law <law@redhat.com>
28290 Jakub Jelinek <jakub@redhat.com>
28292 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
28293 we look for cc setter after the compare-elim changes.
28294 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
28295 within the vector to match what compare-elim now expects.
28296 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
28297 (xorsi3_flags, one_cmplsi2_flags): Likewise.
28299 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
28300 after the compare-elim changes.
28301 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
28302 the vector to match what compare-elim now expects.
28303 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
28304 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
28305 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
28306 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
28307 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
28309 * config/visium/visium.c (single_set_and_flags): Fix where
28310 we look for cc setter after the compare-elim changes.
28311 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
28312 with the vector to match what compare-elim now expects.
28313 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
28314 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
28315 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
28316 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
28317 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
28318 (neg<mode>2_insn_set_overflow): Likewise.
28320 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
28322 PR middle-end/79794
28323 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
28324 maybe_expand_insn call, set ops[0].target. If still set after call,
28325 set alt_rtl. Add extra arg to recursive calls.
28326 (extract_bit_field): Add alt_rtl argument. Pass to
28328 * expmed.h (extract_bit_field): Fix prototype.
28329 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
28330 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
28331 to extract_bit_field_calls.
28332 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
28333 Pass alt_rtl to extract_bit_field calls.
28334 * calls.c (store_unaligned_arguments_into_psuedos)
28335 load_register_parameters): Pass extra NULL to extract_bit_field calls.
28336 * optabs.c (maybe_legitimize_operand): Clear op->target when call
28338 * optabs.h (struct expand_operand): Add target bitfield.
28340 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28342 * compare-elim.c (try_eliminate_compare): Canonicalize
28343 operation with embedded compare to
28344 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
28345 (set (reg) (operation)].
28347 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
28349 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28352 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
28353 cost of adding a carry flag for ADC instruction.
28354 [case MINUS]: Ignore the cost of subtracting a carry flag
28355 for SBB instruction.
28357 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
28359 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
28361 * config/rs6000/bmiintrin.h: New file.
28362 * config/rs6000/bmi2intrin.h: New file.
28363 * config/rs6000/x86intrin.h: New file.
28365 2017-05-12 Jeff Law <law@redhat.com>
28367 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
28370 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
28372 PR middle-end/80707
28373 * tree-cfg.c: Remove cfg edges of unreachable case statements.
28375 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28377 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28378 early expansion of vector divide builtins.
28379 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
28380 builtins identified as having unsigned arguments.
28382 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28384 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
28385 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
28386 expansion of vector logical operations (and, andc, or, xor,
28389 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28391 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
28392 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
28394 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28396 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28397 early GIMPLE expansion of vector multiplies.
28399 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28401 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
28402 TARGET_HAVE_MOVT conditional.
28403 (movt splitter): Likewise.
28405 2017-05-12 Richard Biener <rguenther@suse.de>
28407 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
28408 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28409 Fold all stmts not inplace.
28411 2017-05-12 Richard Biener <rguenther@suse.de>
28413 PR tree-optimization/80713
28414 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
28415 inserted_exprs bit for not removed stmts.
28417 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
28419 PR middle-end/69921
28420 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
28421 parallelized" attribute for parallelized OpenACC kernels.
28422 * omp-offload.c (execute_oacc_device_lower): Use it.
28424 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
28425 Set "oacc kernels" attribute.
28426 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
28427 parameter. Adjust all users.
28428 (oacc_fn_attrib_kernels_p): Remove function.
28429 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
28430 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
28431 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
28432 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
28433 assert "oacc kernels" attribute is set.
28435 2017-05-11 Carl Love <cel@us.ibm.com>
28437 * config/rs6000/rs6000-c: Add support for built-in functions
28438 vector unsigned char vec_popcnt (vector signed char)
28439 vector unsigned char vec_popcnt (vector unsigned char)
28440 vector unsigned short vec_popcnt (vector signed short)
28441 vector unsigned short vec_popcnt (vector unsigned short)
28442 vector unsigned int vec_popcnt (vector signed int)
28443 vector unsigned int vec_popcnt (vector unsigned int)
28444 vector unsigned long long vec_popcnt (vector signed long long)
28445 vector unsigned long long vec_popcnt (vector unsigned long long)
28446 vector signed long long vec_slo (vector signed long long,
28447 vector signed char)
28448 vector signed long long vec_slo (vector signed long long,
28449 vector unsigned char)
28450 vector unsigned long long vec_slo (vector unsigned long long,
28451 vector signed char)
28452 vector unsigned long long vec_slo (vector unsigned long long,
28453 vector unsigned char)
28454 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
28455 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
28456 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
28457 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
28458 * doc/extend.texi: Update the built-in documentation file for the
28459 new built-in functions.
28461 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
28463 * attribs.h (sorted_attr_string): Move machine independent
28464 functions for target clone support from the i386 port to common
28465 code. Rename ix86_function_versions to common_function_versions.
28466 Rename make_name to make_unique_name.
28467 (common_function_versions): Likewise.
28468 (make_unique_name): Likewise.
28469 (make_dispatcher_decl): Likewise.
28470 (is_function_default_version): Likewise.
28471 * attribs.c (attr_strcmp): Likewise.
28472 (sorted_attr_string): Likewise.
28473 (common_function_versions): Likewise.
28474 (make_unique_name): Likewise.
28475 (make_dispatcher_decl): Likewise.
28476 (is_function_default_version): Likewise.
28477 * config/i386/i386.c (attr_strcmp): Likewise.
28478 (sorted_attr_string): Likewise.
28479 (ix86_function_versions): Likewise.
28480 (make_name): Likewise.
28481 (make_dispatcher_decl): Likewise.
28482 (is_function_default_version): Likewise.
28483 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
28485 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28488 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
28489 Account for direct move costs for vec_construct of integer
28492 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
28495 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
28496 (UNSPEC_STX_ATOMIC): Ditto.
28497 (loaddi_via_sse): New insn.
28498 (storedi_via_sse): Ditto.
28499 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
28500 Update corresponding peephole2 patterns.
28501 (atomic_storedi_fpu): Ditto.
28503 2017-05-11 Julia Koval <julia.koval@intel.com>
28505 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
28506 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
28508 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
28509 (__builtin_ia32_rsqrt14ss_mask): New builtins.
28510 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
28512 2017-05-11 Nathan Sidwell <nathan@acm.org>
28514 * graphite-poly.c: Include dumpfile.h.
28516 * dumpfle.h (dump_function): Declare here ...
28517 * tree-dump.h (dump_function): ... not here.
28518 * dumpfile.c: #include tree-cfg.h.
28519 (dump_function): Move here from ...
28520 * tree-dump.c (dump_function): ... here.
28521 * gimplify.c: #include splay-tree.h, not tree-dump.h.
28522 * graphite-poly.c: Don't include tree-dump.h.
28523 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
28524 * print-tree.c: Likewise.
28525 * stor-layout.c: Likewise.
28526 * tree-nested.c: Likewise.
28528 * dumpfile.c (dump_start): Use TDF_FLAGS.
28529 (dump_enable_all): Fix TDF_KIND check thinko.
28531 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
28533 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
28534 array entries to represent two legal parameterizations of the
28535 overloaded __builtin_cmpb function, as represented by the
28536 P6_OV_BUILTIN_CMPB constant.
28537 (altivec_resolve_overloaded_builtin): Add special case handling
28538 for the __builtin_cmpb function, as represented by the
28539 P6_OV_BUILTIN_CMPB constant.
28540 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
28541 (BU_P6_64BIT_2): New macro.
28542 (BU_P6_OVERLOAD_2): New macro
28543 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
28544 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
28545 (CMPB): Add overload support to represent both 32-bit and 64-bit
28546 compare-bytes function.
28547 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
28548 support for TARGET_CMPB.
28549 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
28550 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28551 documentation of the __builtin_cmpb overloaded built-in function.
28553 2017-05-11 Richard Biener <rguenther@suse.de>
28555 PR tree-optimization/80705
28556 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
28557 bases are not vectorizable.
28559 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28561 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
28562 when counting register pressure.
28564 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28566 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
28567 (iv_ca_more_deps): Renamed to ...
28568 (iv_ca_compare_deps): ... this.
28569 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
28571 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28573 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
28575 (determine_group_iv_costs): ... here.
28576 (find_inv_vars_cb): Record inv var if it's not recorded before.
28578 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28580 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
28581 (get_shiftadd_cost): Ditto.
28583 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28585 * tree-ssa-address.c: Include header file.
28586 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
28588 (add_to_parts): Refactor.
28589 (addr_to_parts): New parameter. Update use of move_hint_to_base.
28590 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
28593 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28595 PR tree-optimization/53090
28596 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
28598 (extract_cond_operands): Detect condition with IV on both sides
28599 and return COMP_IV_EXPR_2.
28600 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
28601 (rewrite_use_compare): Simplify by removing call to function
28602 extract_cond_operands.
28604 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28606 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
28607 (extract_cond_operands): Detect condition comparing against non-
28608 invariant bound and return appropriate enum value.
28609 (find_interesting_uses_cond): Update use of extract_cond_operands.
28610 Handle its return value accordingly.
28611 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
28613 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28615 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
28616 nonlinear iv_use computation in loop invariant sensitive way.
28618 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28620 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
28621 (find_iv_candidates): Call relate_compare_use_with_all_cands.
28623 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28625 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
28626 (dump_cand): Support iv_cand.inv_exprs.
28627 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
28629 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
28632 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28634 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
28636 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
28637 as local function. Include necessary header files.
28638 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
28640 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28642 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
28644 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28646 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
28647 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
28648 RSHIFT_EXPR and BIT_NOT_EXPR.
28650 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28652 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
28653 (adjust_setup_cost): New parameter supporting round up adjustment.
28654 (struct address_cost_data): Delete.
28655 (force_expr_to_var_cost): Don't bound cost with spill_cost.
28656 (split_address_cost, ptr_difference_cost): Delete.
28657 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
28658 (struct ainc_cost_data): New struct.
28659 (get_address_cost_ainc): New function.
28660 (get_address_cost, get_computation_cost): Reimplement.
28661 (determine_group_iv_cost_address): Record inv_expr for all uses of
28663 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
28664 (iv_ca_has_deps): Reimplemented to ...
28665 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
28667 (iv_ca_extend): Call iv_ca_more_deps.
28669 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28671 * tree-ssa-address.c (struct mem_address): Move to header file.
28672 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
28673 * tree-ssa-address.h (struct mem_address): Move from C file.
28674 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
28676 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28678 * tree-affine.h (aff_combination_type): New interface.
28679 (aff_combination_zero_p): Remove static.
28680 (aff_combination_const_p): New interface.
28681 (aff_combination_singleton_var_p): New interfaces.
28683 2017-05-11 Richard Biener <rguenther@suse.de>
28685 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28686 Skip unreachable blocks and destinations.
28687 (eliminate): Move stmt removal and fixup ...
28688 (fini_eliminate): ... here. Skip inserted exprs.
28689 (pass_pre::execute): Move fini_pre after fini_eliminate.
28690 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
28691 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
28692 PRE to get rid of dead code that has invalid SSA form and
28693 split critical edges again.
28695 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28697 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28699 2017-05-11 Richard Biener <rguenther@suse.de>
28701 * passes.c (execute_function_todo): Verify loops if they are
28702 said to be up-to-date.
28703 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
28704 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
28706 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
28709 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
28710 handle calling assemble_external ourself.
28713 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
28714 modes with zero size. Enhance comment.
28716 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28718 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
28719 built-ins for vec_xl and vec_xst with short and char pointer
28722 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
28724 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
28725 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
28726 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
28727 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
28728 (_mm_maskz_min_round_ss): New intrinsics.
28729 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
28730 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
28731 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
28732 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
28733 (__builtin_ia32_minss_mask_round): New builtins.
28734 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28735 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28736 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
28738 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
28739 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
28741 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
28744 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
28746 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
28747 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
28748 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
28749 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
28750 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
28751 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
28752 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
28753 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28754 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
28755 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
28756 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
28757 (__builtin_ia32_mulss_mask_round): New builtins.
28758 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28759 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28760 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
28762 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
28763 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
28765 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
28768 2017-05-10 Julia Koval <julia.koval@intel.com>
28770 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
28771 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
28772 (_mm256_setr_m128i): New intrinsics.
28774 2017-05-10 Julia Koval <julia.koval@intel.com>
28776 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
28777 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
28778 (_mm_maskz_rcp14_ss): New intrinsics.
28779 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
28780 (__builtin_ia32_rcp14ss_mask): New builtins.
28781 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
28783 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
28785 PR tree-optimization/51513
28786 * tree-cfg.c (gimple_seq_unreachable_p): New function.
28787 (assert_unreachable_fallthru_edge_p): Use it.
28788 (group_case_labels_stmt): Likewise.
28789 * tree-cfg.h: Prototype it.
28790 * stmt.c: Include cfghooks.h and tree-cfg.h.
28791 (emit_case_dispatch_table) <gap_label>: New local variable.
28792 Use it to fill dispatch table gaps.
28793 Test for default_label before updating probabilities.
28794 (expand_case) <default_label>: Remove unneeded initialization.
28795 Test for unreachable default case statement and remove its edge.
28796 Set default_label accordingly.
28797 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
28799 2017-05-10 Carl Love <cel@us.ibm.com>
28801 * config/rs6000/rs6000-c: Add support for built-in functions
28802 vector signed char vec_neg (vector signed char)
28803 vector signed short int vec_neg (vector short int)
28804 vector signed int vec_neg (vector signed int)
28805 vector signed long long vec_neg (vector signed long long)
28806 vector float vec_neg (vector float)
28807 vector double vec_neg (vector double)
28808 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
28810 * config/rs6000/altivec.h: Add define for vec_neg
28811 * doc/extend.texi: Update the built-in documentation for the
28812 new built-in functions.
28814 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28816 PR tree-optimization/77644
28817 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
28819 2017-05-10 Nathan Sidwell <nathan@acm.org>
28821 * dumpfile.h (TDI_lang_all): New.
28822 (TDF_KIND): New. Renumber others
28823 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
28825 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
28827 (get_dump_file_name): Adjust suffix generation.
28828 (dump_enable_all): Use TDF_KIND.
28829 * doc/invoke.texi (-fdump-lang-all): Document.
28831 * dumpfile.h: Tabify.
28833 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
28836 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
28837 Move member access before delete.
28839 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
28841 * tree-inline.c (expand_call_inline): Split block at stmt
28844 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
28847 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
28848 are now unused after splitting mov{sf,sd}_hardfloat.
28849 (f32_lr2): Likewise.
28850 (f32_lm): Likewise.
28851 (f32_lm2): Likewise.
28852 (f32_li): Likewise.
28853 (f32_li2): Likewise.
28854 (f32_lv): Likewise.
28855 (f32_sr): Likewise.
28856 (f32_sr2): Likewise.
28857 (f32_sm): Likewise.
28858 (f32_sm2): Likewise.
28859 (f32_si): Likewise.
28860 (f32_si2): Likewise.
28861 (f32_sv): Likewise.
28862 (f32_dm): Likewise.
28863 (f32_vsx): Likewise.
28864 (f32_av): Likewise.
28865 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
28866 For movsf, order stores so the VSX stores occur before the GPR
28867 store which encourages the register allocator to use a traditional
28868 FPR instead of a GPR. For movsd, order the stores so that the GPR
28869 store comes before the VSX stores to allow the power6 to work.
28870 This is due to the power6 not having a 32-bit integer store
28871 instruction from a FPR.
28872 (movsf_hardfloat): Likewise.
28873 (movsd_hardfloat): Likewise.
28875 2017-05-09 Martin Sebor <msebor@redhat.com>
28877 PR translation/80280
28878 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
28881 PR translation/80280
28882 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
28883 data member added in r247778.
28884 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
28886 2017-05-09 Nathan Sidwell <nathan@acm.org>
28888 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
28890 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
28893 2017-05-09 Marek Polacek <polacek@redhat.com>
28895 * doc/invoke.texi: Fix typo.
28897 2017-05-09 Richard Biener <rguenther@suse.de>
28899 * tree-vrp.c (vrp_val_is_max): Adjust comment.
28900 (vrp_val_is_min): Likewise.
28901 (set_value_range_to_value): Likewise.
28902 (set_value_range_to_nonnegative): Likewise.
28903 (gimple_assign_nonzero_p): Likewise.
28904 (gimple_stmt_nonzero_p): Likewise.
28905 (vrp_int_const_binop): Likewise. Remove unreachable case.
28906 (adjust_range_with_scev): Adjust comments.
28907 (compare_range_with_value): Likewise.
28908 (extract_range_from_phi_node): Likewise.
28909 (test_for_singularity): Likewise.
28911 2017-05-09 Richard Biener <rguenther@suse.de>
28913 * tree-vrp.c (get_single_symbol): Add assert that we don't
28914 get overflowed constants as invariant part.
28915 (compare_values_warnv): Add comment before the TREE_NO_WARNING
28916 checks. Use wi::cmp instead of recursing for integer constants.
28917 (compare_values): Just ignore whether we assumed undefined
28918 overflow instead of failing the compare.
28919 (extract_range_for_var_from_comparison_expr): Add comment before the
28920 TREE_NO_WARNING sets.
28921 (test_for_singularity): Likewise.
28922 (extract_range_from_comparison): Do not disable optimization
28923 when we assumed undefined overflow.
28924 (extract_range_basic): Remove init of unused var.
28926 2017-05-09 Richard Biener <rguenther@suse.de>
28928 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
28929 (extract_range_from_multiplicative_op_1): Adjust.
28930 (extract_range_from_binary_expr_1): Use int_const_binop.
28932 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
28935 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
28936 rs6000_store_data_bypass_p in seven define_bypass directives and
28937 in several comments.
28938 * config/rs6000/rs6000-protos.h: Add prototype for
28939 rs6000_store_data_bypass_p function.
28940 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
28941 function implements slightly different (rs6000-specific) semantics
28942 than store_data_bypass_p, returning false rather than aborting
28943 with assertion error when arguments do not satisfy the
28944 requirements of store data bypass.
28945 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
28946 rs6000_store_data_bypass_p.
28948 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
28950 * config/xtensa/xtensa-protos.h
28951 (xtensa_initial_elimination_offset): New declaration.
28952 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28953 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
28954 macro definition, add case for FRAME_POINTER_REGNUM when
28955 FRAME_GROWS_DOWNWARD.
28956 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
28957 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
28958 xtensa_initial_elimination_offset.
28960 2017-05-08 Nathan Sidwell <nathan@acm.org>
28962 * doc/invoke.texi: Alphabetize -fdump options.
28964 2017-05-08 Martin Sebor <msebor@redhat.com>
28966 PR translation/80280
28967 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
28969 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
28971 * target.def (compute_frame_layout): New optional target hook.
28972 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
28973 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
28974 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
28976 * reload1.c (verify_initial_elim_offsets): Likewise.
28977 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
28978 (use_simple_return_p): Call arm_compute_frame_layout if needed.
28979 (arm_get_frame_offsets): Split up into this ...
28980 (arm_compute_frame_layout): ... and this function.
28982 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
28984 * config/aarch64/constraints.md (Usa): New constraint.
28985 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
28987 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
28989 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
28990 with_multilib_list after it has been checked.
28992 2017-05-08 Richard Biener <rguenther@suse.de>
28994 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
28995 (bitmap_set_subtract_values): Likewise.
28997 2017-05-08 Richard Biener <rguenther@suse.de>
28999 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
29000 (gimple_assign_nonzero): ... this and remove strict_overflow_p
29002 (gimple_stmt_nonzero_warnv_p): Rename to ...
29003 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
29005 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
29006 (extract_range_basic): Adjust, do not disable propagation on
29007 strict overflow sensitive simplification.
29008 (vrp_visit_cond_stmt): Likewise.
29010 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
29012 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
29013 body size unconditionally.
29015 2017-05-07 Jeff Law <law@redhat.com>
29018 2017-05-06 Jeff Law <law@redhat.com>
29019 PR tree-optimization/78496
29020 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
29023 PR tree-optimization/78496
29024 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
29025 (simplify_stmt_using_ranges): Call it.
29026 (vrp_dom_walker::before_dom_children): Extract equivalences
29027 from an ASSERT_EXPR with an equality comparison against a
29030 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 Richard Sandiford <richard.sandiford@linaro.org>
29045 * lra-constraints.c (lra_copy_reg_equiv): New function.
29046 (split_reg): Use it to copy equivalence information from the
29047 original register to the spill register.
29049 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
29051 PR rtl-optimization/75964
29052 * simplify-rtx.c (simplify_const_relational_operation): Remove
29053 invalid handling of comparisons of integer ABS.
29055 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
29057 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
29058 initialize to zero.
29059 (init_regs): Remove declaration.
29060 (function_arg_advance_32): Initialize error_p as boolean variable.
29062 2017-05-05 Nathan Sidwell <nathan@acm.org>
29064 * store-motion.c (remove_reachable_equiv_notes): Reformat long
29065 lines. Use for (;;).
29067 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29069 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
29070 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
29071 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
29072 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
29073 VF=2 that require versioning.
29075 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29077 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
29080 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29082 * diagnostic.h (diagnostic_override_option_index): Convert from
29083 macro to inline function.
29085 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29087 * diagnostic.c (last_module_changed_p): New function.
29088 (set_last_module): New function.
29089 (diagnostic_report_current_module): Convert macro usage to
29090 the above functions.
29091 * diagnostic.h (diagnostic_context::last_module): Strengthen
29092 from const line_map * to const line_map_ordinary *.
29093 (diagnostic_last_module_changed): Delete macro.
29094 (diagnostic_set_last_module): Delete macro.
29096 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29098 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
29099 with diagnostic_report_diagnostic.
29100 (diagnostic_n_impl_richloc): Likewise.
29101 * diagnostic.h (report_diagnostic): Delete macro.
29102 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
29103 with diagnostic_report_diagnostic.
29104 * substring-locations.c (format_warning_va): Likewise.
29106 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29108 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
29109 save/restor of format_spec. Move option-printing code to...
29110 (print_option_information): ...this new function, and
29111 reimplement by simply printing to the pretty_printer,
29112 rather than appending to the format string.
29114 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29116 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
29117 handling logic into...
29118 (update_effective_level_from_pragmas): ...this new function.
29120 2017-05-04 Andrew Waterman <andrew@sifive.com>
29122 * config/riscv/riscv.opt (mstrict-align): New option.
29123 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
29124 (SLOW_UNALIGNED_ACCESS): Define.
29125 (riscv_slow_unaligned_access): Declare.
29126 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
29128 (riscv_slow_unaligned_access): New variable.
29129 (rocket_tune_info): Set slow_unaligned_access to true.
29130 (optimize_size_tune_info): Set slow_unaligned_access to false.
29131 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
29132 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
29133 (riscv_option_override): Set riscv_slow_unaligned_access.
29134 * doc/invoke.texi: Add -mstrict-align to RISC-V.
29136 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
29138 * config/riscv/riscv.md: Unify indentation.
29140 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29145 * config/rs6000/rs6000.md (u code attribute): Add FIX and
29147 (extendsi<mode>2): Add support for doing sign extension via
29148 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
29149 don't have ISA 3.0 instructions.
29150 (extendsi<mode>2 splitter): Likewise.
29151 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
29152 generate the normal insns since SImode can now go in vector
29153 registers. Disallow the special UNSPECs needed for previous
29154 machines to hide SImode being used. Add new insns
29155 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
29156 (fix_trunc<mode>si2_stfiwx): Likewise.
29157 (fix_trunc<mode>si2_internal): Likewise.
29158 (fixuns_trunc<mode>si2): Likewise.
29159 (fixuns_trunc<mode>si2_stfiwx): Likewise.
29160 (fctiw<u>z_<mode>_smallint): Likewise.
29161 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
29162 of floating point to 32-bit integer from doing a direct move to
29163 the GPR registers to do a store.
29164 (fctiwz_<mode>): Break long line.
29166 2017-05-05 Bin Cheng <bin.cheng@arm.com>
29168 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
29169 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
29170 (addr_list, addr_offset_valid_p): New.
29171 (split_address_groups): Check offset validity with above function.
29172 (gt-tree-ssa-loop-ivopts.h): Include header file.
29174 2017-05-05 Nathan Sidwell <nathan@acm.org>
29176 * config.gcc (arm*-*-*): Add missing 'fi'.
29178 2017-05-05 Steve Ellcey <sellcey@cavium.com>
29180 * doc/invoke.texi (-fopt-info): Explicitly say order of options
29181 included in -fopt-info does not matter.
29182 * doc/optinfo.texi (-fopt-info): Fix description of default
29183 behavour. Explicitly say order of options included in -fopt-info
29186 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29188 * config.gcc: Allow combinations of aprofile and rmprofile values for
29189 --with-multilib-list.
29190 * config/arm/t-multilib: New file.
29191 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
29192 variables. Remove setting of ISA and floating-point ABI in
29193 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
29194 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
29195 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
29196 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
29198 * config/arm/t-rmprofile: Likewise except for the matches changes.
29199 * doc/install.texi (--with-multilib-list): Document the combination of
29200 aprofile and rmprofile values and warn about pitfalls in doing that.
29202 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29204 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
29205 (movdi_aarch64): Likewise.
29207 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29209 PR tree-optimization/80632
29210 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
29212 (build_arrays): Initialize it for virtual phis.
29213 (fix_phi_nodes): Use it for virtual phis.
29215 PR tree-optimization/80558
29216 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
29217 [x, y] op z into [x op, y op z] for op & or | if conditions
29220 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
29221 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29224 * config/arm/arm.md (use_literal_pool): Remove.
29225 (64-bit immediate split): No longer takes cost into consideration
29226 if arm_disable_literal_pool is enabled.
29227 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
29228 used when arm_disable_literal_pool is enabled.
29229 (arm_max_const_double_inline_cost): Remove use of
29230 arm_disable_literal_pool.
29231 (push_minipool_fix): Add assert.
29232 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
29233 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
29234 (no_literal_pool_sf_immediate): New.
29236 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29238 PR tree-optimization/80613
29239 * tree-ssa-dce.c (propagate_necessity): Remove cases for
29240 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
29242 2017-05-05 Richard Biener <rguenther@suse.de>
29244 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
29246 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
29248 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
29249 of this flag from insn conditions due to removal from r247495.
29251 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29253 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
29255 (arm_early_store_addr_dep_ptr): Likewise.
29256 * config/arm/aarch-common-protos.h
29257 (arm_early_load_addr_dep_ptr): Add prototype.
29258 (arm_early_store_addr_dep_ptr): Likewise.
29259 * config/arm/cortex-a53.md: Add new bypasses.
29261 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29263 * tree.c (next_type_uid): Change type to unsigned.
29264 (type_hash_canon): Decrement back next_type_uid if
29265 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
29266 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
29269 2017-05-04 Martin Sebor <msebor@redhat.com>
29271 * builtins.c: Fix a trivial typo in a comment.
29273 PR middle-end/79234
29274 * builtins.c (check_sizes): Adjust to handle reading past the end.
29275 Avoid printing excessive upper bound of ranges. Use %E to print
29276 tree nodes instead of converting them to %wu.
29277 (expand_builtin_memchr): New function.
29278 (compute_dest_size): Rename...
29279 (compute_objsize): ...to this.
29280 (expand_builtin_memcpy): Adjust.
29281 (expand_builtin_mempcpy): Adjust.
29282 (expand_builtin_strcat): Adjust.
29283 (expand_builtin_strcpy): Adjust.
29284 (check_strncat_sizes): Adjust.
29285 (expand_builtin_strncat): Adjust.
29286 (expand_builtin_strncpy): Adjust and simplify.
29287 (expand_builtin_memset): Adjust.
29288 (expand_builtin_bzero): Adjust.
29289 (expand_builtin_memcmp): Adjust.
29290 (expand_builtin): Handle memcmp.
29291 (maybe_emit_chk_warning): Check strncat just once.
29293 2017-05-04 Martin Sebor <msebor@redhat.com>
29295 PR preprocessor/79214
29296 PR middle-end/79222
29297 PR middle-end/79223
29298 * builtins.c (check_sizes): Add inlining context and issue
29299 warnings even when -Wno-system-headers is set.
29300 (check_strncat_sizes): Same.
29301 (expand_builtin_strncat): Same.
29302 (expand_builtin_memmove): New function.
29303 (expand_builtin_stpncpy): Same.
29304 (expand_builtin): Handle memmove and stpncpy.
29306 2017-05-04 Bin Cheng <bin.cheng@arm.com>
29308 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
29309 which is not used any more.
29311 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29313 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
29315 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29317 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
29318 (cortexa53_tunings): Likewise.
29319 (cortexa57_tunings): Likewise.
29320 (cortexa72_tunings): Likewise.
29321 (cortexa73_tunings): Likewise.
29323 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29325 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
29326 Set loop alignment to 8.
29328 2017-05-04 Martin Sebor <msebor@redhat.com>
29330 PR translation/80280
29331 * builtins.c (expand_builtin_object_size): Add missing quoting to
29332 %D and like directives.
29333 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
29334 (hsa_type_for_tree_type): Same.
29335 (verify_function_arguments): Same.
29336 * symtab.c (symbol_table::change_decl_assembler_name): Same.
29337 * varasm.c (get_section): Same.
29340 2017-05-04 Martin Sebor <msebor@redhat.com>
29342 PR translation/80280
29343 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
29345 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29347 * config/aarch64/aarch64.c (generic_addrcost_table):
29348 Change HI/TI mode setting.
29350 2017-05-04 Martin Jambor <mjambor@suse.cz>
29352 PR tree-optimization/80622
29353 * tree-sra.c (comes_initialized_p): New function.
29354 (build_accesses_from_assign): Only set write lazily when
29355 comes_initialized_p is false.
29356 (analyze_access_subtree): Use comes_initialized_p.
29357 (propagate_subaccesses_across_link): Assert !comes_initialized_p
29358 instead of testing for PARM_DECL.
29360 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29362 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
29363 constraint on operand 0 to allow more general addressing modes.
29364 Adjust output template.
29365 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
29367 * config/aarch64/aarch64-protos.h
29368 (aarch64_address_valid_for_prefetch_p): Declare prototype.
29369 * config/aarch64/constraints.md (Dp): New address constraint.
29370 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
29373 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
29375 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
29376 update use of estimate_ipcp_clone_size_and_time.
29377 (estimate_local_effects): Update use of
29378 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
29379 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
29380 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
29381 Return nonspecialized time.
29383 2017-05-04 Richard Biener <rguenther@suse.de>
29385 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
29386 for the last VUSE which def dominates the PHI. Directly call
29388 (get_continuation_for_phi_1): Remove.
29390 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
29392 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
29393 to explain the use of truncating division. Cap the number of
29394 iterations to the maximum given by nb_iterations_upper_bound,
29397 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
29399 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
29400 * configure: Regenerate.
29401 * config.in: Regenerate.
29402 * config/i386/driver-mingw32.c: new file.
29403 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
29404 * config.host: Link driver-mingw32.o on MinGW host.
29405 * doc/install.texi: Document new --enable-mingw-wildcard configure
29408 2017-05-04 Marek Polacek <polacek@redhat.com>
29410 PR tree-optimization/80612
29411 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
29413 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29414 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
29416 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
29417 (movt splitter): Likewise.
29418 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
29419 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
29420 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
29421 block for Thumb-1 with MOVT.
29422 (thumb2_legitimate_address_p): Move code block ...
29423 (can_avoid_literal_pool_for_label_p): ... into this new function.
29424 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
29426 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
29427 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
29428 "M-profile targets with the MOVT instruction".
29430 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29432 * config/arm/arm-builtins.c (arm_init_builtins): Rename
29433 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
29434 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
29436 2017-05-04 Martin Liska <mliska@suse.cz>
29438 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
29439 variable cond_code.
29441 2017-05-04 Richard Biener <rguenther@suse.de>
29443 * tree.c (array_at_struct_end_p): Handle arrays at struct
29444 end with flexarrays more conservatively. Refactor and treat
29445 arrays of arrays or aggregates more strict. Fix
29446 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
29447 * tree.c (array_at_struct_end_p): Adjust prototype.
29448 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
29449 * gimple-fold.c (get_range_strlen): Likewise.
29450 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
29452 2017-05-04 Richard Biener <rguenther@suse.de>
29454 PR tree-optimization/31130
29455 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
29457 (supports_overflow_infinity): Likewise.
29458 (is_negative_overflow_infinity): Likewise.
29459 (is_positive_overflow_infinity): Likewise.
29460 (is_overflow_infinity): Likewise.
29461 (stmt_overflow_infinity): Likewise.
29462 (overflow_infinity_range_p): Likewise.
29463 (usable_range_p): Remove as always returning true.
29464 (make_overflow_infinity): Remove.
29465 (negative_overflow_infinity): Likewise.
29466 (positive_overflow_infinity): Likewise.
29467 (avoid_overflow_infinity): Likewise.
29468 (set_value_range): Adjust accordingly.
29469 (set_value_range_to_nonnegative): Likewise, remove now unused
29470 overflow_infinity arg.
29471 (vrp_operand_equal_p): Adjust.
29472 (update_value_range): Likewise.
29473 (range_int_cst_singleton_p): Likewise.
29474 (operand_less_p): Likewise.
29475 (compare_values_warnv): Likewise.
29476 (extract_range_for_var_from_comparison_expr): Likewise.
29477 (vrp_int_const_binop): Likewise.
29478 (zero_nonzero_bits_from_vr): Likewise.
29479 (extract_range_from_multiplicative_op_1): Likewise.
29480 (extract_range_from_binary_expr_1): Likewise.
29481 (extract_range_from_unary_expr): Likewise.
29482 (extract_range_from_comparison): Likewise.
29483 (extract_range_basic): Likewise.
29484 (adjust_range_with_scev): Likewise.
29485 (compare_ranges): Likewise.
29486 (compare_range_with_value): Likewise.
29487 (dump_value_range): Likewise.
29488 (test_for_singularity): Likewise, remove strict_overflow_p parameter
29490 (simplify_cond_using_ranges): Adjust.
29492 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
29494 * brig-builtins.def: Added a builtin for class_f64.
29495 * builtin-types.def: Added a builtin type needed by class_f64.
29497 2017-05-03 Jason Merrill <jason@redhat.com>
29499 * timevar.def: Add TV_CONSTEXPR.
29501 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29503 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
29505 2017-05-03 Martin Jambor <mjambor@suse.cz>
29507 * ipa-prop.c (ipa_update_after_lto_read): Removed.
29508 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
29509 * ipa-cp.c (ipcp_propagate_stage): Do not call
29510 ipa_update_after_lto_read.
29511 * ipa-inline.c (ipa_inline): Likewise.
29513 2017-05-03 Martin Jambor <mjambor@suse.cz>
29515 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
29516 tag. Added a default constructor and a destructor.
29517 (ipa_edge_args_sum_t): New class;
29518 (ipa_edge_args_sum): Declare.
29519 (ipa_edge_args_vector): Remove declaration.
29520 (IPA_EDGE_REF): Use ipa_edge_args_sum.
29521 (ipa_free_edge_args_substructures): Remove declaration.
29522 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
29523 (ipa_edge_args_info_available_for_edge_p): Likewise.
29524 * ipa-prop.c (ipa_edge_args_vector): Removed.
29525 (edge_removal_hook_holder): Likewise.
29526 (edge_duplication_hook_holder): Likewise.
29527 (ipa_edge_args_sum): New variable.
29528 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
29529 ipa_edge_args_vector.
29530 (ipa_free_edge_args_substructures): Likewise.
29531 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
29532 ipa_edge_args_vector.
29533 (ipa_edge_removal_hook): Turned into method
29534 ipa_edge_args_sum_t::remove.
29535 (ipa_edge_duplication_hook): Turned into method
29536 ipa_edge_args_sum_t::duplicate.
29537 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
29538 registering edge hooks.
29539 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
29540 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
29541 ipa_edge_args_sum instead of ipa_edge_args_vector.
29542 * ipa-profile.c (ipa_profile): Likewise.
29544 2017-05-03 Martin Jambor <mjambor@suse.cz>
29546 * symbol-summary.h (function_summary): New method exists.
29547 (function_summary::symtab_removal): Deallocate through release.
29548 (call_summary): New class.
29549 (gt_ggc_mx): New overload.
29550 (gt_pch_nx): Likewise.
29551 (gt_pch_nx): Likewise.
29553 2017-05-03 Jeff Law <law@redhat.com>
29555 PR tree-optimization/78496
29556 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
29557 from simplify_cond_using_ranges. Split off code to walk
29558 backwards through casts into ...
29559 (simplify_cond_using_ranges_2): New function.
29560 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
29561 (execute_vrp): After identifying jump threads, call
29562 simplify_cond_using_ranges_2.
29564 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
29567 * ipa-inline.h (inline_summary): Add ctor.
29568 (create_ggc): Do not use ggc_cleared_alloc.
29570 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
29571 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29573 * gcc.c (handle_braces): Support escaping in switch matching
29575 * doc/invoke.texi (Spec Files): Document it.
29576 Remove superfluous @code markup in items.
29578 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29580 * diagnostic-show-locus.c (struct column_range): New struct.
29581 (get_affected_columns): New function.
29582 (get_printed_columns): New function.
29583 (struct correction): New struct.
29584 (correction::ensure_capacity): New function.
29585 (correction::ensure_terminated): New function.
29586 (struct line_corrections): New struct.
29587 (line_corrections::~line_corrections): New dtor.
29588 (line_corrections::add_hint): New function.
29589 (layout::print_trailing_fixits): Reimplement in terms of the new
29591 (selftest::test_overlapped_fixit_printing): New function.
29592 (selftest::diagnostic_show_locus_c_tests): Call it.
29594 2017-05-03 Nathan Sidwell <nathan@acm.org>
29596 Canonicalize canonical type hashing
29597 * tree.h (type_hash_canon_hash): Declare.
29598 * tree.c (type_hash_list, attribute_hash_list): Move into
29599 type_hash_canon_hash.
29600 (build_type_attribute_qual_variant): Break out hash code calc into
29601 type_hash_canon_hash.
29602 (type_hash_canon_hash): New. Generic type hash computation.
29603 (build_range_type_1, build_array_type_1, build_function_type,
29604 build_method_type_directly, build_offset_type, build_complex_type,
29605 make_vector_type): Call it.
29607 2017-05-03 Richard Biener <rguenther@suse.de>
29609 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
29610 When all DRs have unknown misaligned do not always peel
29611 when there is a store but apply the same costing model as if
29612 there were only loads.
29614 2017-05-03 Richard Biener <rguenther@suse.de>
29617 PR tree-optimization/80492
29618 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
29619 compare_base_decls returning dont-know properly.
29621 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
29623 * config/arm/iterators.md (CCSI): New mode iterator.
29624 (arch): New mode attribute.
29625 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
29626 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
29627 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
29628 code iterator for success result mode.
29629 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
29630 the corresponding new insn generators.
29632 2017-05-03 Bin Cheng <bin.cheng@arm.com>
29635 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29636 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29638 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
29640 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
29641 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
29642 (DDR_A): Wrap DDR argument in brackets.
29643 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
29644 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
29645 (DDR_REVERSED_P): Likewise.
29647 2017-05-03 Jakub Jelinek <jakub@redhat.com>
29649 PR tree-optimization/79472
29650 * tree-switch-conversion.c (struct switch_conv_info): Add
29651 contiguous_range and default_case_nonstandard fields.
29652 (collect_switch_conv_info): Compute contiguous_range and
29653 default_case_nonstandard fields, don't clear final_bb if
29654 contiguous_range and only the default case doesn't have the required
29656 (check_all_empty_except_final): Set default_case_nonstandard instead
29657 of failing if contiguous_range and the default case doesn't have empty
29659 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
29660 and only the default case doesn't have the required constants. Skip
29662 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
29663 if default_case_nonstandard.
29664 (build_constructors): Build constant 1 just once. Assert that default
29665 values aren't inserted in between cases if contiguous_range. Skip
29667 (build_arrays): Skip virtual phis.
29668 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
29669 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
29670 Handle virtual phis.
29671 (gen_inbound_check): Handle default_case_nonstandard case.
29672 (process_switch): Adjust check_final_bb caller. Call
29673 gather_default_values with the first non-default case instead of
29674 default case if default_case_nonstandard.
29676 2017-05-02 Nathan Sidwell <nathan@acm.org>
29678 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
29679 check. Fix formatting.
29681 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
29683 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
29684 errors when comparing specialized and unspecialized times.
29686 2017-05-02 David Malcolm <dmalcolm@redhat.com>
29688 * diagnostic-show-locus.c
29689 (layout::should_print_annotation_line_p): Make private.
29690 (layout::print_annotation_line): Make private.
29691 (layout::annotation_line_showed_range_p): Make private.
29692 (layout::show_ruler): Make private.
29693 (layout::print_source_line): Make private. Pass in line and
29694 line_width, rather than calling location_get_source_line. Drop
29696 (layout::print_leading_fixits): New method.
29697 (layout::print_any_fixits): Rename to...
29698 (layout::print_trailing_fixits): ...this, and make private.
29699 Don't print newline fixits.
29700 (diagnostic_show_locus): Move logic for printing one row into...
29701 (layout::print_line): ...this new function. Move the
29702 location_get_source_line call and error-handling from
29703 print_source_line to here. Call print_leading_fixits, and rename
29704 print_any_fixits to print_trailing_fixits.
29705 (selftest::test_fixit_insert_containing_newline): Update now that
29706 newlines are partially supported.
29707 (selftest::test_fixit_insert_containing_newline_2): New test.
29708 (selftest::test_fixit_replace_containing_newline): Update comments.
29709 (selftest::diagnostic_show_locus_c_tests): Call the new test.
29710 * edit-context.c (class added_line): New class.
29711 (class edited_line): Describe newline handling in comment.
29712 (edited_line::actually_edited_p): New method.
29713 (edited_line::print_content): Delete redundant decl.
29714 (edited_line::m_predecessors): New field.
29715 (edited_file::print_content): Call edited_line::print_content.
29716 (edited_file::print_diff): Update to support newlines.
29717 (edited_file::print_diff_hunk): Likewise.
29718 (edited_file::print_run_of_changed_lines): New function.
29719 (edited_file::print_diff_line): Convert to...
29720 (print_diff_line): ...this.
29721 (edited_file::get_effective_line_count): New function.
29722 (edited_line::edited_line): Initialize new field m_predecessors.
29723 (edited_line::~edited_line): Clean up m_predecessors.
29724 (edited_line::apply_fixit): Handle newlines.
29725 (edited_line::get_effective_line_count): New function.
29726 (edited_line::print_content): New function.
29727 (edited_line::print_diff_lines): New function.
29728 (selftest::test_applying_fixits_insert_containing_newline): New
29730 (selftest::test_applying_fixits_replace_containing_newline): New
29732 (selftest::insert_line): New function.
29733 (selftest::test_applying_fixits_multiple_lines): Add example of
29735 (selftest::edit_context_c_tests): Call the new tests.
29737 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29739 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
29740 parameter cand. Update dump information.
29741 (get_computation_cost): Update uses.
29743 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29745 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
29746 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
29747 (get_computation_at, rewrite_use_address): Update use of
29748 get_computation_aff.
29750 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29752 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
29753 (get_computation): Delete.
29754 (get_computation_cost): Implement like get_computation_cost_at.
29755 Use get_computation_at.
29756 (get_computation_cost_at): Delete.
29757 (rewrite_use_nonlinear_expr): Use get_computation_at.
29758 (rewrite_use_compare, remove_unused_ivs): Ditto.
29760 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29762 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
29764 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29766 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
29767 (ivopts_global_cost_for_size): Rename parameter and update uses.
29768 (iv_ca_recount_cost): Update uses.
29769 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
29770 candidates seperately in n_invs and n_cands.
29771 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
29773 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29775 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
29776 (find_inv_vars_cb): New.
29777 (find_depends): Renamed to ...
29778 (find_inv_vars): ... this.
29779 (add_candidate_1, force_var_cost): Call find_inv_vars.
29780 (split_address_cost, determine_group_iv_cost_cond): Ditto.
29782 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29784 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
29785 inv_vars. Add inv_exprs.
29786 (struct iv_cand): Rename depends_on to inv_vars.
29787 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
29788 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
29789 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
29790 (dump_cand): Dump inv_vars.
29791 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
29792 (record_invariant, find_depends, add_candidate_1): Ditto.
29793 (set_group_iv_cost, force_var_cost): Ditto.
29794 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
29795 (get_computation_cost_at, get_computation_cost): Ditto.
29796 (determine_group_iv_cost_generic): Ditto.
29797 (determine_group_iv_cost_address): Ditto.
29798 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
29799 (determine_group_iv_costs): Ditto.
29800 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
29801 (iv_ca_set_remove_invariants): Renamed to ...
29802 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
29803 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
29804 (iv_ca_set_add_invariants): Renamed to ...
29805 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
29806 (iv_ca_set_cp): Use iv_ca_set_add_invs.
29807 (iv_ca_has_deps): Support inv_vars and inv_exprs.
29808 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
29809 (create_new_ivs): Remove useless dump.
29811 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29813 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
29815 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
29816 (iv_ca_set_no_cp, create_new_iv): Ditto.
29818 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29820 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29822 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29824 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
29825 function tree_check2.
29827 2017-05-02 Martin Liska <mliska@suse.cz>
29829 * doc/gcov.texi: Add missing preposition.
29830 * gcov.c (function_info::function_info): Properly fill up
29831 all member variables.
29833 2017-05-02 Tamar Christina <tamar.christina@arm.com>
29835 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
29837 2017-05-02 Tamar Christina <tamar.christina@arm.com>
29839 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
29841 2017-05-02 Martin Liska <mliska@suse.cz>
29844 * lto-streamer-in.c (lto_read_tree_1): Remove
29845 LTO_STREAMER_DEBUG.
29846 * lto-streamer.c (struct tree_hash_entry): Likewise.
29847 (struct tree_entry_hasher): Likewise.
29848 (tree_entry_hasher::hash): Likewise.
29849 (tree_entry_hasher::equal): Likewise.
29850 (lto_streamer_init): Likewise.
29851 (lto_orig_address_map): Likewise.
29852 (lto_orig_address_get): Likewise.
29853 (lto_orig_address_remove): Likewise.
29854 * lto-streamer.h: Likewise.
29855 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
29856 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
29858 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
29860 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
29861 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
29862 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
29863 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
29864 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
29865 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
29866 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
29867 (mm_maskz_sub_ss): New intrinsics.
29868 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29869 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
29870 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
29871 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
29872 (__builtin_ia32_subss_mask_round): New builtins.
29873 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29874 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29875 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
29877 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
29878 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
29880 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
29883 2017-05-02 Martin Jambor <mjambor@suse.cz>
29885 PR tree-optimization/78687
29886 * tree-sra.c (access): New field parent.
29887 (process_subtree_disqualification): New function.
29888 (disqualify_candidate): Call it.
29889 (build_accesses_from_assign): Reset write flag if creating an
29891 (build_access_subtree): Fill in parent field and also prpagate
29892 down grp_write flag.
29893 (create_artificial_child_access): New parameter set_grp_write, set
29894 grp_write to its value.
29895 (propagate_subaccesses_across_link): Also propagate grp_write flag
29897 (propagate_all_subaccesses): Push the closest parent back to work
29898 queue if add_access_to_work_queue returned true.
29900 2017-05-02 Richard Biener <rguenther@suse.de>
29902 * common.opt (fstrict-overflow): Alias negative to fwrapv.
29903 * doc/invoke.texi (fstrict-overflow): Remove all traces of
29904 -fstrict-overflow documentation.
29905 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
29906 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
29907 flag_strict_overflow.
29908 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
29909 * lto-opts.c (lto_write_options): Do not stream it.
29910 * lto-wrapper.c (merge_and_complain): Do not handle it.
29911 * opts.c (default_options_table): Do not set -fstrict-overflow.
29912 (finish_options): Likewise do not clear it when sanitizing.
29913 * simplify-rtx.c (simplify_const_relational_operation): Do not
29914 test flag_strict_overflow.
29916 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29918 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
29919 using enabled attribute.
29920 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
29921 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
29922 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
29923 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
29924 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
29925 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
29926 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
29927 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
29928 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
29929 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
29931 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29933 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
29935 2017-05-02 Richard Biener <rguenther@suse.de>
29937 PR tree-optimization/80591
29939 2017-04-10 Richard Biener <rguenther@suse.de>
29941 * tree-ssa-structalias.c (find_func_aliases): Properly handle
29944 2017-05-02 Richard Biener <rguenther@suse.de>
29946 PR tree-optimization/80549
29947 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
29948 (cleanup_tree_cfg_noloop): Create forwarders to known loop
29949 headers if they do not have a preheader.
29951 2017-05-02 Martin Liska <mliska@suse.cz>
29954 * common.opt: Fix typo.
29955 * doc/invoke.texi: Likewise.
29957 2017-05-01 Jan Beulich <jbeulich@suse.com>
29959 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
29960 swapping, add (x,x,m,x,n) alternative.
29962 2017-05-01 Nathan Sidwell <nathan@acm.org>
29964 * calls.c (combine_pending_stack_adjustment_and_call): Remove
29965 unnecessary unadjusted_alignment check.
29967 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29970 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
29972 * gimplify.c (gimplify_cilk_detach): New function.
29973 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
29974 * tree-core.h: Document EXPR_CILK_SPAWN.
29975 * tree.h (EXPR_CILK_SPAWN): Define.
29977 2017-05-01 David Malcolm <dmalcolm@redhat.com>
29979 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
29980 to use new fixit_hint representation, using the "replace" logic.
29981 (get_line_span_for_fixit_hint): Likewise.
29982 (layout::print_any_fixits): Likewise.
29983 (selftest::test_one_liner_many_fixits): Rename to...
29984 (selftest::test_one_liner_many_fixits_1): ...this, and update
29985 comment and expected output to reflect that the multiple fix-it
29986 hints are now consolidated into one insertion.
29987 (selftest::test_one_liner_many_fixits_2): New test.
29988 (selftest::test_diagnostic_show_locus_one_liner): Update for
29990 (selftest::test_fixit_consolidation): Update for fix-it API
29992 * diagnostic.c (print_parseable_fixits): Likewise.
29993 * edit-context.c (edited_line::m_line_events): Convert from
29994 auto_vec <line_event *> to auto_vec <line_event>.
29995 (class line_event): Convert from abstract base class to a concrete
29996 class, taking over the role of replace_event.
29997 (class insert_event): Delete.
29998 (class replace_event): Rename to class line_event. Convert to
30000 (edit_context::add_fixits): Reimplement.
30001 (edit_context::apply_insert): Delete.
30002 (edit_context::apply_replace): Rename to...
30003 (edit_context::apply_fixit): ...this. Convert to half-open range.
30004 (edited_file::apply_insert): Delete.
30005 (edited_file::apply_replace): Rename to...
30006 (edited_file::apply_fixit): ...this.
30007 (edited_line::~edited_line): Drop deletion of events.
30008 (edited_line::apply_insert): Delete.
30009 (edited_line::apply_replace): Rename to...
30010 (edited_line::apply_fixit): ...this. Convert to half-open range.
30011 Update for change to type of m_line_events.
30012 * edit-context.h (edit_context::apply_insert): Delete.
30013 (edit_context::apply_replace): Rename to...
30014 (edit_context::apply_fixit): ...this.
30016 2017-05-01 Martin Sebor <msebor@redhat.com>
30018 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
30021 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
30024 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
30025 __get_cpuid_max returns 0.
30026 (__get_cpuid_count): Ditto.
30028 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
30030 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
30031 replacement expression is another instance of one of its arguments.
30033 2017-05-01 Jakub Jelinek <jakub@redhat.com>
30036 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
30037 check for stack push/pop autoinc.
30038 * config/i386/i386.c (ix86_agi_dependent): Return false
30039 if the only reason why modified_in_p returned true is that
30040 addr is SP based and set_insn is a push or pop.
30042 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
30044 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
30047 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
30050 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
30051 (account_size_time): Use two predicates - exec_pred and
30053 (evaluate_conditions_for_known_args): Compute both clause and
30055 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
30056 (inline_summary_t::duplicate): Update.
30057 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
30059 (compute_inline_parameters): Likewise.
30060 (estimate_edge_size_and_time): Update caluclation of time.
30061 (estimate_node_size_and_time): Compute both time and nonspecialized
30063 (estimate_ipcp_clone_size_and_time): Update.
30064 (inline_merge_summary): Update.
30065 (do_estimate_edge_time): Update.
30066 (do_estimate_edge_size): Update.
30067 (do_estimate_edge_hints): Update.
30068 (inline_read_section, inline_write_summary): Stream both new predicates.
30069 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
30071 (compute_inlined_call_time): Cleanup.
30072 (big_speedup_p): Update.
30073 (edge_badness): Update.
30074 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
30075 (size_time_entry): Replace predicate by exec_predicate and
30076 nonconst_predicate.
30077 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
30078 (estimate_edge_time): Return also nonspec_time.
30079 (reset_edge_growth_cache): Update.
30081 2017-04-29 Jakub Jelinek <jakub@redhat.com>
30083 PR rtl-optimization/80491
30084 * ifcvt.c (noce_process_if_block): When looking for x setter
30085 with missing else_bb, don't check only the insn right before
30086 cond_earliest, but look for the last insn that x is modified in
30087 within the same bb.
30089 PR rtl-optimization/80491
30090 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
30092 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
30094 PR tree-optimization/80487
30095 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
30097 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30099 PR tree-optimization/79697
30100 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
30101 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
30102 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
30104 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
30105 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
30107 2017-04-28 Martin Sebor <msebor@redhat.com>
30109 PR tree-optimization/80523
30110 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
30111 (init_target_to_host_charmap, target_to_host, target_strtol10): New
30113 (maybe_warn, format_directive, parse_directive): Use new functions.
30114 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
30116 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
30118 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
30120 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
30122 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
30123 target_header_dir): Set correctly.
30124 * configure: Regenerated.
30125 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30126 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30127 instead of SYSTEM_HEADER_DIR.
30129 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
30131 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
30132 (estimate_local_effects): Likewise.
30133 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
30134 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
30135 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
30136 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
30137 do_estimate_edge_time, estimate_edge_time): Likewise.
30138 * ipa-inline-analysis.c (estimate_node_size_and_time,
30139 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
30140 (estimate_time_after_inlining): Remove.
30142 2017-04-28 Martin Liska <mliska@suse.cz>
30144 * doc/gcov.texi: Enhance documentation of gcov.
30146 2017-04-28 Martin Liska <mliska@suse.cz>
30148 * doc/gcov.texi: Sort options in alphabetic order.
30149 * doc/gcov-dump.texi: Likewise.
30150 * doc/gcov-tool.texi: Likewise.
30151 * gcov.c (print_usage): Likewise.
30152 * gcov-dump.c (print_usage): Likewise.
30153 * gcov-tool.c (print_merge_usage_message): Likewise.
30154 (print_rewrite_usage_message): Likewise.
30155 (print_overlap_usage_message): Likewise.
30157 2017-04-28 Martin Liska <mliska@suse.cz>
30159 PR gcov-profile/53915
30160 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
30162 2017-04-28 Martin Liska <mliska@suse.cz>
30164 PR gcov-profile/79891
30165 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
30166 is marked by compiler as living on a line.
30167 (get_cycles_count): Remove usage of the union.
30168 (output_intermediate_file): Likewise.
30169 (find_source): Fix GNU coding style.
30170 (accumulate_line_counts): Remove old non-all block mode.
30171 (output_lines): Remove usage of the union.
30172 * profile.c (output_location): Include all BBs, even if
30173 belonging to a same line (and file) as a previous BB.
30175 2017-04-28 Martin Liska <mliska@suse.cz>
30177 * gcov.c (process_args): Handle new argument 'w'.
30178 (read_graph_file): Assign ID to BBs.
30179 (output_branch_count): Display BB # if verbose flag is set.
30180 (output_lines): Likewise for arcs.
30181 (print_usage): Add '--verbose' option help.
30182 * doc/gcov.texi: Document --verbose (-w) option.
30184 2017-04-28 Martin Liska <mliska@suse.cz>
30186 * gcov.c (struct block_location_info): New struct.
30187 (process_file): Fill up the new structure.
30188 (read_graph_file): Replace usage of encoding by the newly added
30190 (add_line_counts): Likewise.
30191 (accumulate_line_counts): Remove usage of the union.
30192 (function_info::function_info): New function.
30193 (function_info::~function_info): Likewise.
30194 (process_file): Call delete instead of release_function.
30195 (release_function): Release the function.
30196 (release_structures): Call delete instead of release_function.
30197 (solve_flow_graph): Replace usage of num_blocks.
30198 (find_exception_blocks): Likewise.
30199 (output_lines): Fix GNU coding style.
30201 2017-04-28 Martin Liska <mliska@suse.cz>
30204 * coverage.c (coverage_remove_note_file): New function.
30205 * coverage.h: Declare the function.
30206 * toplev.c (finalize): Clean if an error has been seen.
30208 2017-04-28 Martin Liska <mliska@suse.cz>
30210 PR gcov-profile/80031
30211 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
30212 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
30213 * gcov.c (read_graph_file): Read just number of blocks.
30214 * profile.c (branch_prob): Do not stream 0 flags per a basic
30217 2017-04-28 Martin Liska <mliska@suse.cz>
30219 * gcov-dump.c (tag_*): Add new argument to declarations.
30220 (dump_gcov_file): Likewise.
30221 (tag_blocks): Add and use new argument depth.
30222 (tag_arcs): Likewise.
30223 (tag_lines): Likewise.
30224 (tag_counters): Likewise.
30225 (tag_summary): Likewise.
30226 (dump_working_sets): Use depth to do a proper indentation.
30228 2017-04-28 Jakub Jelinek <jakub@redhat.com>
30231 * cgraph.h (symtab_node::debug_symtab): No longer inline.
30232 * symtab.c (symtab_node::debug_symtab): Move definition here.
30234 2017-04-28 Richard Biener <rguenther@suse.de>
30236 * lto-streamer.h (LTO_major_version): Bump to 7.
30238 2017-04-28 Richard Biener <rguenther@suse.de>
30240 * tree-vrp.c (assert_info): New struct.
30241 (add_assert_info): New helper.
30242 (register_edge_assert_for_2): Refactor to add asserts to a vector
30244 (register_edge_assert_for_1): Likewise.
30245 (register_edge_assert_for): Likewise.
30246 (finish_register_edge_assert_for): New helper actually registering
30247 asserts where live on edge.
30248 (find_conditional_asserts): Adjust.
30249 (find_switch_asserts): Likewise.
30250 (evrp_dom_walker::try_find_new_range): Generalize.
30251 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
30253 2017-04-27 Marek Polacek <polacek@redhat.com>
30256 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
30257 arg10 and arg11 to itype.
30259 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
30261 * doc/extend.texi (Object Size Checking): Improve grammar.
30263 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
30266 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
30267 that the logic for permitting reciprocal estimates matches that
30270 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30273 * tree.c (type_cache_hasher::equal): Only compare
30274 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
30275 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
30276 non-aggregate element types.
30277 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
30278 about the flag on ARRAY_TYPEs in the comment, formatting fix.
30280 2017-04-27 Richard Biener <rguenther@suse.de>
30282 PR middle-end/80533
30283 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
30284 stripping ARRAY_REFs from MEM_EXPR make sure we're not
30285 keeping a reference to a trailing array.
30287 2017-04-27 Richard Biener <rguenther@suse.de>
30289 PR middle-end/80539
30290 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
30291 being in loop-closed SSA form conservatively.
30292 (chrec_fold_multiply_poly_poly): Likewise.
30294 2017-04-27 Tamar Christina <tamar.christina@arm.com>
30296 PR middle-end/79665
30297 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
30298 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
30300 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30303 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
30304 (aarch64_function_arg_alignment): Return unsigned int again, but still
30305 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
30306 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30307 Don't emit -Wpsabi note.
30308 (aarch64_function_arg_boundary): Likewise.
30309 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30312 2017-04-26 Nathan Sidwell <nathan@acm.org>
30314 * tree.h (crc32_unsigned_n): Declare.
30315 (crc32_unsigned, crc32_unsigned): Make inline.
30316 * tree.c (crc32_unsigned_bits): Replace with ...
30317 (crc32_unsigned_n): ... this.
30318 (crc32_unsigned, crc32_byte): Remove.
30319 (crc32_string): Remove unnecessary braces.
30321 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30323 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
30324 * ipa-inline-analysis.c (MAX_TIME): Remove.
30325 (account_size_time): Use sreal for time.
30326 (dump_inline_summary): Update.
30327 (estimate_function_body_sizes): Update.
30328 (estimate_edge_size_and_time): Update.
30329 (estimate_calls_size_and_time): Update.
30330 (estimate_node_size_and_time): Update.
30331 (inline_merge_summary): Update.
30332 (inline_update_overall_summary): Update.
30333 (estimate_time_after_inlining): Update.
30334 (inline_read_section): Update.
30335 (inline_write_summary): Update.
30336 * ipa-inline.c (compute_uninlined_call_time): Update.
30337 (compute_inlined_call_time): Update.
30338 (recursive_inlining): Update.
30339 (inline_small_functions): Update.
30340 (dump_overall_stats): Update.
30341 * ipa-inline.h: Include sreal.h.
30342 (size_time_entry): Turn time to sreal.
30343 (inline_summary): Turn self_time nad time to sreal.
30345 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30347 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
30349 (sreal::stream_out, sreal::stream_in): New.
30350 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
30352 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30354 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
30357 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
30360 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
30362 (dimode_scalar_chain::compute_convert_gain): Ditto.
30363 (dimode_scalar_chain::make_vector_copies): Ditto.
30364 (dimode_scalar_chain::convert_reg): Ditto.
30365 (dimode_scalar_chain::convert_insn): Ditto.
30366 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
30367 (VI248_AVX512BW_1): New mode iterator.
30368 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
30369 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
30372 2017-04-25 Martin Sebor <msebor@redhat.com>
30374 PR tree-optimization/80497
30375 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
30376 constants are representable in HOST_WIDE_INT.
30377 (parse_directive): Ditto.
30379 2017-04-25 Martin Sebor <msebor@redhat.com>
30382 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
30383 (new_zero_array): Adjust signature.
30384 (dom_info::dom_init): Used unsigned rather that size_t.
30385 (dom_info::dom_info): Same.
30387 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30388 Jakub Jelinek <jakub@redhat.com>
30391 * config/arm/arm.c: Include gimple.h.
30392 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30393 returns negative, increment ncrn only if it returned positive.
30394 (arm_needs_doubleword_align): Return int instead of bool,
30395 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
30396 members, but if there is any such non-FIELD_DECL
30397 > PARM_BOUNDARY aligned decl, return -1 instead of false.
30398 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30399 returns negative, increment nregs only if it returned positive.
30400 (arm_setup_incoming_varargs): Likewise.
30401 (arm_function_arg_boundary): Emit -Wpsabi note if
30402 arm_needs_doubleword_align returns negative, return
30403 DOUBLEWORD_ALIGNMENT only if it returned positive.
30405 2017-04-25 Marek Polacek <polacek@redhat.com>
30408 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30409 first argument to type.
30411 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
30414 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
30415 type checks to test for compatibility instead of equality.
30417 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30418 Jakub Jelinek <jakub@redhat.com>
30421 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
30423 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
30424 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
30425 the alignment computation, but return their maximum in warn_alignment.
30426 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30427 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
30429 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
30430 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30433 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30435 * config/arc/simdext.md (dmpyh): Fix typo.
30437 2017-04-25 Richard Biener <rguenther@suse.de>
30439 PR tree-optimization/80492
30440 * alias.c (compare_base_decls): Handle registers with asm
30441 specification conservatively.
30442 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
30443 compare_base_decls returning dont-know properly.
30445 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30447 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
30448 (legitimate_offset_address_p): New function.
30449 (arc_legitimate_address_p): Use above function.
30451 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30453 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
30455 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30457 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
30458 ACCH registers whenever they are available.
30460 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30462 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
30463 double regs fix when not used.
30465 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30467 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
30469 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
30470 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
30472 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30474 * config/arc/arc.c (arc_output_addsi): Check for h-register class
30475 when emitting short ADD instructions.
30477 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30479 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
30481 (cmpsi_cc_c_insn): Likewise.
30482 (cbranchsi4_scratch): Compute proper instruction length using
30483 compact_hreg_operand.
30484 * config/arc/predicates.md (compact_hreg_operand): New predicate.
30486 2017-04-25 Richard Biener <rguenther@suse.de>
30488 PR middle-end/80509
30489 * passes.c (pass_manager::pass_manager): Initialize
30490 m_name_to_pass_map.
30492 2017-04-25 Richard Biener <rguenther@suse.de>
30494 PR tree-optimization/79201
30495 * tree-ssa-sink.c (statement_sink_location): Handle calls.
30497 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30500 * config/s390/vector.md: Split MEM->GPR vector moves for
30501 non-s_operand addresses.
30503 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30506 * config/s390/predicates.md (reload_const_wide_int_operand): New
30508 * config/s390/s390.md ("movti"): Remove d/P alternative.
30509 ("movti_bigconst"): New pattern definition.
30511 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30514 * s390-protos.h (s390_expand_cs_hqi): Removed.
30515 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
30516 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
30517 modes as well as CCZ1mode and CCZmode.
30518 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
30519 signature of s390_emit_compare_and_swap.
30520 (s390_expand_cs_hqi): Likewise, make static.
30521 (s390_expand_cs_tdsi): Generate an explicit compare before trying
30522 compare-and-swap, in some cases.
30523 (s390_expand_cs): Wrapper function.
30524 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
30526 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
30527 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
30528 patterns for small and large integers. Forbid symref memory operands.
30529 Move expander to s390.c. Require cc register.
30530 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
30531 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
30532 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
30533 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
30534 symref memory operands. Remove CC mode and call s390_match_ccmode
30536 ("atomic_exchange<mode>"): Allow and implement all integer modes.
30538 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30540 * config/s390/s390.md (define_peephole2): New peephole to help
30541 combining the load-and-test pattern with volatile memory.
30543 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30545 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
30546 with CCZmode for TARGET_Z196.
30548 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30550 PR rtl-optimization/80501
30551 * combine.c (make_compound_operation_int): Set subreg_code to SET
30552 even for AND with mask of the sign bit of mode.
30554 PR rtl-optimization/80500
30555 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
30556 sum's initial value.
30558 2017-04-25 Julian Brown <julian@codesourcery.com>
30559 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30561 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
30563 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
30565 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
30567 2017-04-25 Julian Brown <julian@codesourcery.com>
30568 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30570 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
30571 (thunderx2t99_sha): New Reservation.
30573 2017-04-25 Julian Brown <julian@codesourcery.com>
30574 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30576 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
30577 type for 1-element load.
30579 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
30581 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
30583 2017-04-24 Martin Jambor <mjambor@suse.cz>
30585 PR tree-optimization/80293
30586 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
30587 char arrays not totally scalarizable if it is false.
30588 (analyze_all_variable_accesses): Pass correct value in the new
30589 parameter. Add a statistics counter.
30591 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
30593 PR middle-end/79931
30594 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
30596 2017-04-24 Richard Biener <rguenther@suse.de>
30598 PR tree-optimization/80494
30599 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
30600 out for complex types.
30602 2017-04-24 Richard Biener <rguenther@suse.de>
30604 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
30605 * tree-ssa-sccvn.c (print_scc): Print SCC size.
30606 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
30607 (DFS): Adjust and never fail.
30608 (sccvn_dom_walker::fail): Remove.
30609 (sccvn_dom_walker::before_dom_children): Adjust.
30610 (run_scc_vn): Likewise and never fail.
30611 * tree-ssa-pre.c (pass_pre::execute): Adjust.
30612 (pass_fre::execute): Likewise.
30614 2017-04-24 Richard Biener <rguenther@suse.de>
30616 PR tree-optimization/79725
30617 * tree-ssa-sink.c (statement_sink_location): Return whether
30618 failure reason was zero uses. Move that check later.
30619 (sink_code_in_bb): Deal with zero uses by removing the stmt
30622 2017-04-24 Richard Biener <rguenther@suse.de>
30625 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
30626 pointer-based references.
30628 2017-04-24 Richard Biener <rguenther@suse.de>
30631 * pass_manager.h (pass_manager::operator new): Remove.
30632 (pass_manager::operator delete): Likewise.
30633 * passes.c (pass_manager::operator new): Remove.
30634 (pass_manager::operator delete): Likewise.
30635 (pass_manager::pass_manager): Zero individual pass members.
30637 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
30640 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
30641 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
30642 Check "XEXP (src, 1)" operand here.
30643 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
30644 Check "XEXP (src, 1)" operand here.
30645 (dimode_scalar_chain::make_vector_copies): Detect count register
30646 of a shift instruction. Zero extend count register from QImode
30647 to DImode to satisfy vector shift pattern count operand predicate.
30648 Substitute vector shift count operand with a DImode copy.
30649 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
30652 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
30654 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
30655 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
30656 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
30657 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
30658 (UNSPEC_NOREX_MEM): Remove definition.
30660 2017-04-21 Richard Biener <rguenther@suse.de>
30662 PR tree-optimization/79547
30663 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
30664 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
30665 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
30666 without any constraints.
30668 2017-04-21 Richard Biener <rguenther@suse.de>
30670 PR tree-optimization/78847
30671 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
30673 2017-04-21 Richard Biener <rguenther@suse.de>
30675 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
30676 (build_distinct_type_copy): Likewise.
30677 (build_variant_type_copy): Likewise.
30678 * tree.c (build_qualified_type): Pass down mem-stat info.
30679 (build_distinct_type_copy): Likewise.
30680 (build_variant_type_copy): Likewise.
30682 2017-04-21 Richard Biener <rguenther@suse.de>
30684 PR tree-optimization/80237
30685 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
30687 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
30688 for a simplified result.
30690 2016-04-21 Richard Biener <rguenther@suse.de>
30692 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
30693 sth as strict as a simple_iv but a chrec without symbols and an
30694 operand defined in the loop we are peeling (and not some subloop).
30695 (propagate_constants_for_unrolling): Propagate all constants.
30697 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
30700 * config/i386/i386.c (print_reg): Remove assert for disalowed
30701 regno values, call output_operand_lossage instead.
30703 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
30706 * config/i386/constraints.md (Yc): New register constraint.
30707 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
30708 Use Yc constraint for alternative 2 of operand 0. Remove
30709 preferred_for_speed attribute.
30711 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
30713 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
30714 lastprivate clauses in SIMT case.
30716 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
30718 * doc/invoke.texi (-Wextra-semi): Document new warning option.
30720 2017-04-20 Richard Biener <rguenther@suse.de>
30722 PR tree-optimization/57796
30723 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
30724 as N scalar stores.
30725 (vect_model_load_cost): Cost gathers as N scalar loads.
30727 2017-04-20 Richard Biener <rguenther@suse.de>
30729 * ggc-page.c (ggc_allocated_p): Rename to ...
30730 (safe_lookup_page_table_entry): ... this and return the lookup
30732 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
30734 2017-04-20 Richard Biener <rguenther@suse.de>
30736 PR tree-optimization/80453
30737 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
30738 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
30739 from the conditions.
30740 (vn_phi_eq): Pass them down.
30741 (vn_phi_lookup): Record them.
30742 (vn_phi_insert): Likewise.
30744 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
30746 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
30747 uninitialized variable warning to avoid buffer overrun.
30749 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
30752 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
30753 is suppressed for '{ 0 }' in C.
30755 2017-04-20 Jakub Jelinek <jakub@redhat.com>
30757 * BASE-VER: Set to 8.0.0.
30759 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
30761 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
30762 priority .init_array and .fini_array section with SECTION_NOTYPE
30765 2017-04-20 Jakub Jelinek <jakub@redhat.com>
30767 PR middle-end/80423
30768 * tree.h (build_array_type): Add typeless_storage default argument.
30769 * tree.c (type_cache_hasher::equal): Also compare
30770 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
30771 (build_array_type): Add typeless_storage argument, set
30772 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
30774 (build_nonshared_array_type): Adjust build_array_type_1 caller.
30775 (build_array_type): Likewise. Add typeless_storage argument.
30777 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
30778 Jakub Jelinek <jakub@redhat.com>
30780 PR tree-optimization/80426
30781 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
30782 operation on symbolic operands, also compute the overflow for the
30783 invariant part when the operation degenerates into a negation.
30785 2017-04-19 Jakub Jelinek <jakub@redhat.com>
30788 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
30789 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
30792 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
30794 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
30797 * config/avr/avr.c (tree.h): Include it.
30798 (cgraph.h): Include it.
30799 (avr_encode_section_info): Don't warn for uninitialized progmem
30800 variable if it's just an alias.
30802 2017-04-19 Richard Biener <rguenther@suse.de>
30805 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
30806 when needed by AutoPGO.
30808 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
30811 * doc/lto.texi: Remove an extra 'that'.
30813 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
30815 PR rtl-optimization/80429
30816 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
30817 are only used in debug insns.
30819 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
30820 Vladimir Makarov <vmakarov@redhat.com>
30822 * config/sparc/predicates.md (input_operand): Add comment. Return
30823 true for any memory operand when LRA is in progress.
30824 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
30826 2017-04-18 Jeff Law <law@redhat.com>
30829 * mips.md ({return,simple_return}_internal): Do not overwrite
30832 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30834 PR tree-optimization/80443
30835 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
30836 instead of adding 1, subtract -1 and similarly instead of subtracting
30839 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
30841 PR rtl-optimization/80357
30842 * haifa-sched.c (tmp_bitmap): New variable.
30843 (model_recompute): Handle duplicate use records.
30844 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
30845 (free_global_sched_pressure_data): Free it.
30847 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
30850 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
30851 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30852 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30853 instead of SYSTEM_HEADER_DIR.
30855 2017-04-18 Jeff Law <law@redhat.com>
30857 PR middle-end/80422
30858 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
30859 predecessors after walking up the insn chain.
30861 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30864 * dwarf2out.c (modified_type_die): Try harder not to emit internal
30865 sizetype type into debug info.
30867 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30870 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
30871 unneeded test for TARGET_UPPER_REGS_SF.
30872 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
30874 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30877 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
30878 instead of gsi_after_labels.
30880 2017-04-18 Jeff Law <law@redhat.com>
30882 * regcprop.c (maybe_mode_change): Avoid creating copies of the
30886 2017-04-13 Jeff Law <law@redhat.com>
30887 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30888 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30890 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
30893 * config/avr/avr.c (intl.h): Include it.
30894 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
30896 2017-04-18 Martin Liska <mliska@suse.cz>
30898 PR gcov-profile/78783
30899 * gcov-tool.c (gcov_output_files): Validate that destination
30900 file is either removed by the tool or by a user.
30902 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30903 Guy Benyei <guybe@mellanox.com>
30905 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
30906 block, and do not negate it, the stored id is already negative.
30908 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30910 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
30912 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
30915 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
30916 masks of options that should be turned off if the VSX vector
30917 options are turned off.
30918 (OTHER_P8_VECTOR_MASKS): Likewise.
30919 (OTHER_VSX_VECTOR_MASKS): Likewise.
30920 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
30921 rs6000_disable_incompatible_switches to validate no type switches
30923 (rs6000_incompatible_switch): New function to disallow turning on
30924 other vector options if -mno-vsx, -mno-power8-vector, or
30925 -mno-power9-vector are specified.
30927 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30929 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
30931 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30933 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
30934 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
30935 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
30936 (ARG_POINTER_CFA_OFFSET): Likewise.
30938 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30940 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
30941 conditions to take advantage of various optimizations.
30943 2017-04-13 Jeff Law <law@redhat.com>
30945 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30946 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30947 (zero_extendsidi2_dext): Likewise.
30949 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30952 * fold-const.c (fold_ternary_loc): Revert
30953 use op0 instead of fold_convert_loc (loc, type, arg0) part of
30956 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
30958 PR rtl-optimization/80343
30959 * lra-remat.c (update_scratch_ops): Assign original hard reg to
30960 new scratch pseudo.
30962 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
30965 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
30966 to ubsan_encode_value.
30968 2017-04-13 Jeff Law <law@redhat.com>
30970 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
30971 appearing in DEBUG_INSNs.
30973 2017-04-13 Martin Liska <mliska@suse.cz>
30975 PR gcov-profile/80413
30976 * gcov-io.c (gcov_write_string): Copy to buffer just when
30977 allocated size is greater than zero.
30979 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30982 * dwarf2out.c (decls_for_scope): Ignore declarations of
30983 current_function_decl in BLOCK_NONLOCALIZED_VARS.
30985 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
30988 * ipa-visibility.c (non_local_p): Fix typos.
30989 (localize_node): When localizing symbol in same comdat group,
30990 dissolve the group only when we know external symbols are going
30992 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
30994 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30996 PR tree-optimization/79390
30997 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
30998 order does not result in usable sequence, retry with reversed operand
31004 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
31005 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
31006 op0 instead of fold_convert_loc (loc, type, arg0).
31008 2017-04-12 Jeff Law <law@redhat.com>
31010 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
31011 has a delay slot in the generated code.
31013 * config/cris/cris.md (cris_preferred_reload_class): Return
31014 GENNONACR_REGS rather than GENERAL_REGS.
31016 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31019 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
31020 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
31021 signedness of the result type.
31023 2017-04-12 Richard Biener <rguenther@suse.de>
31024 Jeff Law <law@redhat.com>
31026 PR tree-optimization/80359
31027 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
31028 trim stores to TARGET_MEM_REFs.
31030 2017-04-12 Richard Biener <rguenther@suse.de>
31032 PR tree-optimization/79390
31033 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
31034 threading case even more.
31036 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
31039 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
31040 for quad_address_p for TImode, instead of just not indexed_address.
31042 2017-04-12 Richard Biener <rguenther@suse.de>
31043 Bernd Edlinger <bernd.edlinger@hotmail.de>
31045 PR middle-end/79671
31046 * alias.c (component_uses_parent_alias_set_from): Handle
31047 TYPE_TYPELESS_STORAGE.
31048 (get_alias_set): Likewise.
31049 * tree-core.h (tree_type_common): Add typeless_storage flag.
31050 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
31051 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
31052 for types containing members with TYPE_TYPELESS_STORAGE.
31053 (place_field): Likewise.
31054 (layout_type): Likewise for ARRAY_TYPE.
31055 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
31056 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
31057 TYPE_TYPELESS_STORAGE.
31058 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
31060 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31063 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
31064 first argument to type.
31066 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31070 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
31071 CONST0_RTX (mode) rather than const0_rtx where appropriate.
31072 (rs6000_expand_binop_builtin): Likewise.
31073 (rs6000_expand_ternop_builtin): Likewise; also add missing
31074 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
31075 vshasigma built-ins.
31076 * doc/extend.texi: Document that vec_xxpermdi's third argument
31077 must be a constant.
31079 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
31081 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
31082 Use shift_const cost parameter when calculating gain of STV shifts.
31084 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
31086 PR rtl-optimization/70478
31087 * lra-constraints.c (process_alt_operands): Check memory for
31088 disfavoring memory insn operand.
31090 2017-04-11 Jakub Jelinek <jakub@redhat.com>
31092 PR middle-end/80100
31093 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
31094 left shift in unsigned HOST_WIDE_INT type.
31096 PR rtl-optimization/80385
31097 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
31098 (not (neg X)) into (plus X -1) for complex or non-integral modes.
31101 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
31102 if they have any depend clauses.
31104 2017-04-11 Martin Liska <mliska@suse.cz>
31107 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
31108 * ipa-split.c (split_function): Create a local comdat symbol
31109 if caller is in a comdat group.
31111 2017-04-11 Martin Liska <mliska@suse.cz>
31114 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
31117 2017-04-11 Martin Sebor <msebor@redhat.com>
31119 PR middle-end/80364
31120 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
31121 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
31123 (directive::set_width, directive::set_precision, format_character):
31125 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
31128 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
31131 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
31132 conflict, set target->arch_name instead of target->cpu_name.
31134 2017-04-11 Richard Biener <rguenther@suse.de>
31136 PR tree-optimization/80374
31137 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
31138 build_zero_cst, remove fold_convertible_p check again.
31140 2017-04-11 Martin Liska <mliska@suse.cz>
31143 * ubsan.c (instrument_object_size): Do not instrument register
31146 2017-04-11 Jakub Jelinek <jakub@redhat.com>
31149 * config/i386/i386-builtin-types.def
31150 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
31151 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
31152 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
31153 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
31154 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
31155 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
31156 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
31157 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
31158 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
31159 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
31160 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
31161 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
31162 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
31163 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
31164 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
31165 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
31166 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
31167 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
31168 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
31169 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
31170 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
31171 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
31172 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
31173 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
31174 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
31175 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
31176 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
31177 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
31178 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
31179 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
31180 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
31181 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
31182 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
31183 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
31184 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
31185 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
31186 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
31187 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
31188 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
31189 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
31190 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
31191 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
31192 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
31193 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
31194 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
31195 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
31197 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
31198 flag to second_arg_count, handle 4 argument function type _COUNT
31199 aliases, handle second_arg_count on second argument rather than last.
31201 2017-04-10 Jeff Law <law@redhat.com>
31203 PR tree-optimization/80374
31204 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
31205 record anything if we can not convert integer_zero_node to the
31208 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
31211 * config/rs6000/rs6000.c (rs6000_option_override_internal):
31212 Enhance special handling given to the TARGET_P9_MINMAX option in
31213 relation to certain other options.
31215 2017-04-10 Bin Cheng <bin.cheng@arm.com>
31217 PR tree-optimization/80153
31218 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
31219 remove POINTER_PLUS_EXPR's base part directly, rather than through
31222 2017-04-10 Richard Biener <rguenther@suse.de>
31223 Bin Cheng <bin.cheng@arm.com>
31225 PR tree-optimization/80153
31226 * tree-affine.c (aff_combination_to_tree): Get base pointer from
31227 the first element of pointer type aff_tree. Build result expr in
31229 (add_elt_to_tree): Convert to type unconditionally. Remove other
31230 fold_convert calls.
31231 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
31232 (rewrite_use_nonlinear_expr): Check invariant using iv information.
31234 2017-04-10 Richard Biener <rguenther@suse.de>
31236 * tree-ssa-structalias.c (find_func_aliases): Properly handle
31239 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
31241 PR rtl-optimization/70478
31242 * lra-constraints.c (curr_small_class_check): New.
31243 (update_and_check_small_class_inputs): New.
31244 (process_alt_operands): Update curr_small_class_check. Disfavor
31245 alternative insn memory operands. Check available regs for small
31248 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31251 * config/mips/mips.opt (-mvirt): Update description.
31252 * doc/invoke.texi (-mvirt): Likewise.
31254 2017-04-10 Richard Biener <rguenther@suse.de>
31256 PR middle-end/80362
31257 * fold-const.c (fold_binary_loc): Look at unstripped ops when
31258 looking for NEGATE_EXPR in -A / -B to A / B folding.
31260 2017-04-10 Martin Liska <mliska@suse.cz>
31262 PR gcov-profile/80224
31263 * gcov.c (print_usage): Fix usage string.
31264 (get_gcov_intermediate_filename): Remove.
31265 (output_gcov_file): Use both for normal and intermediate format.
31266 (generate_results): Do not initialize special file for
31267 intermediate format.
31269 2017-04-10 Richard Biener <rguenther@suse.de>
31271 PR tree-optimization/80304
31272 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
31275 2017-04-10 Nathan Sidwell <nathan@acm.org>
31278 * config/rs6000/rs6000.c (rs6000_vector_type): New.
31279 (rs6000_init_builtins): Use it.
31281 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31283 * config/arm/arm.md (<mrc>): Add mode to SET source.
31284 (<mrrc>): Likewise.
31286 2017-04-10 Richard Biener <rguenther@suse.de>
31288 PR middle-end/80344
31289 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
31291 2017-04-10 Jakub Jelinek <jakub@redhat.com>
31294 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
31295 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
31296 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
31297 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
31298 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
31299 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
31300 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
31301 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
31302 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
31303 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
31304 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
31305 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
31306 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
31307 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
31308 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
31309 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
31310 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
31311 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
31312 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
31313 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
31314 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
31315 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
31316 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
31318 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
31320 PR rtl-optimization/70478
31321 * lra-constraints.c: Reverse the last patch.
31323 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
31325 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
31326 Add comment for WCHAR_T.
31328 2017-04-08 Martin Liska <mliska@suse.cz>
31331 2017-04-07 Martin Liska <mliska@suse.cz>
31334 * ipa-split.c (split_function): Add function part to a same comdat
31337 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31340 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
31342 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
31344 * rs6000/rs6000.c (vec_load_pendulum): Rename...
31345 (vec_pairing): ...to this.
31346 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
31347 (rs6000_sched_init): Adjust for name change.
31348 (struct rs6000_sched_context): Likewise.
31349 (rs6000_init_sched_context): Likewise.
31350 (rs6000_set_sched_context): Likewise.
31352 2017-04-07 Jakub Jelinek <jakub@redhat.com>
31358 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
31360 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
31361 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
31362 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
31364 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
31366 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
31368 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31370 PR rtl-optimization/70703
31371 * ira-color.c (update_conflict_hard_regno_costs): Use
31372 int64_t instead of HOST_WIDE_INT.
31374 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31376 PR rtl-optimization/70478
31377 * lra-constraints.c (process_alt_operands): Disfavor alternative
31378 insn memory operands.
31380 2017-04-07 Jeff Law <law@redhat.com>
31382 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
31383 CALL and NOTE_INSN_CALL_ARG_LOCATION.
31385 2017-04-07 Martin Liska <mliska@suse.cz>
31388 * config/aarch64/aarch64.c (aarch64_process_target_attr):
31389 Show error message instead of an ICE.
31391 2017-04-07 Martin Liska <mliska@suse.cz>
31394 * ipa-split.c (split_function): Add function part to a same comdat
31397 2017-04-07 Richard Biener <rguenther@suse.de>
31399 PR middle-end/80341
31400 * tree.c (get_unwidened): Also handle ! for_type case for
31402 * convert.c (do_narrow): Split out from ...
31403 (convert_to_integer_1): ... here. Do not pass final truncation
31404 type to get_unwidened for TRUNC_DIV_EXPR.
31406 2017-04-07 Richard Biener <rguenther@suse.de>
31408 * tree-affine.c (wide_int_ext_for_comb): Take type rather
31410 (aff_combination_const): Adjust.
31411 (aff_combination_scale): Likewise.
31412 (aff_combination_add_elt): Likewise.
31413 (aff_combination_add_cst): Likewise.
31414 (aff_combination_convert): Likewise.
31415 (add_elt_to_tree): Likewise. Remove unused argument.
31416 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
31418 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
31420 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
31422 * config/arm/arm.c (arm_default_short_enums): Use
31423 ARM_DEFAULT_SHORT_ENUMS.
31424 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
31426 2017-04-06 Jakub Jelinek <jakub@redhat.com>
31429 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
31430 members with redundant out-of-class redeclaration.
31432 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31435 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
31436 * config/i386/i386.md (*zero_extendsidi2):
31437 Add (?*x,*x) and (?*v,*v) alternatives.
31439 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31442 * config/i386/i386.c (ix86_expand_builtin)
31443 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
31444 mode from insn data. Convert operands to insn operand mode.
31445 Copy operands that don't satisfy insn predicate to a register.
31447 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
31449 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
31452 2017-04-06 Richard Biener <rguenther@suse.de>
31454 PR tree-optimization/80334
31455 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
31456 preserve alignment of accesses.
31458 2017-04-06 Richard Biener <rguenther@suse.de>
31460 PR tree-optimization/80262
31461 * tree-sra.c (build_ref_for_offset): Preserve address-space
31463 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
31464 Drop useless address-space information on MEM_REF offsets.
31466 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
31468 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
31470 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31472 PR rtl-optimization/70703
31473 * ira-color.c (update_conflict_hard_regno_costs): Use
31474 HOST_WIDE_INT instead of long.
31476 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
31479 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
31480 not defined for x86_64 target. Add -mmmx target option when __SSE2__
31482 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
31483 for x86_64 target. Handle -m3dnowa option.
31485 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31487 PR rtl-optimization/70703
31488 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
31489 (update_conflict_hard_regno_costs): Use long instead of unsigned
31490 arithmetic for cost calculation.
31492 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31493 Bernd Edlinger <bernd.edlinger@hotmail.de>
31496 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
31499 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
31502 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
31503 ptr_mode with Pmode throughout.
31504 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
31505 into probe_stack_range and use DImode.
31507 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
31510 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
31511 call_eh_return is true.
31513 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31515 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
31516 Initialize last_match_fntype_index.
31518 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31521 * tree-nvr.c: Include internal-fn.h.
31522 (pass_return_slot::execute): Ignore internal calls without
31525 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31526 Richard Biener <rguenther@suse.de>
31529 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
31530 captures used multiple times, except for the last use.
31531 * generic-match-head.c: Include gimplify.h.
31533 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31535 PR tree-optimization/79390
31536 * target.h (struct noce_if_info): Declare.
31537 * targhooks.h (default_noce_conversion_profitable_p): Declare.
31538 * target.def (noce_conversion_profitable_p): New target hook.
31539 * ifcvt.h (struct noce_if_info): New type, moved from ...
31540 * ifcvt.c (struct noce_if_info): ... here.
31541 (noce_conversion_profitable_p): Renamed to ...
31542 (default_noce_conversion_profitable_p): ... this. No longer
31544 (noce_try_store_flag_constants, noce_try_addcc,
31545 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
31546 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
31547 instead of noce_conversion_profitable_p.
31548 * config/i386/i386.c: Include ifcvt.h.
31549 (ix86_option_override_internal): Don't override
31550 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
31551 (ix86_noce_conversion_profitable_p): New function.
31552 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
31553 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
31554 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
31555 * doc/tm.texi: Regenerated.
31557 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31559 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
31562 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31565 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
31566 instructions for small multiply cores.
31568 2017-04-04 Jeff Law <law@redhat.com>
31570 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
31572 (mips_expand_vec_perm_const): Initialize elements in orig_perm
31573 that are not set by the loop over the elements.
31575 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31578 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
31579 int mode, convert_modes it to mode as unsigned, otherwise use
31580 lowpart_subreg to mode rather than SImode.
31581 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
31582 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
31583 Use DImode instead of SImode for the shift count operand.
31584 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
31587 2017-04-04 Richard Biener <rguenther@suse.de>
31589 PR middle-end/80281
31590 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
31591 arithmetic done for the negate or the plus. Simplify.
31592 (A - (-B) -> A + B): Likewise.
31593 * fold-const.c (split_tree): Make sure to not negate pointers.
31595 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
31597 PR rtl-optimization/60818
31598 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
31599 a compare of comparisons with the thing compared if this results
31600 in a different machine mode.
31602 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
31604 * alias.c (base_alias_check): Fix typo in comment.
31605 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
31606 * cgraphunit.c (symbol_table::compile): Likewise.
31607 * collect2.c (maybe_run_lto_and_relink): Likewise.
31608 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
31609 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
31610 * config/avr/avr.c (avr_map_op_t): Likewise.
31611 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
31612 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
31613 * config/epiphany/epiphany.md (movcc): Likewise.
31614 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
31615 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
31617 * config/mips/mips.c (mips_save_restore_reg): Likewise.
31618 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
31619 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
31620 * config/sh/sh.c (sh_rtx_costs): Likewise.
31621 * fold-const.c (fold_truth_andor): Likewise.
31622 * genautomata.c (collapse_flag): Likewise.
31623 * gengtype.h (struct type::u::s): Likewise.
31624 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
31625 * input.c (FORMAT_AMOUNT): Likewise.
31626 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
31627 (known_aggs_to_agg_replacement_list): Likewise.
31628 * ipa-inline-analysis.c: Likewise.
31629 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
31630 * ipa-polymorphic-call.c
31631 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
31632 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
31633 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
31635 * modulo-sched.c (apply_reg_moves): Likewise.
31636 * omp-expand.c (build_omp_regions_1): Likewise.
31637 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
31638 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
31639 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
31640 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
31641 * value-prof.c: Likewise.
31642 * var-tracking.c (val_reset): Likewise.
31644 2017-04-03 Richard Biener <rguenther@suse.de>
31646 PR tree-optimization/80275
31647 * fold-const.c (split_address_to_core_and_offset): Handle
31650 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
31652 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
31653 descriptors is at least equal to that of functions.
31655 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
31657 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
31659 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
31662 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
31663 (mov<IMOD4:mode>): New expander.
31664 (*mov<IMOD4:mode>_internal): New insn and split pattern.
31666 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
31668 PR rtl-optimization/79405
31669 * fwprop.c (propagations_left): New variable.
31670 (forward_propagate_into): Decrement it.
31671 (fwprop_init): Initialize it.
31672 (fw_prop): If the variable has reached zero, stop propagating.
31673 (fwprop_addr): Ditto.
31675 2017-03-31 Jakub Jelinek <jakub@redhat.com>
31678 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
31679 a FUNCTION_DECL, pass it as decl instead of origin to
31682 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
31684 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
31687 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
31690 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
31691 TARGET_VSX_SMALL_INTEGER.
31693 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31695 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
31696 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
31698 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31700 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
31701 extraction from odd-numbered MSA register.
31703 2017-03-31 Jakub Jelinek <jakub@redhat.com>
31705 PR middle-end/80173
31706 * expmed.c (store_bit_field_1): Don't attempt to create
31707 a word subreg out of hard registers wider than word if they
31708 have HARD_REGNO_NREGS of 1 for their mode.
31710 PR middle-end/80163
31711 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
31712 conversions to integer types wider than word and pointer.
31715 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
31716 (rtx_equal_for_cselib_p): Pass 0 to it.
31717 * cselib.c (cselib_hasher::equal): Likewise.
31718 (rtx_equal_for_cselib_1): Add depth argument. If depth
31719 is 128, don't look up VALUE locs and punt. Increment
31720 depth in recursive calls when walking VALUE locs.
31722 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
31724 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
31725 (make_gcov_file_name): Use the canonical path name for generating
31727 (read_line): Fix handling of files with ascii null bytes.
31729 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
31731 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
31732 to initialise a vector register instead
31733 of using a const_int.
31735 2017-03-30 Jakub Jelinek <jakub@redhat.com>
31737 PR translation/80189
31738 * gimplify.c (omp_default_clause): Use %qs instead of %s in
31739 diagnostic messages.
31741 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
31744 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
31745 (dfp_diex_<mode>): Update mode of operand 1.
31746 * doc/extend.texi (dxex, dxexq): Document change to return type.
31747 (diex, diexq): Document change to argument type.
31749 2017-03-30 Martin Jambor <mjambor@suse.cz>
31752 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
31753 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
31754 it reflects the signature changes performed at the callee side.
31755 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
31756 to cgraph_build_function_type_skip_args.
31757 (build_function_decl_skip_args): Adjust call to the above function.
31759 2017-03-30 Jakub Jelinek <jakub@redhat.com>
31762 * config/i386/sse.md
31763 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
31764 register as dest whenever it is a MEM not rtx_equal_p to the
31765 corresponding dup operand, and when forcing into reg move the
31766 reg into the memory afterwards.
31767 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
31768 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
31769 for the force_reg mode.
31770 (avx512vl_vextractf128<mode>): Use register as dest either
31771 always when a MEM, or when it is a MEM not rtx_equal_p to the
31772 corresponding dup operand, or even not when it is a CONST_VECTOR
31773 depending on the mode and lo vs. hi.
31774 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
31776 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
31777 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
31778 Likewise. Require that operands[2] is even.
31779 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
31780 Remove extraneous parens. Require that operands[2] is a multiple
31782 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
31783 operands[0] is a MEM if <mask_applied>, the predicates/constraints
31784 disallow memory then.
31786 2017-03-30 Richard Biener <rguenther@suse.de>
31788 PR tree-optimization/77498
31789 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
31790 to non-constants over backedges.
31792 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
31794 PR rtl-optimization/80233
31795 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
31796 as last_combined_insn. Do not test for BARRIER_P separately.
31798 2017-03-29 Andreas Schwab <schwab@suse.de>
31801 * calls.c (prepare_call_address): Convert funexp to Pmode before
31802 copying to temp reg.
31804 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31806 PR tree-optimization/80158
31807 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
31808 Handle possible future case of more than one alternate
31810 (replace_rhs_if_not_dup): Likewise.
31811 (replace_one_candidate): Likewise.
31813 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
31815 PR rtl-optimization/80193
31816 * ira.c (ira): Do not check allocation for LRA.
31818 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
31820 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
31821 (nvptx_output_simt_exit): Declare.
31822 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
31823 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
31824 (init_softstack_frame): Move initialization of crtl->is_leaf to...
31825 (nvptx_declare_function_name): ...here. Emit declaration of local
31826 memory space buffer for omp_simt_enter insn.
31827 (nvptx_output_unisimt_switch): New.
31828 (nvptx_output_softstack_switch): New.
31829 (nvptx_output_simt_enter): New.
31830 (nvptx_output_simt_exit): New.
31831 * config/nvptx/nvptx.h (struct machine_function): New fields
31832 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
31833 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
31834 (UNSPECV_SIMT_EXIT): Ditto.
31835 (omp_simt_enter_insn): New insn.
31836 (omp_simt_enter): New expansion.
31837 (omp_simt_exit): New insn.
31838 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
31840 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
31841 (expand_GOMP_SIMT_ENTER_ALLOC): New.
31842 (expand_GOMP_SIMT_EXIT): New.
31843 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
31844 (GOMP_SIMT_ENTER_ALLOC): Ditto.
31845 (GOMP_SIMT_EXIT): Ditto.
31846 * target-insns.def (omp_simt_enter): New insn.
31847 (omp_simt_exit): Ditto.
31848 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
31850 (lower_rec_simd_input_clauses): Implement SIMT privatization.
31851 (lower_rec_input_clauses): Likewise.
31852 (lower_lastprivate_clauses): Handle SIMT privatization.
31854 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
31855 (ompdevlow_adjust_simt_enter): New.
31856 (find_simtpriv_var_op): New.
31857 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
31858 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
31860 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
31861 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
31862 (copy_decl_for_dup_finish): Ditto.
31864 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
31866 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
31869 * config/i386/i386.c (ix86_option_override_internal): Always
31870 allow -mpreferred-stack-boundary=3 for 64-bit targets.
31872 2017-03-28 Bin Cheng <bin.cheng@arm.com>
31874 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
31876 2017-03-28 Bin Cheng <bin.cheng@arm.com>
31878 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
31879 mark new edge's irreducible flag accordign to it.
31880 (vect_do_peeling): Check loop preheader edge's irreducible flag
31881 and pass it to function slpeel_add_loop_guard.
31883 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
31885 PR tree-optimization/80218
31886 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
31887 Update block frequencies and counts.
31889 2017-03-28 Richard Biener <rguenther@suse.de>
31891 PR tree-optimization/78644
31892 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
31893 of a simplification result we may not use it at all.
31895 2017-03-28 Richard Biener <rguenther@suse.de>
31898 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
31899 without arguments, generate default definition of a SSA name.
31901 2017-03-28 Richard Biener <rguenther@suse.de>
31903 PR middle-end/80222
31904 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
31905 TYPE_REF_CAN_ALIAS_ALL references.
31906 * fold-const.c (fold_indirect_ref_1): Likewise.
31908 2017-03-28 Martin Liska <mliska@suse.cz>
31911 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
31912 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
31914 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31915 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
31917 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
31918 (EXTRA_SPECS): Define.
31919 (SUBTARGET_EXTRA_SPECS): Likewise.
31920 (SUBTARGET_CPP_SPEC): Likewise.
31921 * config/arc/elf.h (EXTRA_SPECS): Renamed to
31922 SUBTARGET_EXTRA_SPECS.
31923 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
31925 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31927 * config/arc/simdext.md (vst64_insn): Update pattern.
31928 (vld32wh_insn): Likewise.
31929 (vld32wl_insn): Likewise.
31930 (vld64_insn): Likewise.
31931 (vld32_insn): Likewise.
31933 2017-03-28 Marek Polacek <polacek@redhat.com>
31936 * fold-const.c (fold_comparison): Use protected_set_expr_location
31937 instead of SET_EXPR_LOCATION.
31939 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
31941 * tree.c (add_expr): Avoid name lookup warning.
31943 2017-03-27 Jeff Law <law@redhat.com>
31945 PR tree-optimization/80216
31946 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
31947 function name. Limit recursion depth.
31948 (record_temporary_equivalences): Corresponding changes.
31950 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
31952 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
31955 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31958 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
31960 * cfgcleanup.c (reg_note_cfa_p): New array.
31961 (insns_have_identical_cfa_notes): New function.
31962 (old_insns_match_p): Don't cross-jump in between /f
31963 and non-/f instructions. If both i1 and i2 are frame related,
31964 verify all CFA notes, their order and content.
31966 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
31969 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
31970 HImode and SImode with zero extend to DImode to one insn.
31971 (bswap<mode>2_extenddi): Likewise.
31972 (bswapsi2_extenddi): Likewise.
31973 (bswaphi2_extendsi): Likewise.
31974 (bswaphi2): Combine bswap HImode and SImode into one insn.
31975 Separate memory insns from swapping register.
31976 (bswapsi2): Likewise.
31977 (bswap<mode>2): Likewise.
31978 (bswaphi2_internal): Delete, no longer used.
31979 (bswapsi2_internal): Likewise.
31980 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
31981 store, and gpr<-gpr swap insns.
31982 (bswap<mode>2_store): Likewise.
31983 (bswaphi2_reg): Register only splitter, combine with the splitter.
31984 (bswaphi2 splitter): Likewise.
31985 (bswapsi2_reg): Likewise.
31986 (bswapsi2 splitter): Likewise.
31987 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
31988 the insns into load, store, and register/register insns.
31989 (bswapdi2_ldbrx): Likewise.
31990 (bswapdi2_load): Likewise.
31991 (bswapdi2_store): Likewise.
31992 (bswapdi2_reg): Likewise.
31994 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
31996 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
31997 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
31999 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
32002 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
32004 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32005 special handling for target option conflicts between dform
32006 options (-mpower9-dform, -mpower9-dform-vector,
32007 -mpower9-dform-scalar) and -mno-direct-move.
32009 2017-03-27 Richard Biener <rguenther@suse.de>
32011 PR tree-optimization/80181
32012 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
32014 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32016 * config/arc/predicates.md (move_double_src_operand): Replace the
32017 call to move_double_src_operand with a call to address_operand.
32019 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32021 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
32022 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
32023 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
32025 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32027 * config/arc/predicates.md (long_immediate_loadstore_operand):
32028 Consider scaled addresses cases.
32030 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32032 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
32033 restored when in interrupt.
32034 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
32035 doesn't have delay slot.
32037 2017-03-27 Richard Biener <rguenther@suse.de>
32040 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
32041 inlined thunk clones.
32043 2017-03-27 Jakub Jelinek <jakub@redhat.com>
32046 * asan.c (instrument_derefs): Copy over last operand from
32047 original COMPONENT_REF to the new COMPONENT_REF with
32048 DECL_BIT_FIELD_REPRESENTATIVE.
32049 * ubsan.c (instrument_object_size): Likewise.
32051 2017-03-27 Richard Biener <rguenther@suse.de>
32053 PR tree-optimization/80170
32054 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
32055 sure DR/SCEV didnt fold in constants we do not see when looking
32056 at the reference base alignment.
32058 2017-03-27 Richard Biener <rguenther@suse.de>
32060 PR middle-end/80171
32061 * gimple-fold.c (fold_ctor_reference): Properly guard against
32062 NULL return value from canonicalize_constructor_val.
32064 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
32067 * config/i386/i386.c (ix86_expand_builtin)
32068 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
32069 flags reg setting and flags reg using instructions.
32070 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
32071 clobbering instructions to zero extend op2.
32073 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
32075 * doc/install.texi (Configuration) <--with-aix-soname>:
32076 Update link to AIX ld.
32078 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
32080 PR rtl-optimization/80160
32081 PR rtl-optimization/80159
32082 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
32083 reg_alternate_class into account.
32085 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
32088 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
32089 to consider in curr_insn_transform.
32091 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32093 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
32094 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
32095 and emit_mode_inner.
32097 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32099 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
32100 argument to the overloaded builtin variants. Use the new flag to
32101 deprecate certain builtin variants.
32102 * config/s390/s390-builtin-types.def: Add new builtin types.
32103 * config/s390/s390-builtins.h: Support new flags field for
32104 overloaded builtins.
32105 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
32106 (s390_macro_to_expand): Enable vector float data type.
32107 (s390_cpu_cpp_builtins_internal): Indicate support of the new
32108 builtins by incrementing the __VEC__ version number.
32109 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
32111 (s390_resolve_overloaded_builtin): Emit error messages depending
32112 on the builtin flags.
32113 * config/s390/s390.c (s390_expand_builtin): Support additional
32114 flags argument. Change error message to match the messages
32115 emitted in s390-c.c.
32116 * config/s390/s390.md: New UNSPEC_* constants.
32117 (op_type): Add new instruction types.
32118 * config/s390/vecintrin.h: Add new builtins and test data class
32120 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
32121 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
32122 (VEC_INEXACT, VEC_NOINEXACT): New constants.
32123 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
32124 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
32125 ("vec_mergel<mode>"): V_HW -> VEC_HW.
32127 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
32128 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
32129 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
32130 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
32132 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
32133 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
32134 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
32135 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
32137 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
32138 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
32139 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
32140 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
32141 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
32142 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
32143 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
32145 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
32146 ("vec_scatter_element<V_HW_4:mode>_DI")
32147 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
32148 ("vec_fpint<mode>", "vflls")
32149 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
32150 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
32151 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
32152 ("*vec_cmphe<mode>_cc"): ... these.
32154 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
32155 mode constant instead of magic value.
32157 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32159 * config/s390/s390.c (s390_expand_vec_compare): Support other
32160 vector floating point modes than just V2DF.
32161 (s390_expand_vcond): Likewise.
32162 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
32163 (s390_cannot_change_mode_class): Prevent mode changes between TF
32164 and V1TF in vector registers.
32165 * config/s390/s390.md (DF, SF): New mode attributes.
32166 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
32167 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
32168 SFmode support for VRs.
32169 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
32171 (VFT, VF_HW): New mode iterators.
32172 (vw, sdx): New mode attributes.
32173 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
32174 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
32175 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
32176 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
32177 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
32178 also the new vector floating point modes. Renaming to ...
32180 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
32181 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
32182 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
32183 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
32184 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
32185 ("vec_unordered<mode>"): ... these.
32187 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
32188 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
32189 ("*vec_extendv2df"): New insn definitions.
32191 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32193 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
32194 ("mulditi3_2", "*muldi3_sign"): New patterns.
32195 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
32196 rename the pattern definition.
32198 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32200 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
32202 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
32204 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32206 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
32207 instruction if possible.
32208 * config/s390/vector.md (vec_halfnumelts): New mode
32210 ("*vec_vllezlf<mode>"): New pattern.
32212 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32214 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
32215 ("popcountv4si2", "popcountv2di2"): Rename to ...
32216 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
32217 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
32219 ("popcount<mode>2_vxe"): New pattern.
32221 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32223 * common/config/s390/s390-common.c (processor_flags_table): Add
32225 * config.gcc: Add arch12.
32226 * config/s390/driver-native.c (s390_host_detect_local_cpu):
32227 Default to arch12 for unknown CPU model numbers.
32228 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
32229 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
32230 PROCESSOR_max sanity check.
32231 * config/s390/s390-opts.h (enum processor_type): Add
32233 * config/s390/s390.c (processor_table): Add arch12.
32234 (s390_expand_builtin): Add check for B_VXE flag.
32235 (s390_issue_rate): Add PROCESSOR_ARCH12.
32236 (s390_get_sched_attrmask): Likewise.
32237 (s390_get_unit_mask): Likewise.
32238 (s390_sched_score): Enable z13 scheduling for arch12.
32239 (s390_sched_reorder): Likewise.
32240 (s390_sched_variable_issue): Likewise.
32241 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
32243 (s390_tune_attr): Use z13 scheduling also for arch12.
32244 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
32245 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
32246 (TARGET_VXE_P): New macros.
32247 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
32248 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
32249 * config/s390/s390.opt: Add arch12 as processor_type.
32251 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32253 * config/s390/s390.md
32254 ("fixuns_truncdddi2", "fixuns_trunctddi2")
32255 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
32256 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
32258 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
32259 Rename expanders to ...
32261 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
32262 ("fixuns_truncdddi2_emu"): ... these.
32264 ("fixuns_trunc<mode>si2_emu"): New expander.
32266 ("*fixuns_truncdfdi2_z13"): Rename to ...
32267 ("*fixuns_truncdfdi2_vx"): ... this.
32269 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32271 * config/s390/2964.md: Remove the single element vector compare
32272 instructions which are no longer used.
32273 * config/s390/s390.c (s390_select_ccmode): Remove handling of
32275 (s390_canonicalize_comparison): Remove handling of DFmode
32277 (s390_expand_vec_compare_scalar): Remove function.
32278 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
32279 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
32281 ("*cmp<mode>_ccs"): Add wfcdb instruction.
32283 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32285 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
32287 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
32288 will anyway by matched by mov<mode>_64dfp.
32290 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32292 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
32293 vlef/vstef. Add missing operand to vleif.
32295 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32297 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
32298 pair for all vector types with 64 bit elements.
32299 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
32300 * config/s390/vector.md (V_HW_64): ... here.
32301 (V_128_NOSINGLE): New mode iterator.
32302 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
32303 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
32304 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
32305 ("*vec_load_pairv2di"): Change to ...
32306 ("*vec_load_pair<mode>"): ... this one.
32308 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32310 * config/s390/constraints.md: Add comments.
32311 (jKK): Reject element sizes > 8 bytes.
32312 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
32314 * config/s390/s390.md: Add the s_operand checks formerly in
32315 s390_split_ok_p to various splitters where they are still
32317 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
32318 for 128 bit vectors. Plus two splitters.
32320 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32322 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
32325 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32328 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
32329 error if the boundary argument is not constant.
32331 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32333 PR rtl-optimization/80112
32334 * loop-doloop.c (doloop_condition_get): Don't check condition
32335 if cmp isn't SET with IF_THEN_ELSE src.
32337 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32339 PR tree-optimization/80158
32340 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
32341 replacing a candidate statement, also replace it for the
32342 candidate's alternate interpretation.
32343 (replace_rhs_if_not_dup): Likewise.
32344 (replace_one_candidate): Likewise.
32346 2017-03-24 Richard Biener <rguenther@suse.de>
32348 PR tree-optimization/80167
32349 * graphite-isl-ast-to-gimple.c
32350 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
32352 (translate_isl_ast_to_gimple::get_rename): Likewise.
32354 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32356 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
32357 handling of certain combinations of target options, including the
32358 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
32359 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
32361 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32364 * config/arm/arm.md (*load_multiple): Add reload_completed to
32365 matching condition.
32367 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32368 Richard Biener <rguenth@suse.de>
32370 PR tree-optimization/79908
32371 PR tree-optimization/80136
32372 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32373 been cast away, gimplify_and_add suffices.
32375 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
32377 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
32379 2017-03-23 Richard Biener <rguenther@suse.de>
32381 PR tree-optimization/80032
32382 * gimplify.c (gimple_push_cleanup): Forced unconditional
32383 cleanups still have to go to the conditional_cleanups
32386 2017-03-22 Jakub Jelinek <jakub@redhat.com>
32388 PR tree-optimization/80072
32389 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
32391 (next_operand_entry_id): Change type to unsigned int.
32392 (sort_by_operand_rank): Make sure to return the right return value
32393 even if unsigned fields are bigger than INT_MAX.
32394 (struct oecount): Change cnt and id type to unsigned int.
32395 (oecount_hasher::equal): Formatting fix.
32396 (oecount_cmp): Make sure to return the right return value
32397 even if unsigned fields are bigger than INT_MAX.
32398 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
32401 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
32402 TREE_READONLY on result if writing it more than once.
32405 * doc/invoke.texi (-fsanitize=thread): Document that with
32406 -fnon-call-exceptions atomics are not able to throw
32410 * tsan.c: Include tree-eh.h.
32411 (instrument_builtin_call): Call maybe_clean_eh_stmt or
32412 maybe_clean_or_replace_eh_stmt where needed.
32413 (instrument_memory_accesses): Add cfg_changed argument.
32414 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
32415 if it returned true.
32416 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
32418 PR rtl-optimization/63191
32419 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
32420 wrapper function, moved the whole old content into ...
32421 (ix86_delegitimize_address_1): ... this. New inline function.
32422 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
32423 true as last argument instead of ix86_delegitimize_address.
32425 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32427 * config/aarch64/aarch64.c (generic_branch_cost): Copy
32428 cortexa57_branch_cost.
32430 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32432 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
32434 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32437 * doc/md.texi (Constraints): Document wA constraint.
32438 * config/rs6000/constraints.md (wA): New.
32439 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
32440 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
32441 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
32442 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
32444 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
32447 * gimplify.c (is_oacc_declared): New function.
32448 (oacc_default_clause): Use it to set default flags for acc declared
32449 variables inside parallel regions.
32450 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
32451 declared variables.
32452 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
32453 declare attribute to any decl as necessary.
32455 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
32458 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
32459 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
32460 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
32461 (arm_arch_lpae): This.
32462 * config/arm/arm.c (arm_arch7ve): Rename into ...
32463 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
32464 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
32467 2017-03-22 Martin Liska <mliska@suse.cz>
32470 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
32471 error message instead of an ICE.
32473 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32475 * doc/extend.texi (6.11 Additional Floating Types): Revise.
32477 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
32479 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
32481 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32484 2017-03-21 Martin Sebor <msebor@redhat.com>
32486 * doc/extend.texi: Use "cannot" instead of "can't."
32487 * doc/hostconfig.texi: Same.
32488 * doc/install.texi: Same.
32489 * doc/invoke.texi: Same.
32490 * doc/loop.texi: Same.
32491 * doc/md.texi: Same.
32492 * doc/objc.texi: Same.
32493 * doc/rtl.texi: Same.
32494 * doc/tm.texi: Same.
32495 * doc/tm.texi.in: Same.
32496 * doc/trouble.texi: Same.
32498 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
32501 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
32502 (collect_checksum_attributes): Set it.
32503 (die_checksum_ordered): Use it.
32505 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32507 PR tree-optimization/79908
32508 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
32509 change: For a VA_ARG whose LHS has been cast away, use
32510 force_gimple_operand to construct the side effects.
32512 2017-03-21 David Malcolm <dmalcolm@redhat.com>
32514 PR translation/80001
32515 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
32516 more amenable to translation.
32517 (oacc_loop_auto_partitions): Likewise.
32519 2017-03-21 Marek Polacek <polacek@redhat.com>
32520 Martin Sebor <msebor@redhat.com>
32522 PR tree-optimization/80109
32523 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
32524 on INTEGRAL_TYPE_P.
32526 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32527 Segher Boessenkool <segher@kernel.crashing.org>
32530 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
32531 check reg_used_between_p between insn and one of succ or succ2
32532 depending on if succ is artificial insn not inserted into insn
32535 2017-03-21 Martin Liska <mliska@suse.cz>
32537 PR gcov-profile/80081
32538 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
32539 * doc/gcc.texi: Include gcov-dump stuff.
32540 * doc/gcov-dump.texi: New file.
32542 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
32544 PR rtl-optimization/79150
32545 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
32546 conditional jump, if the jump is the last insn of the loop.
32548 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32549 Richard Biener <rguenth@suse.de>
32551 PR tree-optimization/79908
32552 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32553 been cast away, use force_gimple_operand to construct the side
32556 2017-03-21 Martin Liska <mliska@suse.cz>
32558 PR libfortran/79956
32559 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
32562 2017-03-21 Brad Spengler <spender@grsecurity.net>
32565 * plugin.c (htab_hash_plugin): New function.
32566 (add_new_plugin): Use it and adjust.
32567 (parse_plugin_arg_opt): Adjust.
32568 (init_one_plugin): Likewise.
32570 2017-03-21 Richard Biener <rguenther@suse.de>
32572 PR tree-optimization/80032
32573 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
32574 if set force the cleanup to happen unconditionally.
32575 (gimplify_target_expr): Push inserted clobbers with force_uncond
32576 to avoid them being removed by control-dependent DCE.
32578 2017-03-21 Richard Biener <rguenther@suse.de>
32580 PR tree-optimization/80122
32581 * tree-inline.c (copy_bb): Do not expans va-arg packs or
32582 va_arg_pack_len when the inlined call stmt requires pack
32584 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
32586 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32589 * tsan.c (instrument_builtin_call): If the memory model argument
32590 is not a constant, assume it is valid.
32593 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
32596 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
32598 PR rtl-optimization/79910
32599 * combine.c (can_combine_p): Do not allow combining an I0 or I1
32600 if its dest is used by an insn before I2 (other than the combined
32601 insns themselves, which are properly handled already).
32603 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
32606 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
32608 * combine.c (record_used_regs): New static function.
32609 (try_combine): Handle situations where there is an additional
32610 instruction between I2 and I3 which needs to have a LOG_LINK
32614 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
32616 * combine.c (try_combine): Delete redundant i1 test. Call
32617 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
32619 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32622 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
32623 alternatives 13/14.
32625 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32627 PR tree-optimization/80054
32628 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
32629 the optimization if a PHI or any of its arguments is not dominated
32630 by the candidate's basis. Use gphi* rather than gimple* as
32632 (replace_profitable_candidates): Clean up a gimple* variable that
32633 should be a gphi* variable.
32635 2017-03-20 Martin Sebor <msebor@redhat.com>
32638 * doc/extend.texi (attribute constructor): Document present limitation.
32640 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
32643 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
32644 __POWER9_VECTOR__ #ifdef control, change template definition to
32645 use Power9-specific built-in function.
32646 (vec_any_eq): Likewise.
32647 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
32648 to control outcomes from this test.
32649 (vector_ae_<mode>p): For VEC_F modes, likewise.
32651 2017-03-20 Ian Lance Taylor <iant@google.com>
32653 * config/i386/i386.c (ix86_function_regparm): Save an extra
32654 register for -fsplit-stack with DECL_STATIC_CHAIN.
32656 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
32659 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
32660 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
32662 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
32664 * config/riscv/riscv.c (riscv_print_operand): Use "fence
32666 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
32669 2017-03-20 Marek Polacek <polacek@redhat.com>
32672 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
32674 2017-03-20 Richard Biener <rguenther@suse.de>
32676 PR tree-optimization/80113
32677 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
32678 allocate extra SSA name for PHI def.
32679 (add_close_phis_to_outer_loops): Likewise.
32680 (add_close_phis_to_merge_points): Likewise.
32681 (copy_loop_close_phi_args): Likewise.
32682 (copy_cond_phi_nodes): Likewise.
32684 2017-03-20 Martin Liska <mliska@suse.cz>
32686 PR middle-end/79753
32687 * tree-chkp.c (chkp_build_returned_bound): Do not build
32688 returned bounds for a LHS that's not a BOUNDED_P type.
32690 2017-03-20 Martin Liska <mliska@suse.cz>
32694 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
32695 COMPLEX_CST and VECTOR_CST.
32697 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32700 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
32701 target operand. A new splitter adds the clobber statement in case
32702 the target operand is dead anyway.
32704 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
32706 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
32707 to age-old versions of binutils and glibc.
32709 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
32711 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
32713 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
32715 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
32717 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
32719 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
32720 requirement for binutils 2.13.
32722 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
32724 * combine.c (try_combine): Delete redundant i1 test. Call
32725 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
32727 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
32729 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
32730 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
32732 <riscv64-*-elf>: Re-arrange section
32733 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
32734 <riscv32-*-linux>: Likewise.
32735 <riscv64-*-elf>: Likewise
32736 <riscv64-*-linux>: Likewise.
32738 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
32741 * aarch64.opt(verbose-cost-dump): Fix typo.
32743 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
32746 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
32747 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
32749 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
32751 * reload.c (find_reloads): When reloading a nonoffsettable address,
32752 use RELOAD_OTHER for it and its address reloads.
32754 PR rtl-optimization/79910
32755 * combine.c (record_used_regs): New static function.
32756 (try_combine): Handle situations where there is an additional
32757 instruction between I2 and I3 which needs to have a LOG_LINK
32760 2017-03-17 Jeff Law <law@redhat.com>
32762 PR tree-optimization/71437
32763 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
32764 conditional in the hash table first.
32765 (vrp_dom_walker::before_dom_children): Extract condition from
32766 ASSERT_EXPR. Record condition, its inverion and any implied
32767 conditions as well.
32769 2017-03-17 Marek Polacek <polacek@redhat.com>
32770 Markus Trippelsdorf <markus@trippelsdorf.de>
32772 PR tree-optimization/80079
32773 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
32776 2017-03-17 Richard Biener <rguenther@suse.de>
32778 PR middle-end/80075
32779 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
32780 Properly verify the LHS before the RHS possibly claims to be
32782 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
32785 2017-03-17 Martin Jambor <mjambor@suse.cz>
32787 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
32788 (List of -O2 options): Likewise.
32789 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
32792 2017-03-17 Tom de Vries <tom@codesourcery.com>
32794 * gcov-dump.c (print_usage): Print bug_report_url.
32796 2017-03-17 Richard Biener <rguenther@suse.de>
32798 PR middle-end/80050
32799 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
32800 (parser::peek): Likewise.
32802 2017-03-17 Richard Biener <rguenther@suse.de>
32804 PR tree-optimization/80048
32805 * sese.c (free_sese_info): Properly release rename_map and
32806 copied_bb_map elements.
32808 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
32810 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
32811 Add linked-list forward and backlinks. Insert on
32812 construction, remove on destruction.
32813 (class pass_store_merging): Add m_stores_head field.
32814 (pass_store_merging::terminate_and_process_all_chains):
32815 Iterate over m_stores_head list.
32816 (pass_store_merging::terminate_all_aliasing_chains):
32818 (pass_store_merging::execute): Check for debug stmts first.
32819 Push new chains onto the m_stores_head stack.
32821 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
32824 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
32825 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
32826 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
32828 2017-03-16 Jeff Law <law@redhat.com>
32830 PR tree-optimization/71437
32831 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
32832 member function. Implementation moved into after_dom_children
32833 member function and into the threader's thread_outgoing_edges
32835 (dom_opt_dom_walker::after_dom_children): Simplify by moving
32836 some code into new thread_outgoing_edges.
32837 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
32838 definition. Simplify marker handling (do it here). Assume we always
32839 have the available expression and the const/copies tables.
32840 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
32842 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
32843 * tree-vrp.c (equiv_stack): No longer file scoped.
32844 (vrp_dom_walker): New class.
32845 (vrp_dom_walker::before_dom_children): New member function.
32846 (vrp_dom_walker::after_dom_children): Likewise.
32847 (identify_jump_threads): Setup domwalker. Use it rather than
32848 walking edges in a random order by hand. Simplify setup/finalization.
32849 (finalize_jump_threads): Remove.
32850 (vrp_finalize): Do not call identify_jump_threads here.
32851 (execute_vrp): Do it here instead and call thread_through_all_blocks
32854 PR tree-optimization/71437
32855 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
32857 (simplify_stmt_for_jump_threading): Add basic_block argument. All
32859 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
32860 (dom_opt_dom_walker::thread_across_edge): Remove
32861 handle_dominating_asserts argument. All callers changed.
32862 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
32863 changes. Remove calls to lhs_of_dominating_assert. Other
32864 uses of handle_dominating_asserts turn into unconditional code
32865 (simplify_control_stmt_condition_1): Likewise.
32866 (simplify_control_stmt_condition): Likewise.
32867 (thread_through_normal_block, thread_across_edge): Likewise.
32868 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
32869 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
32870 object if it is not an SSA_NAME.
32871 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
32872 before calling into the VRP specific simplifiers.
32873 (identify_jump_threads): Remove handle_dominating_asserts
32876 2017-03-16 Jakub Jelinek <jakub@redhat.com>
32879 * tree-diagnostic.c (default_tree_printer): No longer static.
32880 * tree-diagnostic.h (default_tree_printer): New prototype.
32882 2017-03-16 Tamar Christina <tamar.christina@arm.com>
32884 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
32885 Change ins into fmov.
32887 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32889 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
32890 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
32891 Use h_con constraint for operand 1.
32892 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
32893 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
32895 2017-03-15 Jeff Law <law@redhat.com>
32897 PR tree-optimization/71437
32898 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
32899 (record_temporary_equivalences): Use it.
32901 PR tree-optimization/71437
32902 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
32903 tree-ssa-scopedtables.
32904 (lookup_avail_expr, build_and_record_new_cond): Likewise.
32905 (record_conditions, record_cond, vuse_eq): Likewise.
32906 (record_edge_info): Adjust to API tweak of record_conditions.
32907 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
32908 (record_temporary_equivalences, optimize_stmt): Likewise.
32909 (eliminate_redundant_computations): Likewise.
32910 (record_equivalences_from_stmt): Likewise.
32911 * tree-ssa-scopedtables.c: Include options.h and params.h.
32912 (vuse_eq): New function, moved from tree-ssa-dom.c
32913 (build_and_record_new_cond): Likewise.
32914 (record_conditions): Likewise. Accept vector of conditions rather
32915 than edge_equivalence structure for first argument.
32916 for the first argument.
32917 (avail_exprs_stack::lookup_avail_expr): New member function, moved
32918 from tree-ssa-dom.c.
32919 (avail_exprs_stack::record_cond): Likewise.
32920 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
32921 from tree-ssa-dom.c.
32922 (avail_exprs_stack): Add new member functions lookup_avail_expr
32924 (record_conditions): Declare.
32926 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
32929 * lra-constraints.c (process_alt_operands): Increase reject for
32930 reloading an input/output operand.
32932 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
32935 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
32936 insns to convert from signed/unsigned char/short to IEEE 128-bit
32938 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
32940 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
32943 * config/i386/i386.c (ix86_vector_duplicate_value): Create
32944 subreg of inner mode for values already in registers.
32946 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
32948 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
32949 iteration reg is used after the loop.
32951 2017-03-14 Martin Sebor <msebor@redhat.com>
32953 PR tree-optimization/79800
32954 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
32955 precision in negative-positive range.
32956 (format_floating): Call non-const overload with adjusted precision.
32958 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
32961 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
32964 2017-03-14 Martin Sebor <msebor@redhat.com>
32966 PR middle-end/80020
32967 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
32968 * builtins.def (aligned_alloc): Use it.
32971 * Makefile.in (GTFILES): Add calls.c.
32972 * calls.c: Include "gt-calls.h".
32974 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
32976 PR rtl-optimization/79728
32977 * regs.h (struct target_regs): New field
32978 x_contains_allocatable_regs_of_mode.
32979 (contains_allocatable_regs_of_mode): New macro.
32980 * reginfo.c (init_reg_sets_1): Initialize it, and change
32981 contains_reg_of_mode so it includes global regs as well.
32982 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
32983 rather than contains_regs_of_mode.
32985 2017-03-14 Martin Liska <mliska@suse.cz>
32987 * doc/invoke.texi: Document options that can't be combined with
32988 -fcheck-pointer-bounds.
32990 2017-03-14 Martin Liska <mliska@suse.cz>
32992 PR middle-end/79831
32993 * doc/invoke.texi (-Wchkp): Document the option.
32995 2017-03-14 Martin Liska <mliska@suse.cz>
32997 * Makefile.in: Install gcov-dump.
32999 2017-03-14 Martin Liska <mliska@suse.cz>
33001 * multiple_target.c (expand_target_clones): Bail out for
33002 an invalid attribute.
33004 2017-03-14 Richard Biener <rguenther@suse.de>
33006 * alias.c (struct alias_set_entry): Pack properly.
33007 * cfgloop.h (struct loop): Likewise.
33008 * cse.c (struct set): Likewise.
33009 * ipa-utils.c (struct searchc_env): Likewise.
33010 * loop-invariant.c (struct invariant): Likewise.
33011 * lra-remat.c (struct cand): Likewise.
33012 * recog.c (struct change_t): Likewise.
33013 * rtl.h (struct address_info): Likewise.
33014 * symbol-summary.h (function_summary): Likewise.
33015 * tree-loop-distribution.c (struct partition): Likewise.
33016 * tree-object-size.c (struct object_size_info): Likewise.
33017 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
33018 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
33019 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
33020 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
33021 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
33022 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
33023 (struct _stmt_vec_info): Likewise.
33025 2017-03-14 Martin Liska <mliska@suse.cz>
33028 * multiple_target.c (create_dispatcher_calls): Check that
33029 a target can create a function dispatcher.
33031 2017-03-14 Martin Liska <mliska@suse.cz>
33034 * multiple_target.c (expand_target_clones): Drop local.local
33035 flag for default implementation.
33037 2017-03-14 Richard Biener <rguenther@suse.de>
33039 PR tree-optimization/80030
33040 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
33042 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
33044 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
33045 gcc_fallthrough() instead of __attribute__((fallthrough));
33047 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
33049 * doc/gcc.texi: Remove "up" link to (DIR).
33050 * doc/gccint.texi: Ditto.
33052 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
33054 * doc/install.texi (Specific) <avr>: Remove reference to
33057 2017-03-13 Jeff Law <law@redhat.com>
33059 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
33060 attribute rather than comments.
33062 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
33063 match_scratch operand is highest.
33065 2017-03-13 Martin Liska <mliska@suse.cz>
33067 PR middle-end/78339
33068 * ipa-pure-const.c (warn_function_noreturn): If the declarations
33069 is a CHKP clone, use original declaration.
33071 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33073 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
33074 (arc_conditional_register_usage): Use a different allocation order
33075 when optimizing for size.
33076 * common/config/arc/arc-common.c (arc_option_optimization_table):
33077 Section anchors default on when optimizing for size.
33079 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33081 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
33083 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33085 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
33086 * config/arc/arc.md (cpu_facility): Add cd variant.
33087 (*movqi_insn): Add code density variant.
33088 (*movhi_insn): Likewise.
33089 (*movqi_insn): Likewise.
33090 (*addsi3_mixed): Likewise.
33091 (subsi3_insn): Likewise.
33093 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33095 * config/arc/arc.md (movsi_cond_exec): Update constraint.
33097 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33099 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
33100 expressions with MINUS and UNARY ops.
33102 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33105 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
33107 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
33108 between vec_select and vector argument.
33109 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
33110 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
33111 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
33112 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
33113 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
33114 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
33116 2017-03-13 Richard Biener <rguenther@suse.de>
33119 * params.def (vect-max-peeling-for-alignment): Fix typo.
33121 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33123 * doc/install.texi (Specific) <mips-*-*>: Remove description of
33124 issue that only occurred with binutils below 2.18.
33126 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33128 * doc/install.texi (Specific) <cris-axis-elf>: No longer
33129 refer to binutils 2.11/2.12 minimum.
33131 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33133 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
33134 ftp.kernel.org and simplify binutils requirement.
33136 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
33138 * doc/invoke.texi (Warning Options): Fix spelling of link-time
33140 (Optimize Options): Ditto. Also remove redundancy.
33142 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33144 PR translation/79848
33145 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
33147 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
33148 to G_ to avoid double translation.
33150 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33152 PR translation/79923
33153 * auto-profile.c (get_combined_location): Convert leading
33154 character of diagnostics to lower case and remove trailing period.
33155 (read_profile): Likewise for various diagnostics.
33156 * config/arm/arm.c (arm_option_override): Remove trailing period
33157 from various diagnostics.
33158 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
33159 (msp430_expand_delay_cycles): Likewise.
33161 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33164 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
33165 full command-line argument, rather than just "str".
33166 (aarch64_validate_march): Likewise.
33167 (aarch64_validate_mtune): Likewise.
33169 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
33171 PR rtl-optimization/78911
33172 * lra-assigns.c (must_not_spill_p): New function.
33173 (spill_for): Use it.
33175 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33177 PR tree-optimization/79981
33178 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
33179 ATOMIC_COMPARE_EXCHANGE ifn result.
33180 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
33181 IFN_ATOMIC_COMPARE_EXCHANGE.
33183 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33186 * opts.c (parse_sanitizer_options): Add missing question mark to
33187 "did you mean" message.
33189 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33191 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
33193 (VMULEUH_UNS): Likewise.
33194 (VMULOUB_UNS): Likewise.
33195 (VMULOUH_UNS): Likewise.
33196 * config/rs6000/rs6000.c (builtin_function_type): Remove
33197 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
33199 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33202 * read-rtl-function.c (function_reader::read_rtx_operand): Update
33203 x with result of extra_parsing_for_operand_code_0.
33204 (function_reader::extra_parsing_for_operand_code_0): Convert
33205 return type from void to rtx, returning x. When reading
33206 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
33207 larger size containing struct block_symbol.
33209 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
33211 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
33212 -mfloat128-hardware without -m64.
33214 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
33217 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
33218 entries to the case statement that marks unsigned arguments to
33219 overloaded functions.
33221 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33223 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
33224 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
33226 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
33229 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
33230 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
33232 2017-03-10 Martin Liska <mliska@suse.cz>
33236 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
33237 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
33240 2017-03-10 Olivier Hainque <hainque@adacore.com>
33242 * tree-switch-conversion (array_value_type): Start by resetting
33243 candidate type to it's main variant.
33245 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33247 PR rtl-optimization/79909
33248 * combine.c (try_combine): Use simplify_replace_rtx on individual
33249 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
33250 of the whole CALL_INSN_FUNCTION_USAGE.
33252 PR tree-optimization/79972
33253 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
33254 get_range_info on SSA_NAMEs. Formatting fixes.
33256 2017-03-10 Richard Biener <rguenther@suse.de>
33257 Jakub Jelinek <jakub@redhat.com>
33259 PR tree-optimization/77975
33260 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
33261 edge to be constant.
33262 (get_val_for): For constant x return it. Formatting fix.
33263 (loop_niter_by_eval): Avoid pointless looping if the next iteration
33264 would use the same bases as the current one.
33266 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33268 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
33269 instead of vec_select for V1TImode.
33270 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
33272 (VSX_LE_128): Add V1TI to this mode iterator.
33273 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
33274 (*vsx_le_perm_store_<mode>): Likewise.
33275 (pre-reload splitter for VSX stores): Likewise.
33276 (post-reload splitter for VSX stores): Likewise.
33277 (*vsx_xxpermdi2_le_<mode>): Likewise.
33278 (*vsx_lxvd2x2_le_<mode>): Likewise.
33279 (*vsx_stxvd2x2_le_<mode>): Likewise.
33281 2017-03-09 Michael Eager <eager@eagercon.com>
33283 Correct failures with --enable-checking=yes,rtl.
33285 * config/microblaze/microblaze.c (microblaze_expand_shift):
33286 Replace GET_CODE test with CONST_INT_P and INTVAL test with
33287 test for const0_rtx.
33288 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
33289 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
33291 2017-03-09 Richard Biener <rguenther@suse.de>
33293 PR tree-optimization/79977
33294 * graphite-scop-detection.c (scop_detection::merge_sese):
33295 Handle the case of extra exits to blocks dominating the entry.
33297 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
33299 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
33302 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
33304 PR rtl-optimization/79949
33305 * lra-constraints.c (process_alt_operands): Check memory when
33306 trying to predict a cycle. Print about the overall increase.
33308 2017-03-09 Richard Biener <rguenther@suse.de>
33310 PR middle-end/79971
33311 * gimple-expr.c (useless_type_conversion_p): Preserve
33312 TYPE_SATURATING for fixed-point types.
33314 2017-03-09 Richard Biener <rguenther@suse.de>
33317 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
33318 alignment of BLKmode params.
33320 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33323 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
33324 (VALL_NO_V2Q): Likewise.
33326 * config/aarch64/aarch64-simd.md
33327 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
33329 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
33330 VALL_NO_V2Q mode iterator.
33331 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
33333 2017-03-09 Martin Liska <mliska@suse.cz>
33335 PR tree-optimization/79631
33336 * tree-chkp-opt.c (chkp_is_constant_addr): Call
33337 tree_int_cst_sign_bit just for INTEGER constants.
33339 2017-03-09 Martin Liska <mliska@suse.cz>
33343 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
33346 2017-03-09 Marek Polacek <polacek@redhat.com>
33349 * tree.c (inchash::add_expr): Handle TREE_VEC.
33351 2017-03-09 Martin Liska <mliska@suse.cz>
33354 (chkp_narrow_size_and_offset): New function.
33355 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
33356 (void chkp_parse_bit_field_ref): New function.
33357 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
33358 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
33360 2017-03-09 Martin Liska <mliska@suse.cz>
33363 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
33364 (chkp_find_bounds_1): Remove gcc_unreachable.
33366 2017-03-09 Jakub Jelinek <jakub@redhat.com>
33369 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
33370 BUILT_IN_SYNC*, determine the access type from the size suffix and
33371 always build a MEM_REF with that type. Handle forgotten
33372 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
33375 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
33376 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
33377 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
33378 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
33379 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
33380 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
33381 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
33382 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
33383 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
33384 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
33385 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
33386 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
33387 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
33388 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
33389 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
33390 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
33391 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
33392 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
33393 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
33394 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
33395 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
33396 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
33397 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
33398 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
33399 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
33400 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
33401 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
33402 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
33403 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
33404 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
33405 definitions outside of __OPTIMIZE__ guarded section.
33408 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
33409 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
33410 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
33411 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
33414 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33416 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
33417 ("vfenez<mode>"): Add missing constraints.
33419 2017-03-08 Martin Sebor <msebor@redhat.com>
33422 * config/nds32/nds32.c (nds32_option_override):
33423 Fix misspelled diagnostic.
33425 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33428 * gimplify.c (gimplify_omp_for): Replace index var in outer
33429 taskloop statement with an artificial variable and add
33430 OMP_CLAUSE_PRIVATE clause for it.
33432 2017-03-08 Richard Biener <rguenther@suse.de>
33434 PR tree-optimization/79955
33435 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
33436 for accesses that are completely outside of the variable.
33438 2017-03-08 Andrew Haley <aph@redhat.com>
33440 PR tree-optimization/79943
33441 * tree-ssa-loop-split.c (compute_new_first_bound): When
33442 calculating the new upper bound, (END-BEG) should be added, not
33445 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33447 * config/avr/avr.md (setmemhi): Make sure match_dup
33448 operand number comes before match_scratch.
33450 2017-03-08 Richard Biener <rguenther@suse.de>
33452 PR tree-optimization/79920
33453 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
33454 with ncopies == 1 to ...
33455 (vect_transform_slp_perm_load): ... here. Properly compute
33456 all element loads by iterating VF times over the group. Do
33457 not handle ncopies (computed in a broken way) in
33458 vect_create_mask_and_perm.
33460 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33463 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
33464 is a uniform vector, use uniform_vector_p return value instead of
33465 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
33467 2017-03-07 Marek Polacek <polacek@redhat.com>
33469 PR middle-end/79809
33470 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
33471 (alloca_call_type): Likewise.
33473 2017-03-07 Martin Liska <mliska@suse.cz>
33475 * gcov.c (process_args): Put comment to correct location.
33477 2017-03-07 Martin Liska <mliska@suse.cz>
33479 PR middle-end/68270
33480 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
33481 Use array_at_struct_end_p instead of DECL_CHAIN (field).
33482 (chkp_narrow_bounds_for_field): Likewise.
33483 (chkp_parse_array_and_component_ref): Pass one more argument to
33486 2017-03-07 Richard Biener <rguenther@suse.de>
33488 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
33491 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
33493 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
33494 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
33496 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33499 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
33500 to end of description.
33501 (PARAM_MAX_STORES_TO_MERGE): Likewise.
33503 2017-03-07 Jakub Jelinek <jakub@redhat.com>
33505 PR rtl-optimization/79901
33506 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
33508 (*avx512f_<code><mode>3<mask_name>): ... this.
33509 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
33510 iterator instead of VI8_AVX2_AVX512BW.
33512 PR rtl-optimization/79901
33513 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
33514 min/max expander, expand it using expand_vec_cond_expr.
33517 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
33520 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33523 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
33524 to void * for PCH reasons.
33525 * dwarf2out.c (output_loc_operands, output_die): Cast
33526 v.val_vec.array to unsigned char *.
33528 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
33531 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
33534 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
33536 PR rtl-optimization/79571
33537 * lra-constraints.c (process_alt_operands): Calculate static
33538 reject and subtract it from overall when only addresses will be
33541 2017-03-06 Julia Koval <julia.koval@intel.com>
33544 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
33545 incoming stack boundary to 128 for 64-bit targets.
33547 2017-03-06 Richard Biener <rguenther@suse.de>
33549 PR tree-optimization/79894
33550 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
33551 to NULL after folding it.
33553 2017-03-06 Richard Biener <rguenther@suse.de>
33555 PR tree-optimization/79824
33556 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
33557 check disabling peeling for gaps.
33559 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
33561 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
33562 attributes): Document gettimeofday.
33564 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33566 * config/s390/s390.c (s390_option_override_internal): Set
33567 PARAM_MIN_VECT_LOOP_BOUND
33569 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33571 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
33572 * config/s390/s390.md: Likewise.
33574 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33577 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
33578 (<avx2_avx512>_perm<mode>): Rename to ...
33579 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
33581 (<avx512>_perm<mode>_mask): Rename to ...
33582 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
33584 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
33585 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
33586 instead of VI8F_256_512.
33587 (avx512f_perm<mode>): New define_expand.
33588 (avx512f_perm<mode>_mask): Likewise.
33589 (avx512f_perm<mode>_1<mask_name>): New define_insn.
33590 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
33592 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33594 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
33595 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
33597 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
33599 2017-03-06 Martin Liska <mliska@suse.cz>
33602 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
33603 when having a SSA NAME w/o VAR_DECL assigned to it.
33605 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33607 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
33608 msa_dpsub_<su>_d): Fix MODE for vec_select.
33610 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33612 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
33614 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
33616 2017-03-06 Richard Biener <rguenther@suse.de>
33618 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
33619 * plugin.c (register_plugin_info): Likewise.
33620 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
33622 2017-03-05 Jakub Jelinek <jakub@redhat.com>
33624 * config/i386/sse.md (sse_storehps, sse_storelps,
33625 avx_<castmode><avxsizesuffix>_<castmode>,
33626 avx512f_<castmode><avxsizesuffix>_<castmode>,
33627 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
33628 in condition that at least one operand is not a MEM.
33630 2017-03-03 Jakub Jelinek <jakub@redhat.com>
33632 PR middle-end/79805
33633 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
33634 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
33636 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
33637 gimple_call_nothrow_p flag based on whether original builtin can throw.
33638 If it can, emit following stmts on the fallthrough edge.
33639 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
33640 don't create new bb if inserting just debug stmts on the edge, try to
33641 insert them on the fallthru bb or just reset debug stmts.
33643 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
33646 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
33647 restore recog_data (including the operand rtxes inside it) around
33648 the call to get_insn_template.
33650 2017-03-03 Martin Sebor <msebor@redhat.com>
33652 PR tree-optimization/79699
33653 * context.c (context::~context): Free MPFR caches to avoid
33654 a memory leak on program exit.
33656 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33658 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
33659 Use wide_int::ulow () instead of .elt (0).
33661 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
33663 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
33664 (*pushxf): Limit oF constraint to 32bit targets and add oC
33665 constraint for 64bit targets.
33666 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
33667 (*pushdf): Change rmF constraint to rmC.
33669 2017-03-03 Martin Liska <mliska@suse.cz>
33671 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
33672 Remove unused variable.
33674 2017-03-03 Jakub Jelinek <jakub@redhat.com>
33677 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
33678 is a memory operand, increase num_memory.
33679 (ix86_expand_args_builtin): Likewise.
33681 2017-03-03 Jan Hubicka <jh@suse.cz>
33684 * ipa-devirt.c (maybe_record_node): Properly handle
33685 __cxa_pure_virtual visibility.
33687 2017-03-03 Martin Liska <mliska@suse.cz>
33689 PR tree-optimization/79803
33690 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
33692 (pass_loop_prefetch::execute): Disabled optimization if an
33693 assumption about L1 cache size is not met.
33695 2017-03-03 Martin Liska <mliska@suse.cz>
33697 PR rtl-optimization/79574
33698 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
33699 (hash_scan_set): Likewise.
33700 (dump_hash_table): Likewise.
33701 (hoist_code): Likewise.
33703 2017-03-03 Richard Biener <rguenther@suse.de>
33705 * fixed-value.c (fixed_from_string): Restore use of elt (1)
33706 in place of uhigh ().
33707 (fixed_convert_from_real): Likewise.
33709 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
33712 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
33714 2017-03-03 Richard Biener <rguenther@suse.de>
33716 PR middle-end/79818
33717 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
33718 TYPE_OVERFLOW_UNDEFINED check.
33720 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33722 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
33724 (vector_ae_<mode>_p): Likewise.
33725 (vector_nez_<mode>_p): Likewise.
33726 (vector_ne_v2di_p): Likewise.
33727 (vector_ae_v2di_p): Likewise.
33728 (vector_ne_<mode>_p): Likewise.
33729 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
33731 (vsx_tsqrt<mode>2_fe): Likewise.
33733 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
33736 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
33738 2017-03-02 Jakub Jelinek <jakub@redhat.com>
33740 PR rtl-optimization/79780
33741 * cprop.c (one_cprop_pass): When second and further conditional trap
33742 in a single basic block is turned into an unconditional trap, turn it
33743 into a deleted note to avoid RTL verification failures.
33745 2017-03-02 Richard Biener <rguenther@suse.de>
33747 * fold-const.c (const_binop): Use ulow () instead of elt (0).
33749 2017-03-02 Richard Biener <rguenther@suse.de>
33751 PR tree-optimization/79345
33753 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
33754 param and abort the walk, returning -1 if it is hit.
33755 (walk_aliased_vdefs): Take a limit param and pass it on.
33756 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
33757 defaulting to 0 and return a signed int.
33758 * tree-ssa-uninit.c (struct check_defs_data): New struct.
33759 (check_defs): New helper.
33760 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
33761 about uninitialized memory.
33762 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
33763 bogus uninitialized warning.
33764 (fixed_convert_from_real): Likewise.
33766 2017-03-02 Bin Cheng <bin.cheng@arm.com>
33768 PR tree-optimization/66768
33769 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
33770 iv_use if base object can't be determined.
33772 2017-03-02 Jakub Jelinek <jakub@redhat.com>
33774 PR tree-optimization/79345
33775 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
33776 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
33777 (get_pattern_stats): Initialize it.
33778 * genemit.c (gen_expand): Verify match_scratch numbers come after
33779 match_operand/match_dup numbers.
33780 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
33781 match_scratch numbers.
33782 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
33784 * config/s390/s390.md (trunctdsd2): Likewise.
33786 2017-03-02 Richard Biener <rguenther@suse.de>
33788 * wide-int.h (wide_int_storage::operator=): Implement in terms
33791 2017-03-02 Richard Biener <rguenther@suse.de>
33793 PR tree-optimization/79777
33794 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
33795 the to insert expression to sth existing.
33797 2017-03-01 Martin Sebor <msebor@redhat.com>
33799 PR middle-end/79692
33800 * gimple-ssa-sprintf.c
33801 (directive::known_width_and_precision): New function.
33802 (format_integer): Use it.
33803 (get_mpfr_format_length): Consider the full range of precision
33804 when computing %g output with the # flag. Set the likely byte
33805 count to 3 rather than 1 when precision is indeterminate.
33806 (format_floating): Correct the lower bound of precision.
33808 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33810 * doc/invoke.texi: Document default code model for 64-bit Linux.
33812 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33815 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
33816 udiv rather than div since input pattern is unsigned.
33818 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
33820 * config/i386/i386.c (print_reg): Warn for values of
33821 unsupported size in integer register.
33823 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
33826 * config/rs6000/predicates.md (current_file_function_operand): Do
33827 not allow self calls to be local if the function is replaceable.
33829 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
33832 * config/rs6000/altivec.h (vec_ctz and others): Change the
33833 preprocessor macro that controls conditional compilation from
33834 _ARCH_PWR9 to __POWER9_VECTOR__.
33835 (vec_all_ne): Change parameterization of __altivec_scalar_pred
33836 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
33837 control (instead of _ARCH_PWR9 control) so that template
33838 definition uses power9-specific function.
33839 (vec_any_eq): Likewise.
33840 (vec_all_ne): Change macro definition to use a power9-specific
33841 expansion under #ifdef __POWER9_VECTOR__ control (instead of
33842 _ARCH_PWR9 control).
33843 (vec_any_eq) Likewise.
33844 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
33845 expansion for CMPNEF to remove support for xvcmpnesp instruction.
33846 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
33847 support for xvcmpnedp instruction.
33848 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
33849 macro expansion so that Power9 implementation of vec_all_ne does
33850 not use the AltiVec predicate framework.
33851 (VCMPNEH_P): Likewise.
33852 (VCMPNEW_P): Likewise.
33853 (VCMPNED_P): Likewise.
33854 (VCMPNEFP_P): Likewise.
33855 (VCMPNEDP_P): Likewise.
33856 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
33857 implementation of vec_any_eq to not use AltiVec predicate
33859 (VCMPAEH_P): Likewise.
33860 (VCMPAEW_P): Likewise.
33861 (VCMPAED_P): Likewise.
33862 (VCMPAEFP_P): Likewise.
33863 (VCMPAEDP_P): Likewise.
33864 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
33865 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
33866 not use the AltiVec predicate framework.
33867 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
33868 of vec_any_eq to not use AltiVec predicate framework.
33869 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
33870 support for predefined __POWER9_VECTOR__ macro to indicate that
33871 Power9 instruction selection is enabled.
33872 (altivec_overloaded_builtins): Remove extraneous
33873 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
33874 function argument types RS6000_BTI_bool_V16QI and
33875 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
33876 entry for overloaded function argument types RS6000_BTI_bool_V4SI
33877 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
33878 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
33879 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
33880 Power9 for implementations of vec_cmpne. Change the signature for
33881 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
33882 (representing vec_all_ne) to remove the previously described first
33883 argument of type RS6000_BTI_INTSI, as this was an artifact of
33884 reliance on the AltiVec predicate framework, which is no longer
33885 used in the implementation of these functions. Add
33886 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
33887 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
33888 since, unlike the AltiVec predicate framework implementation, we
33889 do not share function descriptors between vec_alle and vec_anyeq.
33890 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
33891 set of modes that receive special treatment even when
33892 TARGET_P9_VECTOR is true. The special treatment emits code that
33893 does not depend on Power9 instructions.
33894 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
33895 define_expand to not rely on AltiVec predicate framework.
33896 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33898 (vector_ne_v2di_p): Change this define_expand to not rely on
33899 AltiVec predicate framework.
33900 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
33902 (vector_ne_<mode>_p): Change this define_expand to not rely on
33903 AltiVec predicate framework.
33904 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33906 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
33907 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
33908 define_insn pattern.
33909 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
33910 define_insn pattern because the xvcmpne<VSs>. instruction is not
33912 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
33913 instruction is not supported.
33915 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33917 * config/nvptx/nvptx.c: Include intl.h.
33919 2017-03-01 Martin Jambor <mjambor@suse.cz>
33922 * ipa-prop.h (ipa_bits): Removed field known.
33923 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
33924 to pointers. Adjusted their comments to warn about their sharing.
33925 (ipcp_transformation_summary): Change bits to a vector of pointers.
33926 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
33927 (ipa_get_ipa_bits_for_value): Declare.
33928 * tree-vrp.h (value_range): Mark as GTY((for_user)).
33929 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
33930 (ipa_bits_hash_table): Likewise.
33931 (ipa_vr_ggc_hash_traits): Likewise.
33932 (ipa_vr_hash_table): Likewise.
33933 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
33934 being pointers and vr_known being removed.
33935 (ipa_set_jf_unknown): Likewise.
33936 (ipa_get_ipa_bits_for_value): New function.
33937 (ipa_set_jfunc_bits): Likewise.
33938 (ipa_get_value_range): New overloaded functions.
33939 (ipa_set_jfunc_vr): Likewise.
33940 (ipa_compute_jump_functions_for_edge): Use the above functions to
33941 construct bits and vr parts of jump functions.
33942 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
33943 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33945 (ipcp_grow_transformations_if_necessary): Also allocate
33946 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33948 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
33949 them. Fix too long lines.
33950 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
33951 vr_known being removed.
33952 (ipa_read_jump_function): Use new setter functions to construct bits
33953 and vr parts of jump functions or set them to NULL.
33954 (write_ipcp_transformation_info): Adjust for bits being pointers.
33955 (read_ipcp_transformation_info): Likewise.
33956 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
33958 Include gt-ipa-prop.h.
33959 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
33961 (ipcp_store_bits_results): Likewise.
33962 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
33963 Do not write to existing jump functions but use a temporary instead.
33965 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33968 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
33969 attempt to use its first operand as BIT_FIELD_REF base.
33971 2017-03-01 Richard Biener <rguenther@suse.de>
33973 PR middle-end/79721
33974 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
33975 interpolating formula in wrapping arithmetic.
33976 (chrec_apply): Convert chrec_evaluate return value to wanted type.
33978 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33980 PR tree-optimization/79734
33981 * tree-vect-generic.c (expand_vector_condition): Optimize
33982 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
33983 Handle VEC_COND_EXPR where comparison has different inner width from
33984 type's inner width.
33986 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
33988 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
33989 markup, and similar issues. Remove @opindex entries for things
33990 that aren't options. Add missing -mmpy-option entries.
33992 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33994 PR tree-optimization/79737
33995 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
33996 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
33997 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
33998 instead of byte_size. Formatting fix.
33999 (shift_bytes_in_array_right): Formatting fix.
34001 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
34004 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
34005 condition on optimize for the leaf function test.
34007 2017-02-28 Martin Liska <mliska@suse.cz>
34010 * read-rtl-function.c (function_reader::handle_unknown_directive):
34011 Bail out when one uses -flto.
34013 2017-02-28 Martin Liska <mliska@suse.cz>
34015 * common.opt: Replace space with tabular for options of <number>
34017 * config/i386/i386.opt: Show <number> value for
34018 -mlarge-data-threshold.
34019 * opts.c (print_filtered_help): Do not display number in hexadecimal
34022 2017-02-28 Martin Liska <mliska@suse.cz>
34024 * common.opt: Fix --help=option -Q for options which are of
34027 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
34029 * config/i386/i386.c (print_reg): Error out for values
34030 of 8-bit size in invalid integer register.
34032 2017-02-28 Martin Sebor <msebor@redhat.com>
34034 PR tree-optimization/79691
34035 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
34037 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34040 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
34041 gcc_unreachable with output_operand_lossage.
34043 2017-02-28 Richard Biener <rguenther@suse.de>
34045 PR tree-optimization/79740
34046 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
34048 (visit_nary_op): Insert the nary into the hashtable if we
34049 pattern-matched sth.
34050 * tree-ssa-pre.c (eliminate_insert): Robustify.
34052 2017-02-28 Richard Biener <rguenther@suse.de>
34054 PR middle-end/79731
34055 * fold-const.c (decode_field_reference): Reject out-of-bound
34058 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34060 * config/i386/i386.c: Include intl.h.
34061 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
34062 instead of just cond ? "..." : "...".
34063 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
34064 * coverage.c (read_counts_file): Likewise.
34065 * omp-offload.c: Include intl.h.
34066 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
34067 of just cond ? "..." : "...".
34068 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
34069 of just cond ? "..." : "...".
34071 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
34074 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
34076 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
34077 'tune for' CPU name.
34078 * config/arm/arm-cpu-data.h: Regenerated.
34080 2017-02-28 Richard Biener <rguenther@suse.de>
34082 PR tree-optimization/79732
34083 * tree-inline.c (expand_call_inline): Do not shadow var.
34085 2017-02-28 Richard Biener <rguenther@suse.de>
34087 PR tree-optimization/79723
34088 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
34089 address-space properly.
34091 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
34093 * doc/optinfo.texi (Optimization groups): Fix option used for
34095 * doc/invoke.texi (-fopt-info): Document "omp".
34096 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
34097 (OPTGROUP_ALL): Add OPTGROUP_OMP.
34098 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
34099 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
34100 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
34102 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
34104 * dumpfile.c (optgroup_options): Instead of "openmp", associate
34105 OPTGROUP_OMP with "omp".
34107 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
34110 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
34111 for arithmetic shift of unsigned V2DI.
34113 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
34115 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
34116 arc/linux.h headers.
34117 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
34118 (LINK_SPEC): Likewise.
34119 (ARC_TLS_EXTRA_START_SPEC): Likewise.
34120 (EXTRA_SPECS): Likewise.
34121 (STARTFILE_SPEC): Likewise.
34122 (ENDFILE_SPEC): Likewise.
34123 (LIB_SPEC): Likewise.
34124 (TARGET_SDATA_DEFAULT): Likewise.
34125 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
34126 (MULTILIB_DEFAULTS): Likewise.
34127 (DWARF2_UNWIND_INFO): Likewise.
34128 * config/arc/big.h: New file.
34129 * config/arc/elf.h: Likewise.
34130 * config/arc/linux.h: Likewise.
34131 * config/arc/t-uClibc: Remove.
34133 2017-02-27 Bin Cheng <bin.cheng@arm.com>
34135 PR tree-optimization/77536
34136 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
34137 (tree_transform_and_unroll_loop): Use above function to compute the
34138 estimated niter of unrolled loop and use it when scaling profile.
34139 Also use count info rather than frequency if it's non-zero.
34140 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
34141 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
34142 (vect_transform_loop): Call above function.
34144 2017-02-27 Richard Biener <rguenther@suse.de>
34146 PR tree-optimization/45397
34147 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
34148 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
34149 (visit_nary_op): Add pattern matching for CSEing sign-changed
34150 or truncated operations with wider ones.
34152 2017-02-27 Richard Biener <rguenther@suse.de>
34154 PR tree-optimization/79690
34155 * tree-vect-stmts.c (vectorizable_store): Use vector type
34156 built from the DR with address-space.
34158 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
34160 * doc/invoke.texi (Optimize Options): Refine the description
34161 of asan-use-after-return.
34163 2017-02-25 Alan Modra <amodra@gmail.com>
34165 PR rtl-optimization/79584
34166 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
34167 base, not ad->base_term, the reg within base. Remove assertion
34168 that ad->base == ad->base_term. Replace gen_int_mode using
34169 bogus mode with const0_rtx.
34171 2017-02-25 Jakub Jelinek <jakub@redhat.com>
34173 PR middle-end/79396
34174 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
34175 FMA_EXPR like tcc_binary or tcc_unary.
34177 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
34180 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
34182 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
34183 (output_loc_operands): Handle DW_OP_call_ref and
34184 DW_OP_GNU_variable_value.
34185 (struct variable_value_struct): New type.
34186 (struct variable_value_hasher): Likewise.
34187 (variable_value_hash): New variable.
34188 (string_types): Remove.
34189 (copy_loc_descr): New function.
34190 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
34191 (prepend_loc_descr_to_each): New function.
34192 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
34193 instead of add_loc_descr_to_each if the first argument is single
34194 location list and the second has multiple.
34195 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
34196 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
34197 when looking for variable value which doesn't have other location info.
34198 (loc_list_from_tree): Formatting fix.
34199 (gen_array_type_die): Simplify DW_AT_string_length handling.
34200 (adjust_string_types): Remove.
34201 (gen_subprogram_die): Don't call adjust_string_types nor test/set
34202 string_types. Call resolve_variable_values.
34203 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
34204 (resolve_addr_in_expr): Likewise. Add A argument.
34205 (copy_deref_exprloc): Remove deref argument. Adjust for the
34206 original expression being DW_OP_GNU_variable_value with optionally
34207 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
34208 optionally after it.
34209 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
34210 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
34211 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
34212 (variable_value_hasher::hash, variable_value_hasher::equal): New
34214 (resolve_variable_value_in_expr, resolve_variable_value,
34215 resolve_variable_values, note_variable_value_in_expr,
34216 note_variable_value): New functions.
34217 (dwarf2out_early_finish): Call note_variable_value on all toplevel
34220 2017-02-24 Jakub Jelinek <jakub@redhat.com>
34223 * opts.h (handle_generated_option): Add GENERATED_P argument.
34224 * opts-common.c (handle_option): Adjust function comment.
34225 (handle_generated_option): Add GENERATED_P argument, pass it to
34227 (control_warning_option): Pass false to handle_generated_option
34229 * opts.c (maybe_default_option): Pass true to handle_generated_option
34231 * optc-gen.awk: Likewise.
34233 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34235 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
34236 a REG, look at the REG it is a SUBREG of.
34237 (splitter for cmpeqsi_t): Ditto.
34239 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34241 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
34242 the special USEs with the pattern of the insn, not the insn itself.
34244 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
34247 * doc/invoke.texi: Document -mload-store-pairs.
34249 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34250 Sandra Loosemore <sandra@codesourcery.com>
34252 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
34253 argument isn't a CONST_INT.
34254 (nios2_alternate_compare_const): Assert op is a CONST_INT.
34255 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
34256 (nios2_validate_compare): Bypass alternate compare logic if *op2
34257 is not a CONST_INT.
34258 (ldstwm_operation_p): Return false if first_base is not a REG or
34259 if first_offset is not a CONST_INT.
34261 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34263 * config/cris/cris.md: Use correct operand in a define_peephole2.
34265 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34267 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
34269 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34271 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
34272 this_insn if it is an INSN or JUMP_INSN.
34273 (force_offsettable): Look at base, not at addr.
34274 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
34275 on things that aren't necessarily CONST_INTs.
34277 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
34279 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
34280 -mfpmath=sse is the default also for x86-32 targets with SSE2
34281 instruction set when @option{-ffast-math} is enabled
34283 2017-02-24 Jeff Law <law@redhat.com>
34285 PR rtl-optimizatoin/79286
34286 * ira.c (update_equiv_regs): Drop may_trap_p exception to
34289 2017-02-24 Richard Biener <rguenther@suse.de>
34291 PR tree-optimization/79389
34292 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
34295 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
34297 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
34298 function comment to reflect reality.
34299 (loop_exits_before_overflow): Fix typo in function description.
34301 2017-02-24 Richard Biener <rguenther@suse.de>
34303 PR tree-optimization/79389
34304 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
34305 properly that a threading opportunity exists. Detect conditional
34306 copy/constant propagation opportunities.
34308 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
34310 * config/visium/visium.md (type): Add trap.
34311 (b): New mode attribute.
34312 (*btst): Rename into...
34313 (*btst<mode>): ...this and adjust.
34314 (*cbranchsi4_btst_insn): Rename into...
34315 (*cbranch<mode>4_btst_insn): ...this and adjust.
34316 (trap): New define_insn.
34318 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34320 PR tree-optimization/79389
34321 * ifcvt.c (struct noce_if_info): Add rev_cond field.
34322 (noce_reversed_cond_code): New function.
34323 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
34324 reversed_comparison_code. Formatting fix.
34325 (noce_try_store_flag): Test rev_cond != NULL in addition to
34326 reversed_comparison_code.
34327 (noce_try_store_flag_constants): Likewise.
34328 (noce_try_store_flag_mask): Likewise.
34329 (noce_try_addcc): Use rev_cond if non-NULL instead of
34330 reversed_comparison_code.
34331 (noce_try_cmove_arith): Likewise. Formatting fixes.
34332 (noce_try_minmax, noce_try_abs): Clear rev_cond.
34333 (noce_find_if_block): Initialize rev_cond.
34334 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
34335 instead of false as last argument never attempt to reverse it
34338 2017-02-23 Bin Cheng <bin.cheng@arm.com>
34340 PR tree-optimization/79663
34341 * tree-predcom.c (combine_chains): Process refs in reverse order
34342 only for ZERO length chains, and add explaining comment.
34344 2017-02-23 Jeff Law <law@redhat.com>
34346 PR tree-optimization/79578
34347 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
34348 in call to operand_equal_p.
34350 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
34353 * config/i386/cpuid.h: Fix another undefined behavior.
34355 2017-02-23 Richard Biener <rguenther@suse.de>
34357 PR tree-optimization/79683
34358 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
34359 vector types for data-refs.
34361 2017-02-23 Martin Liska <mliska@suse.cz>
34363 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
34365 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34367 PR middle-end/79665
34368 * internal-fn.c (get_range_pos_neg): Moved to ...
34369 * tree.c (get_range_pos_neg): ... here. No longer static.
34370 * tree.h (get_range_pos_neg): New prototype.
34371 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
34372 are known to be in between 0 and signed maximum inclusive, try to
34373 expand both unsigned and signed divmod and use the cheaper one from
34376 2017-02-22 Jeff Law <law@redhat.com>
34378 PR tree-optimization/79578
34379 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
34380 to compare base operands.
34382 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
34385 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
34386 gpc_reg_operand instead of fpr_reg_operand.
34388 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
34390 * config/mips/mips.c (mips_return_in_memory): Force FP
34391 vector types to be returned in memory for o32 ABI.
34393 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34395 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
34396 instead of DW_TAG_member for static data member declarations and don't
34397 set no_linkage_name for static inline data members.
34398 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
34401 2017-02-22 Martin Liska <mliska@suse.cz>
34403 * doc/invoke.texi: Replace inequality signs with square brackets
34406 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
34409 * lra-constraints.c (simplify_operand_subreg): Handle
34410 WORD_REGISTER_OPERATIONS targets.
34412 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34415 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
34416 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
34417 elimination by swapping fld*.
34419 2017-02-22 Richard Biener <rguenther@suse.de>
34421 PR tree-optimization/79673
34422 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
34423 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
34424 irrelevant address-space qualifiers and avoiding a
34425 ADDR_SPACE_CONVERT_EXPR from fold_convert.
34427 2017-02-22 Richard Biener <rguenther@suse.de>
34429 PR tree-optimization/79666
34430 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
34431 to not symbolically negate if that may introduce undefined
34434 2017-02-22 Martin Liska <mliska@suse.cz>
34437 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
34438 * data-streamer-out.c (streamer_write_gcov_count_stream):
34440 * value-prof.c (stream_out_histogram_value): Make assert more
34441 precise based on type of counter.
34443 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
34446 * config/i386/i386.md (standard_x87sse_constant_load splitter):
34447 Use nonimmediate_operand instead of memory_operand for operand 1.
34448 (float-extend standard_x87sse_constant_load splitter): Ditto.
34450 2017-02-21 Jeff Law <law@redhat.com>
34452 PR tree-optimization/79621
34453 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
34454 blocks with edges to themselves.
34456 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34459 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
34460 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
34461 Use gimple_call_builtin_p.
34464 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
34465 on temporarily removed DEBUG_INSNs.
34467 PR tree-optimization/79649
34468 * tree-loop-distribution.c (classify_partition): Give up on
34469 non-generic address space loads/stores.
34471 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
34473 * doc/loop.texi (Loop manipulation): Remove nonexistent
34474 tree_ssa_loop_version from the documentation.
34475 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
34477 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34480 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
34481 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
34482 * config/rs6000/rs6000.c: Include except.h.
34483 (rs6000_expand_split_stack_prologue): Call
34484 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
34486 2017-02-21 Martin Jambor <mjambor@suse.cz>
34489 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
34490 have been analyzed.
34492 2017-02-21 Martin Jambor <mjambor@suse.cz>
34494 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
34495 for backward compatibility only.
34496 * doc/invoke.texi (Option Summary): Remove all references to
34497 -fipa-cp-alignment.
34499 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
34503 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34505 * lra-constraints.c (curr_insn_transform): Handle
34506 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34508 2017-02-21 Martin Liska <mliska@suse.cz>
34510 * config/i386/i386.opt: Replace -masm-dialect with -masm.
34512 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
34514 PR translation/79638
34515 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
34517 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
34520 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
34521 (arm_function_ok_for_sibcall): Return false for an indirect call by
34522 descriptor if all the argument registers are used.
34523 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
34524 alignment of the function.
34526 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34528 PR tree-optimization/61441
34529 * simplify-rtx.c (simplify_const_unary_operation): For
34530 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
34531 the sNaN unmodified.
34533 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34535 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34536 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34537 instead of SYSTEM_HEADER_DIR.
34539 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
34540 Martin Liška <mliska@suse.cz>
34542 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
34543 Fix typos and grammar, use active voice, and clarify.
34545 2017-02-20 Marek Polacek <polacek@redhat.com>
34547 PR middle-end/79537
34548 * gimplify.c (gimplify_expr): Handle unused *&&L;.
34551 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
34553 2017-02-20 Jakub Jelinek <jakub@redhat.com>
34556 * config/i386/i386.c (ix86_expand_builtin): Handle
34557 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
34558 ix86_builtins_isa[fcode].isa as a requirement of those
34559 flags and any other flag in the bitmask.
34560 (ix86_init_mmx_sse_builtins): Use 0 instead of
34561 ~OPTION_MASK_ISA_64BIT as mask.
34562 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
34563 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
34564 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
34565 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
34567 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34570 * lra-constraints.c (split_reg): Check requested split mode
34571 is supported by the register.
34573 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34575 * lra-constraints.c (simplify_operand_subreg): Remove early
34578 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34581 * lra-constraints.c (curr_insn_transform): Tighten condition
34582 for converting SUBREG reloads from OP_OUT to OP_INOUT.
34584 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34587 * lra-constraints.c (curr_insn_transform): Handle
34588 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34590 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
34593 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
34595 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
34597 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
34600 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
34603 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34605 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
34608 2017-02-18 Jakub Jelinek <jakub@redhat.com>
34610 * final.c (last_columnnum, override_columnnum): New variables.
34611 (final_start_function): Set last_columnnum, pass it to begin_prologue
34612 hook and pass 0 to dwarf2out_begin_prologue.
34613 (final_scan_insn): Update override_columnnum. Pass last_columnnum
34614 to source_line debug hook.
34615 (notice_source_line): Compute last_columnnum and for debug_column_info
34616 return true on column changes.
34617 * debug.h (struct gcc_debug_hooks): Add column argument to
34618 source_line and begin_prologue hooks.
34619 (debug_nothing_int_charstar_int_bool): Remove prototype.
34620 (debug_nothing_int_int_charstar,
34621 debug_nothing_int_int_charstar_int_bool): New prototypes.
34622 (dwarf2out_begin_prologue): Add column argument.
34623 * debug.c (do_nothing_debug_hooks): Adjust source_line and
34624 begin_prologue hooks.
34625 (debug_nothing_int_charstar_int_bool): Remove.
34626 (debug_nothing_int_int_charstar,
34627 debug_nothing_int_int_charstar_int_bool): New functions.
34628 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
34629 through to dwarf2out_source_line.
34630 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
34631 (dwarf2out_source_line): Add column argument, emit it if requested.
34632 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
34634 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
34635 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
34636 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
34637 through to dwarf2out_begin_prologue.
34638 (vmsdbgout_source_line): Add column argument, pass it through to
34639 dwarf2out_source_line.
34640 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
34641 dbxout_source_line caller.
34642 (dbxout_source_line): Add column argument.
34644 * common.opt (gno-column-info, gcolumn-info): New options.
34645 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
34646 (check_die): Also test for multiple DW_AT_decl_column attributes.
34647 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
34648 DW_AT_decl_column if requested.
34649 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
34651 (gen_variable_die): Likewise.
34652 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
34653 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
34656 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
34657 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
34658 (ix86_handle_option): Handle OPT_m3dnowa.
34659 * doc/invoke.texi (-m3dnowa): Document.
34660 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
34661 -m3dnowa instead of -m3dnow -march=athlon.
34664 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
34665 instead of gcc_assert for K, r and R code checks. Formatting fixes.
34667 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34670 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
34671 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
34672 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
34673 generator for vsx_xxpermdi_<mode>_be.
34674 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
34675 force big-endian semantics.
34676 (vsx_xxpermdi_<mode>_be): New define_expand with same
34677 implementation as previous version of vsx_xxpermdi_<mode>.
34679 2017-02-17 Jakub Jelinek <jakub@redhat.com>
34681 PR tree-optimization/79327
34682 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
34683 variable, its initialization and use.
34685 2017-02-17 Julia Koval <julia.koval@intel.com>
34687 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
34688 (OPTION_MASK_ISA_PKU_UNSET): New.
34689 (ix86_handle_option): Handle -mrdpid.
34690 * config/i386/cpuid.h (bit_RDPID): New.
34691 * config/i386/driver-i386.c (host_detect_local_cpu):
34692 Detect RDPID feature.
34693 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
34694 * config/i386/i386-c.c (ix86_target_macros_internal):
34696 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
34697 (ix86_valid_target_attribute_inner_p): Add "rdpid".
34698 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
34699 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
34700 * config/i386/i386.md (define_insn "rdpid"): New.
34701 * config/i386/i386.opt Add -mrdpid.
34702 * config/i386/immintrin.h (_rdpid_u32): New.
34704 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
34706 PR rtl-optimization/79541
34707 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
34708 instead of transforming it into USE.
34710 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
34712 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
34713 If HONOR_SNANS (SFmode) force the input to a register.
34714 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
34715 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
34716 an frsp or similar insn.
34718 2017-02-17 Martin Liska <mliska@suse.cz>
34720 PR rtl-optimization/79577
34721 * params.def (selsched-max-sched-times): Increase minimum to 1.
34723 2017-02-17 Martin Liska <mliska@suse.cz>
34725 PR rtl-optimization/79574
34726 * gcse.c (want_to_gcse_p): Prevent integer overflow.
34728 2017-02-17 Martin Liska <mliska@suse.cz>
34730 PR tree-optimization/79529
34731 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
34732 ssa_defined_default_def_p to handle cases which are implicitly
34734 * tree-ssa.c (ssa_defined_default_def_p): New function.
34735 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
34736 which are implicitly defined.
34737 * tree-ssa.h (ssa_defined_default_def_p): Declare.
34739 2017-02-17 Richard Biener <rguenther@suse.de>
34741 PR middle-end/79576
34742 * params.def (max-ssa-name-query-depth): Limit to 10.
34744 2017-02-17 Richard Biener <rguenther@suse.de>
34746 PR tree-optimization/79552
34747 * tree-ssa-structalias.c (visit_loadstore): Properly verify
34750 2017-02-17 Richard Biener <rguenther@suse.de>
34753 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
34755 2017-02-17 Marek Polacek <polacek@redhat.com>
34757 PR middle-end/79536
34758 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
34759 (fold_negate_expr): New wrapper.
34761 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
34763 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
34764 Correct terminology and de-emphasize pre-standard behavior.
34766 2017-02-16 Alan Modra <amodra@gmail.com>
34768 PR rtl-optimization/79286
34769 * ira.c (def_dominates_uses): New function.
34770 (update_equiv_regs): Don't create an equivalence for insns that
34771 may trap where the register def does not dominate the use.
34773 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
34775 PR rtl-optimization/78127
34776 * lra.c (lra): Call lra_eliminate before finish the loop after
34779 2017-02-16 Richard Biener <rguenther@suse.de>
34781 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
34783 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
34784 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
34785 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
34786 (isl_val_int_from_wi): New function.
34787 (extract_affine_gmp): Rename to ...
34788 (extract_affine_wi): ... this, take a widest_int.
34789 (extract_affine_int): Just wrap extract_affine_wi.
34790 (add_param_constraints): Use isl_val_int_from_wi.
34791 (add_loop_constraints): Likewise, and extract_affine_wi.
34793 2017-02-15 Jeff Law <law@redhat.com>
34795 PR middle-end/79521
34796 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
34797 ira_init_register_move_cost_if_necessary.
34799 2017-02-15 Martin Sebor <msebor@redhat.com>
34801 PR middle-end/32003
34802 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
34803 removed in a prior commit.
34805 2017-02-15 Bin Cheng <bin.cheng@arm.com>
34807 PR tree-optimization/79347
34808 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
34809 counters during peeling.
34811 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
34813 * Makefile.in (site.exp): Remove "set ISLVER".
34815 2017-02-15 Jakub Jelinek <jakub@redhat.com>
34818 * real.c (real_from_integer): Call real_convert even for decimal.
34820 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
34823 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
34825 2017-02-14 Andrew Pinski <apinski@cavium.com>
34827 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
34828 cores and change the partno/implementer to be correct.
34829 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
34830 the 'B" as the implementer.
34831 * config/aarch64/aarch64-tune.md: Regenerate.
34833 2017-02-14 Carl Love <cel@us.ibm.com>
34835 * config/rs6000/rs6000.c: Add case statement entry to make the
34836 xvcvuxdsp built-in argument unsigned.
34837 * config/rs6000/vsx.md: Fix the source and return operand types so they
34838 match the instruction definitions from the ISA document. Fix typo
34839 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
34842 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
34845 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
34846 member early_clobber_alts.
34847 * lra-lives.c (reg_early_clobber_p): New.
34848 (process_bb_lives): Use it.
34849 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
34850 (debug_operand_data): Initialize early_clobber_alts.
34851 (setup_operand_alternative): Set up early_clobber_alts.
34852 (collect_non_operand_hard_regs): Ditto. Pass early clobber
34853 alternatives to new_insn_reg.
34854 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
34856 (lra_update_insn_regno_info): Pass the new arg.
34858 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34860 PR middle-end/79505
34861 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
34862 (new_oacc_loop_raw): Don't clear already cleared fields.
34865 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
34866 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
34867 _mm512_prefetch_i64gather_ps): New inline functions and macros.
34869 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
34872 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
34874 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
34877 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
34878 the extra instruction to the right place to store 128-bit constant
34881 2017-02-14 Martin Sebor <msebor@redhat.com>
34883 PR middle-end/79448
34884 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
34885 warning for strings of unknown length.
34887 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
34889 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
34891 2017-02-14 Jeff Law <law@redhat.com>
34894 * ira-costs.c (scan_one_insn): Initialize register move costs
34895 for pseudos seen in USE/CLOBBER insns.
34897 PR tree-optimization/79095
34898 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
34899 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
34900 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
34901 if the operands are known to be not equal, then the resulting range
34903 (intersect_ranges): If the new range is ~[0,0] and the old range is
34904 wide, then prefer ~[0,0].
34905 * tree-vrp.c (overflow_comparison_p_1): New function.
34906 (overflow_comparison_p): New function.
34907 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
34908 if NAME is used in an overflow test.
34909 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
34910 overflow check that can be expressed as an equality test, then adjust
34911 ops to be that equality test.
34913 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34915 * config/s390/s390-builtin-types.def: Remove flags argument.
34916 * config/s390/s390.c (s390_init_builtins): Likewise.
34918 2017-02-14 Martin Liska <mliska@suse.cz>
34920 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
34921 vector. Fix trailing white spaces.
34923 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
34925 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
34928 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34930 PR rtl-optimization/68664
34931 * config/arm/arm.c (arm_sched_can_speculate_insn):
34932 New function. Declare prototype.
34933 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34935 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34937 PR rtl-optimization/68664
34938 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
34940 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34942 2017-02-14 Amit Pawar <amit.pawar@amd.com>
34944 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
34945 max skip bytes for function, loop and jump.
34947 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34949 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
34950 ABS_EXPR for gimple dump.
34952 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34955 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
34957 PR tree-optimization/79408
34958 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
34959 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
34960 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
34961 also if rhs1 is INTEGER_CST.
34963 2017-02-14 Richard Biener <rguenther@suse.de>
34965 PR middle-end/79432
34966 * tree-into-ssa.c (insert_phi_nodes): When the function can
34967 have abnormal edges rewrite SSA names with broken use-def
34968 dominance out of SSA and register them for PHI insertion.
34970 2017-02-13 Martin Sebor <msebor@redhat.com>
34972 PR middle-end/79496
34973 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
34974 clearing info.nowrite flag when snprintf size argument is a range.
34976 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34978 * cprop.c (cprop_jump): Add missing space in string literal.
34979 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
34980 (get_constraint_for_component_ref): Likewise.
34981 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
34982 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
34983 * lra-constraints.c (process_alt_operands): Likewise.
34984 * ipa-inline.c (inline_small_functions): Likewise.
34985 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
34986 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
34987 * trans-mem.c (diagnose_tm_1_op): Likewise.
34988 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
34989 (grid_parallel_clauses_gridifiable): Likewise.
34991 * config/nvptx/mkoffload.c (process): Add space in between
34994 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
34995 "MOD4_SSE_REGS" and "ALL_REGS".
34997 * spellcheck.c (test_data): Add , in between "foo" and "food".
34999 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35002 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
35003 boundary crossing check and subsequent code generation agree.
35005 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35007 * config/aarch64/aarch64.c (has_memory_op): Delete.
35008 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
35011 2017-02-13 Jakub Jelinek <jakub@redhat.com>
35013 PR rtl-optimization/79388
35014 PR rtl-optimization/79450
35015 * combine.c (distribute_notes): When removing TEM_INSN for which
35016 corresponding dest has last value recorded, invalidate that last
35019 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35021 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
35022 of explicit '@'. Add missing assembly comment marker on branch costs
35025 2017-02-13 Nathan Sidwell <nathan@acm.org>
35027 * gengtype-lex.l (<in_struct>): Add '/'.
35029 2017-02-13 Martin Liska <mliska@suse.cz>
35032 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
35034 2017-02-13 Richard Biener <rguenther@suse.de>
35036 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
35038 * configure: Re-generate.
35039 * config.in: Likewise.
35040 * graphite-dependences.c: Simplify as if
35041 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
35042 * graphite-isl-ast-to-gimple.c: Likewise.
35043 * graphite-optimize-isl.c: Likewise.
35044 * graphite-poly.c: Likewise.
35045 * graphite-sese-to-poly.c: Likewise.
35046 * graphite.h: Likewise.
35047 * toplev.c: Include isl/version.h and use isl_version () for
35048 printing the ISL version.
35049 * doc/install.texi: Update ISL requirement.
35051 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
35053 * doc/standards.texi (Standards): Update reference to
35056 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
35058 * doc/extend.texi (Named Address Spaces): sourceware.org now
35060 * doc/install.texi (Binaries): Ditto.
35063 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
35065 * doc/cpp.texi: Replace "stringify"/"stringification" with C
35066 standard terminology "stringize"/"stringizing" throughout.
35067 * doc/cppinternals.texi: Likewise.
35069 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
35071 * doc/extend.texi: Fix some spelling mistakes and typos.
35072 * doc/invoke.texi: Likewise.
35074 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
35077 * params.def (inline-min-speedup) Change from 10 to 8.
35079 2017-02-11 Jakub Jelinek <jakub@redhat.com>
35081 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
35084 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
35087 * ipa-inline-analysis.c (get_minimal_bb): New function.
35088 (record_modified): Use it.
35089 (remap_edge_change_prob): Handle also ancestor functions.
35091 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
35093 * doc/contrib.texi (Contributors): Remove broken link into
35094 the Mauve CVS repository.
35096 2017-02-11 Jakub Jelinek <jakub@redhat.com>
35098 PR middle-end/79454
35099 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
35100 result computation whenever lhs doesn't have vector mode, not
35101 just when it has BLKmode.
35103 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
35105 * doc/makefile.texi (profiledbootstrap): Refer to the
35106 installation instructions only in textual form.
35108 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35111 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
35113 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
35115 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
35116 (Specific): Update mingw-w64 reference.
35118 (Specific): Remove broken link to Renesas RX processor.
35120 2017-02-10 Richard Biener <rguenther@suse.de>
35122 * toplev.c (process_options): Do not mention obsolete graphite
35123 options when printing sorry message about missing graphite support.
35124 Mention -floop-nest-optimize.
35126 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
35128 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
35129 (vtst_p16): Likewise.
35130 (vtstq_p8): Likewise.
35131 (vtstq_p16): Likewise.
35133 (vtstq_p64): Likewise.
35134 * config/arm/arm_neon.h (vgetq_lane_p64): New.
35135 (vset_lane_p64): New.
35136 (vsetq_lane_p64): New.
35138 2017-02-10 Jakub Jelinek <jakub@redhat.com>
35140 PR tree-optimization/79411
35141 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
35142 stmt operands are SSA_NAMEs used in abnormal phis.
35143 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
35146 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35149 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
35152 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35154 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
35156 2017-02-09 Jakub Jelinek <jakub@redhat.com>
35158 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
35162 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
35163 not arbitrary TREE_CONSTANT.
35166 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
35167 "omp declare target link" attribute unless is_global_var.
35168 * omp-offload.c (find_link_var_op): Likewise.
35170 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
35171 Chung-Lin Tang <cltang@codesourcery.com>
35173 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
35175 (gimplify_adjust_omp_clauses): Don't delete TILE.
35176 (gimplify_omp_for): Deal with TILE.
35177 * internal-fn.c (expand_GOACC_TILE): New function.
35178 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
35180 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
35181 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
35183 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
35184 avoid DIV for outermost collapse var.
35185 (expand_oacc_for): Insert tile element loop as needed. Adjust.
35186 Remove out of date comments, fix whitespace.
35187 * omp-general.c (omp_extract_for_data): Deal with tiling.
35188 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
35189 adjust OLF_DIM_BASE value.
35190 (struct omp_for_data): Add tiling field.
35191 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
35192 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
35193 for auto loops. Remove default auto determining, moved to
35194 oacc_loop_fixed_partitions.
35195 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
35196 stmts, add e_mask field.
35197 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
35198 (oacc_thread_numbers): Use oacc_dim_call.
35199 (oacc_xform_tile): New.
35200 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
35201 (finish_oacc_loop): Adjust for ifns vector.
35202 (oacc_loop_discover_walk): Append loop abstraction sites to list,
35203 add case for GOACC_TILE fns.
35204 (oacc_loop_xform_loop): Delete.
35205 (oacc_loop_process): Iterate over call list directly, and add
35206 handling for GOACC_TILE fns.
35207 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
35209 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
35210 vector partitioning to outer loops. Assign 2 partitions to loops
35211 when available. Add TILE handling.
35212 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
35213 (execite_oacc_device_lower): Process GOACC_TILE fns,
35214 ignore unknown specs.
35215 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
35216 * tree.c (omp_clause_num_ops): Adjust TILE ops.
35217 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
35219 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
35221 * configure.ac (ACX_BUGURL): Update.
35222 * configure: Regenerate.
35224 2017-02-09 Richard Biener <rguenther@suse.de>
35226 PR tree-optimization/69823
35227 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
35228 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
35230 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35232 * config/arc/arc-c.def: Add __NPS400__ definition.
35233 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
35234 (TARGET_NPS400): Define.
35236 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35238 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
35240 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
35241 pointer, arch_info.
35242 (arc_cpu_types): Fill the arch_info field with a pointer into the
35243 arc_arch_types table.
35244 (arc_selected_cpu): Declare.
35245 * config/arc/arc.c (arc_selected_cpu): Make global.
35246 (arc_selected_arch): Delete.
35247 (arc_base_cpu): Delete.
35248 (arc_override_options): Remove references to deleted variables,
35249 update access to arch information.
35250 (ARC_OPT): Update access to arch information.
35251 (ARC_OPTX): Likewise.
35252 * config/arc/arc.h (arc_base_cpu): Remove declaration.
35253 (TARGET_ARC600): Update access to arch information.
35254 (TARGET_ARC601): Likewise.
35255 (TARGET_ARC700): Likewise.
35256 (TARGET_EM): Likewise.
35257 (TARGET_HS): Likewise.
35258 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
35261 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
35264 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
35265 condition/operands for integer GE/LE/GEU/LEU operations.
35267 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
35269 PR translation/79397
35270 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
35273 2017-02-08 Martin Jambor <mjambor@suse.cz>
35276 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
35277 whether allocation happened.
35278 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
35279 nothing was allocated.
35281 2017-02-08 Jakub Jelinek <jakub@redhat.com>
35283 PR tree-optimization/79408
35284 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
35285 constant, but SSA_NAME with a known integer range, use the minimum
35286 of that range instead of op1 to determine if modulo can be replaced
35287 with its first operand.
35289 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35291 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
35293 2017-02-08 Richard Biener <rguenther@suse.de>
35295 PR tree-optimization/71824
35296 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35297 Check all loops contained in the merged region.
35299 2017-02-07 Andrew Pinski <apinski@cavium.com>
35301 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
35303 2017-02-07 Andrew Pinski <apinski@cavium.com>
35305 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
35306 (thunderxt88): Likewise.
35307 (thunderxt81): Disable LSE and change v8.1 to v8.
35308 (thunderxt83): Likewise.
35310 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35311 Richard Biener <rguenther@suse.de>
35313 PR middle-end/79399
35314 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
35315 type from int to size_t.
35316 * ira-costs.c (struct_costs_size): Change type from int to size_t.
35318 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35320 PR rtl-optimization/79386
35321 * cprop.c (bypass_conditional_jumps): Initialize
35322 bypass_last_basic_block already before splitting bbs after
35323 unconditional traps...
35324 (bypass_conditional_jumps): ... rather than here.
35327 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
35328 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
35329 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
35330 fix -masm=intel patterns.
35332 2017-02-07 Richard Biener <rguenther@suse.de>
35334 PR tree-optimization/79256
35335 PR middle-end/79278
35336 * builtins.c (get_object_alignment_2): Use min_align_of_type
35337 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
35338 and ADJUST_FIELD_ALIGN.
35340 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
35342 * doc/tm.texi: Regenerate.
35343 * stor-layout.c (layout_decl): Adjust.
35344 (update_alignment_for_field): Likewise.
35345 (place_field): Likewise.
35346 (min_align_of_type): Likewise.
35347 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
35348 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
35349 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
35350 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
35351 * config/frv/frv.c (frv_adjust_field_align): Likewise.
35352 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
35353 * config/i386/i386.c (x86_field_alignment): Likewise.
35354 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
35355 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
35356 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
35357 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
35358 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
35359 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
35363 2017-01-30 Richard Biener <rguenther@suse.de>
35365 PR tree-optimization/79256
35366 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35367 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35370 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
35372 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
35373 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
35374 builtins to SImode and emit a zero-extend, if necessary.
35376 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35378 * docs/invoke.texi (RISC-V Options): Alphabetize.
35380 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35382 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
35385 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35387 * config/riscv/riscv.c: New file.
35388 * common/config/riscv/riscv-common.c: Likewise.
35389 * config.gcc: Likewise.
35390 * config/riscv/constraints.md: Likewise.
35391 * config/riscv/elf.h: Likewise.
35392 * config/riscv/generic.md: Likewise.
35393 * config/riscv/linux.h: Likewise.
35394 * config/riscv/multilib-generator: Likewise.
35395 * config/riscv/peephole.md: Likewise.
35396 * config/riscv/pic.md: Likewise.
35397 * config/riscv/predicates.md: Likewise.
35398 * config/riscv/riscv-builtins.c: Likewise.
35399 * config/riscv/riscv-c.c: Likewise.
35400 * config/riscv/riscv-ftypes.def: Likewise.
35401 * config/riscv/riscv-modes.def: Likewise.
35402 * config/riscv/riscv-opts.h: Likewise.
35403 * config/riscv/riscv-protos.h: Likewise.
35404 * config/riscv/riscv.h: Likewise.
35405 * config/riscv/riscv.md: Likewise.
35406 * config/riscv/riscv.opt: Likewise.
35407 * config/riscv/sync.md: Likewise.
35408 * config/riscv/t-elf-multilib: Likewise.
35409 * config/riscv/t-linux: Likewise.
35410 * config/riscv/t-linux-multilib: Likewise.
35411 * config/riscv/t-riscv: Likewise.
35412 * configure.ac: Likewise.
35413 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
35414 Waterman as RISC-V maintainers.
35415 * doc/install.texi: Add RISC-V entries.
35416 * doc/invoke.texi: Add RISC-V options section.
35417 * doc/md.texi: Add RISC-V constraints section.
35418 * configure: Regenerated.
35420 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
35423 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
35424 false values to be constant vectors with all 0 or all 1 bits set.
35425 (vcondu<mode><mode>): Likewise.
35426 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
35428 (fpmask_comparison_operator): Update comment.
35429 (vecint_comparison_operator): New predicate.
35430 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
35431 vector conditionals when the true and false values are constant
35432 vectors with all 0 bits or all 1 bits set.
35434 2017-02-06 Martin Sebor <msebor@redhat.com>
35436 PR tree-optimization/79376
35437 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
35439 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
35441 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
35442 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
35443 to simplify split condition.
35445 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35447 * omp-expand.c (oxpand_omp_atomic_fetch_op,
35448 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
35451 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
35453 PR rtl-optimization/68664
35454 * target.def (can_speculate_insn): New hook.
35455 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
35456 * doc/tm.texi: Regenerate.
35457 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
35458 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
35459 (rs6000_sched_can_speculate_insn): New function.
35461 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35463 PR tree-optimization/79284
35464 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
35465 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
35466 vectorizable_mask_load_store, vectorizable_operation,
35467 vect_is_simple_cond, get_same_sized_vectype): Use it instead
35468 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
35469 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
35470 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
35471 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
35472 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
35473 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
35474 is_gimple_assign (stmt). Replace another such test with
35475 is_gimple_assign (stmt).
35477 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35480 * config/avr/avr.c (rtl-iter.h): Include it.
35481 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
35482 (avr_legitimate_combined_insn): ...and implementation.
35484 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
35486 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
35487 * config/s390/s390.c (s390_const_operand_ok)
35488 (s390_canonicalize_comparison, s390_extract_part)
35489 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
35490 (s390_contiguous_bitmask_p, s390_rtx_costs)
35491 (legitimize_pic_address): Likewise.
35492 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
35493 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
35494 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
35495 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
35496 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
35498 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35500 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
35501 REGNO($0) == REGNO($1).
35503 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35505 * config/s390/linux.h(SIZE_TYPE): Add comment.
35507 2017-02-06 Julian Brown <julian@codesourcery.com>
35508 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35509 Virendra Pathak <virendra.pathak@broadcom.com>
35511 * config/aarch64/aarch64-cores.def: Change the scheduler
35513 * config/aarch64/aarch64.md: Include thunderx2t99.md.
35514 * config/aarch64/thunderx2t99.md: New file.
35516 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35518 * doc/standards.texi (Go Language): Update link to language
35521 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35523 * tree-eh.c (lower_resx): Sanitize profile.
35524 (cleanup_empty_eh_move_lp): Likewise.
35526 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35529 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
35531 * cfgloopmanip.h (loop_version): Update prototype.
35532 * modulo-sched.c (sms_schedule): Update call of loop_version.
35533 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
35534 * tree-parloops.c (gen_parallel_loop): Likewise.
35535 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
35536 * tree-ssa-loop-split.c (split_loop): Likewise.
35537 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
35538 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
35540 2017-02-05 Martin Liska <mliska@suse.cz>
35543 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
35545 (print_operand_address): Initialize a struct to zero.
35547 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35549 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
35550 garbage collector only in textual form.
35552 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35554 * doc/extend.texi (x86 specific memory model extensions for
35555 transactional memory): Simplify a phrase.
35557 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
35560 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
35561 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
35562 (atomic_storedi_1): Likewise.
35564 2017-02-04 Jakub Jelinek <jakub@redhat.com>
35566 PR tree-optimization/79338
35567 * tree-parloops.c (gather_scalar_reductions): Don't call
35568 vect_analyze_loop_form for loop->inner before destroying loop's
35571 2017-02-03 Martin Sebor <msebor@redhat.com>
35573 PR tree-optimization/79327
35574 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
35575 when precision has resulted in leading zeros.
35576 (format_integer): Adjust the likely counter to assume an unknown
35577 argument that may be zero is non-zero.
35579 2017-02-03 Jason Merrill <jason@redhat.com>
35582 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
35583 avoid copying non-taken branch.
35585 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35587 PR tree-optimization/79340
35588 * tree-vect-loop.c (vectorizable_reduction): Release
35589 vec_defs elements after safe_splicing them into other vectors.
35592 PR tree-optimization/79327
35593 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
35594 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
35596 (format_integer): Use wide_int_to_tree instead of build_int_cst
35597 + to_?hwi. If argmin is NULL, just set argmin and argmax to
35598 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
35599 of shortest and longest sequence.
35601 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
35603 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
35604 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
35606 2017-02-03 Walter Lee <walt@tilera.com>
35609 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
35610 after initial stackframe link reg save.
35611 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
35613 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35616 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
35617 wu for stxssp alternative.
35619 2017-02-03 Martin Sebor <msebor@redhat.com>
35621 PR tree-optimization/79352
35622 * gimple-fold.c (get_range_strlen): Add argument.
35623 (get_range_strlen): Change return type to bool.
35624 (get_maxval_strlen): Pass in a dummy argument.
35625 * gimple-fold.h (get_range_strlen): Change return type to bool.
35626 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
35627 * tree.h (array_at_struct_end_p): Add argument.
35628 * tree.c (array_at_struct_end_p): Handle it.
35630 2017-02-03 Martin Liska <mliska@suse.cz>
35633 * multiple_target.c (create_dispatcher_calls): Redirect edge
35634 from a caller of a dispatcher.
35635 (expand_target_clones): Make the clones local.
35636 (ipa_target_clone): Do both target clones and resolvers.
35637 (ipa_dispatcher_calls): Remove the pass.
35638 (pass_dispatcher_calls::gate): Likewise.
35639 (make_pass_dispatcher_calls): Likewise.
35640 * passes.def (pass_target_clone): Put as very first IPA early
35643 2017-02-03 Martin Liska <mliska@suse.cz>
35645 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
35646 in case of a function with ifunc attribute.
35648 2017-02-03 Martin Liska <mliska@suse.cz>
35650 * cgraph.c (cgraph_node::dump): Dump function version info.
35651 * symtab.c (symtab_node::dump_base): Add missing new line.
35653 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
35655 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
35656 (ifcombine_ifandif): Use it.
35658 2017-02-03 Martin Liska <mliska@suse.cz>
35660 * doc/invoke.texi: Document default value for
35661 use-after-scope-direct-emission-threshold.
35663 2017-02-03 Martin Liska <mliska@suse.cz>
35665 PR tree-optimization/79339
35666 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
35667 (format_floating): Likewise.
35669 2017-02-03 Martin Liska <mliska@suse.cz>
35672 * ipa-prop.c (ipa_node_params_t::insert): Remove current
35674 (ipa_node_params_t::remove): Likewise.
35675 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
35676 initialization from removed ipa_node_params_t::insert.
35677 (ipa_node_params::~ipa_node_params): Move from removed
35678 ipa_node_params_t::release.
35679 * symbol-summary.h (symbol_summary::m_released): New member.
35680 Do not release a summary twice. Do not allow to call finalizer
35681 for types of a summary that live in GGC memory.
35683 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35685 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
35688 2017-02-02 Martin Sebor <msebor@redhat.com>
35690 PR middle-end/79275
35691 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
35692 (format_string): Tighten up the range of output for non-constant
35693 strings and correct the expected range for wide non-constant strings.
35695 2017-02-02 Martin Sebor <msebor@redhat.com>
35697 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
35699 PR middle-end/32003
35700 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
35702 (-fdump-tree-@var): Add to index and document how to come up
35703 with pass-specific option and dump file names.
35704 (-fdump-passes): Clarify where to look for output.
35706 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
35708 PR middle-end/77445
35709 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
35710 statistics of the analyzed path; allow threading for speed when
35711 any of BBs along the path are optimized for speed.
35713 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
35715 PR middle-end/78468
35716 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
35717 settings of the virtual registers.
35720 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
35722 * explow.c (get_dynamic_stack_size): Take known alignment of stack
35723 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
35726 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35728 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
35729 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
35731 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35733 * config/s390/s390.md: Add missing comments with the expanded
35735 * config/s390/vector.md: Likewise.
35736 * config/s390/vx-builtins.md: Likewise.
35738 2017-02-02 Jakub Jelinek <jakub@redhat.com>
35741 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
35742 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
35743 conditions on a single line.
35745 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35747 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35748 __S390_VX__ to __VX__.
35750 2017-02-01 Andrew Pinski <apinski@cavium.com>
35752 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
35753 stmt_info to record_stmt_cost.
35754 (vect_get_known_peeling_cost): Pass stmt_info if known to
35756 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
35757 cpu_vector_cost field into
35758 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
35759 field into vec_int_stmt_cost and vec_fp_stmt_cost.
35760 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
35761 splitting of scalar_stmt_cost and vec_stmt_cost.
35762 (thunderx_vector_cost): Likewise.
35763 (cortexa57_vector_cost): LIkewise.
35764 (exynosm1_vector_cost): Likewise.
35765 (xgene1_vector_cost): Likewise.
35766 (thunderx2t99_vector_cost): Improve after the splitting of the two
35768 (aarch64_builtin_vectorization_cost): Update for the splitting of
35769 scalar_stmt_cost and vec_stmt_cost.
35771 2017-02-01 Torvald Riegel <triegel@redhat.com>
35772 Richard Henderson <rth@redhat.com>
35774 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
35775 conditional on existance of a fast atomic load.
35776 * optabs-query.c (can_atomic_load_p): New function.
35777 * optabs-query.h (can_atomic_load_p): Declare it.
35778 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
35779 no fast atomic load is available for the particular size of access.
35780 (expand_atomic_compare_and_swap): Likewise.
35781 (expand_atomic_load): Likewise.
35782 (expand_atomic_store): Likewise.
35783 (expand_atomic_fetch_op): Likewise.
35784 * testsuite/lib/target-supports.exp
35785 (check_effective_target_sync_int_128): Remove x86 because it provides
35786 no fast atomic load.
35787 (check_effective_target_sync_int_128_runtime): Likewise.
35789 2017-02-01 Richard Biener <rguenther@suse.de>
35791 * graphite.c: Include tree-vectorizer.h for find_loop_location.
35792 (graphite_transform_loops): Provide opt-info for optimized nests.
35793 * tree-parloop.c (parallelize_loops): Provide opt-info for
35794 parallelized loops.
35796 2017-02-01 Richard Biener <rguenther@suse.de>
35798 PR middle-end/79315
35799 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
35800 was not set before.
35802 2017-02-01 Richard Biener <rguenther@suse.de>
35804 PR tree-optimization/71824
35805 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35806 Verify the loops are valid in the merged SESE region.
35807 (scop_detection::can_represent_loop_1): Check analyzing the
35808 evolution of the number of iterations in the region succeeds.
35810 2017-01-31 Ian Lance Taylor <iant@golang.org>
35812 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
35813 REG_ARGS_SIZE note to 32-bit push insns and call insn.
35815 2017-01-31 David Malcolm <dmalcolm@redhat.com>
35817 PR preprocessor/79210
35818 * input.c (get_substring_ranges_for_loc): Replace line_width
35819 assertion with error-handling.
35821 2017-01-31 Richard Biener <rguenther@suse.de>
35823 PR tree-optimization/77318
35824 * graphite-sese-to-poly.c (extract_affine): Fix assert.
35825 (create_pw_aff_from_tree): Take loop parameter.
35826 (add_condition_to_pbb): Pass loop of the condition to
35827 create_pw_aff_from_tree.
35829 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35831 * config/s390/s390.c (s390_asan_shadow_offset): New function.
35832 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
35834 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
35838 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
35840 (convert_int_to_float128): Likewise.
35841 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
35842 (convert_int_to_float128): Likewise.
35843 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
35844 (UNSPEC_IEEE128_CONVERT): Likewise.
35845 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
35846 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
35847 Use local variables for IBM extended format.
35848 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
35849 (fix_trunc<mode>si2_fprs): Likewise.
35850 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
35851 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
35852 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
35853 to know that we can now have integers of all sizes in vector
35855 (fix<uns>_<mode>di2_hw): Likewise.
35856 (float<uns>_<mode>si2_hw): Likewise.
35857 (fix_<mode>si2_hw): Likewise.
35858 (fixuns_<mode>si2_hw): Likewise.
35859 (float<uns>_<mode>di2_hw): Likewise.
35860 (float_<mode>di2_hw): Likewise.
35861 (float_<mode>si2_hw): Likewise.
35862 (floatuns_<mode>di2_hw): Likewise.
35863 (floatuns_<mode>si2_hw): Likewise.
35864 (xscvqp<su>wz_<mode>): Delete, no longer used.
35865 (xscvqp<su>dz_<mode>): Likewise.
35866 (xscv<su>dqp_<mode>): Likewise.
35867 (ieee128_mfvsrd_64bit): Likewise.
35868 (ieee128_mfvsrd_32bit): Likewise.
35869 (ieee128_mfvsrwz): Likewise.
35870 (ieee128_mtvsrw): Likewise.
35871 (ieee128_mtvsrd_64bit): Likewise.
35872 (ieee128_mtvsrd_32bit): Likewise.
35874 2017-01-31 Martin Liska <mliska@suse.cz>
35877 * ipa-prop.c (ipa_free_all_node_params): Call release method
35878 instead of ~sumbol_summary to not to trigger double times
35881 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
35883 PR tree-optimization/71691
35884 * bitmap.h (class auto_bitmap): New.
35885 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
35886 is_maybe_undefined instead of ssa_undefined_value_p.
35888 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35890 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35891 __S390_ARCH_LEVEL__ to __ARCH__.
35893 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35895 PR tree-optimization/79267
35896 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
35897 if should_remove_lhs_p is true.
35899 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
35902 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
35903 (add_alignment_attribute): New.
35904 (base_type_die): Add alignment attribute.
35905 (subrange_type_die): Likewise.
35906 (modified_type_die): Likewise.
35907 (gen_array_type_die): Likewise.
35908 (gen_descr_array_type_die: Likewise.
35909 (gen_enumeration_type_die): Likewise.
35910 (gen_subprogram_die): Likewise.
35911 (gen_variable_die): Likewise.
35912 (gen_field_die): Likewise.
35913 (gen_ptr_to_mbr_type_die): Likewise.
35914 (gen_struct_or_union_type_die): Likewise.
35915 (gen_subroutine_type_die): Likewise.
35916 (gen_typedef_die): Likewise.
35917 (base_type_cmp): Compare alignment attribute.
35919 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35922 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
35923 (setb_unsigned) New pattern for setb with CCUNS.
35924 * config/rs6000/rs6000.c (expand_block_compare): Use a different
35925 subfc./subfe sequence to avoid overflow problems. Generate a
35926 shorter sequence with cmpld/setb for power9.
35927 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
35928 for generating subfc. instruction.
35929 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
35930 now uses this instruction.
35932 2017-01-30 Ian Lance Taylor <iant@google.com>
35935 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
35936 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
35938 2017-01-30 Martin Sebor <msebor@redhat.com>
35940 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
35941 Move constant to the right of a relational operator.
35942 (get_mpfr_format_length, format_character, format_string): Ditto.
35943 (should_warn_p, maybe_warn): Same.
35945 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
35947 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
35950 * asan.c (get_translation_unit_decl): Remove function.
35951 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
35953 2017-01-30 Martin Liska <mliska@suse.cz>
35955 PR gcov-profile/79259
35956 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
35957 -fprofile-generate.
35959 2017-01-30 Martin Liska <mliska@suse.cz>
35962 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
35963 Initialize variables with NULL value.
35965 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
35968 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
35970 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
35972 2017-01-30 Richard Biener <rguenther@suse.de>
35974 PR tree-optimization/79276
35975 * tree-vrp.c (process_assert_insertions): Properly adjust common
35976 when removing a duplicate.
35978 2017-01-30 Richard Biener <rguenther@suse.de>
35980 PR tree-optimization/79256
35981 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35982 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35984 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
35986 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
35989 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
35990 ("*r<noxa>sbg_<mode>_sll_bitmask")
35991 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
35992 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
35993 Use contiguous_bitmask_nowrap_operand.
35995 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35998 * config/rs6000/altivec.h (vec_xl): Revise #define.
35999 (vec_xst): Likewise.
36001 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
36003 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
36005 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
36007 PR rtl-optimization/79194
36008 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
36009 traps before call to bypass_conditional_jumps.
36011 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
36013 PR tree-optimization/71374
36014 * lra-constraints.c (check_conflict_input_operands): New.
36015 (match_reload): Use it.
36017 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
36020 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
36021 account to calculate conflict_set.
36023 2017-01-27 Bin Cheng <bin.cheng@arm.com>
36025 PR rtl-optimization/78559
36026 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
36027 other_insn in combine.
36029 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
36031 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
36032 uint16_type_node for BT_UINT16.
36034 2017-01-27 David Malcolm <dmalcolm@redhat.com>
36036 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
36037 "RTL Tests" to menu.
36038 (GIMPLE Tests): New node.
36039 (RTL Tests): New node.
36041 2017-01-27 Richard Biener <rguenther@suse.de>
36043 PR tree-optimization/79245
36044 * tree-loop-distribution.c (distribute_loop): Apply cost
36045 modeling also to detected patterns.
36047 2017-01-27 Richard Biener <rguenther@suse.de>
36049 PR tree-optimization/71433
36050 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
36051 (compare_assert_loc): New function.
36052 (process_assert_insertions): Sort and optimize assert locations
36053 to remove duplicates and push down identical assertions on
36054 edges to their destination block.
36056 2017-01-27 Richard Biener <rguenther@suse.de>
36058 PR tree-optimization/79244
36059 * tree-vrp.c (remove_range_assertions): Forcefully propagate
36060 out SSA names even if abnormal.
36062 2017-01-27 Jakub Jelinek <jakub@redhat.com>
36064 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
36065 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
36066 instead of MPFR_RNDN.
36068 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
36071 * arm.c (arm_option_override): Don't call build_target_option_node
36072 until after doing all option overrides.
36073 (arm_valid_target_attribute_tree): Likewise.
36075 2017-01-27 Martin Liska <mliska@suse.cz>
36077 * doc/invoke.texi (-fprofile-arcs): Document profiling support
36078 for {cd}tors and C++ {cd}tors.
36080 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
36082 * config/s390/s390.md ("*setmem_long_and")
36083 ("*setmem_long_and_31z"): Use zero_extend instead of and.
36085 2017-01-26 Martin Sebor <msebor@redhat.com>
36087 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
36090 2017-01-26 Martin Sebor <msebor@redhat.com>
36092 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
36093 HAVE_DFmode before using XFmode or DFmode.
36094 (parse_directive): Avoid using the z length modifier to avoid
36095 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
36097 PR middle-end/78703
36098 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
36099 to accept adjustment as an array.
36100 (get_int_range): New function.
36101 (struct directive): Make width and prec arrays.
36102 (directive::set_width, directive::set_precision): Call get_int_range.
36103 (format_integer, format_floating): Handle width and precision ranges.
36104 (format_string, parse_directive): Same.
36106 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36109 * dwarf2out.c (generate_skeleton_bottom_up): For children with
36110 comdat_type_p set, just clone them, but keep the children in the
36114 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
36115 which have direct callers with -fvar-tracking-assignments enabled
36117 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
36118 inside of type units.
36120 2017-01-26 Martin Sebor <msebor@redhat.com>
36122 PR middle-end/78703
36123 * gimple-ssa-sprintf.c (struct result_range): Add likely and
36125 (struct format_result): Replace number_chars, number_chars_min,
36126 and number_chars_max with a single member of struct result_range.
36128 (format_result::operator+=): Adjust.
36129 (struct fmtresult): Remove bounded. Handle likely and unlikely
36131 (fmtresult::adjust_for_width_or_precision): New function.
36132 (fmtresult:type_max_digits): New function.
36133 (bytes_remaining): Handle likely and unlikely counters.
36134 (min_bytes_remaining): Remove.
36135 (format_percent): Simplify.
36136 (format_integer, format_floating): Set likely and unlikely counters.
36137 (get_string_length, format_character, format_string): Same.
36138 (format_plain, should_warn_p): New function.
36139 (maybe_warn): Call should_warn_p. Update diagnostic messages
36140 and handle those for all directives, including plain strings.
36141 (format_directive): Handle likely and unlikely counters.
36142 Remove unnecessary quoting from diagnostics. Add an informational
36144 (add_bytes): Remove.
36145 (pass_sprintf_length::compute_format_length): Simplify.
36146 (try_substitute_return_value): Handle likely and unlikely counters.
36148 2017-01-26 Carl Love <cel@us.ibm.com>
36150 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
36151 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
36153 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
36156 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
36157 endianess for subregs into account.
36158 * lra-constraints.c (lra_constraints): Do risky transformations
36159 always on the first iteration.
36160 * lra-lives.c (check_pseudos_live_through_calls): Add arg
36161 last_call_used_reg_set.
36162 (process_bb_lives): Define and use last_call_used_reg_set.
36163 * lra.c (lra): Always continue after lra_constraints on the first
36166 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
36168 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
36170 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
36172 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36174 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
36175 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
36176 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
36177 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
36178 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
36179 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
36180 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
36181 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
36182 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
36184 2017-01-26 Marek Polacek <polacek@redhat.com>
36187 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
36188 for the third operand.
36190 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36192 PR middle-end/79236
36193 * omp-low.c (struct omp_context): Add simt_stmt field.
36194 (scan_omp_for): Return omp_context *.
36195 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
36196 context to the _simt_ SIMD stmt.
36197 (lower_omp_for): For combined SIMD with sibling _simt_
36198 SIMD, make sure to use the same decls in _looptemp_
36199 clauses as in the sibling.
36201 2017-01-26 David Sherwood <david.sherwood@arm.com>
36203 PR middle-end/79212
36204 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
36207 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36210 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
36211 emit fld b; fld a; if possible.
36213 * brig-builtins.def: Update copyright years.
36214 * config/arm/arm_acle_builtins.def: Update copyright years.
36216 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
36219 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
36220 constraint instead of o for the stxsd instruction.
36222 2017-01-25 Carl Love <cel@us.ibm.com>
36224 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
36225 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
36227 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
36229 * doc/invoke.texi (C++ Dialect Options): Fix typo.
36231 2017-01-25 Richard Biener <rguenther@suse.de>
36233 PR tree-optimization/69264
36234 * target.def (vector_alignment_reachable): Improve documentation.
36235 * doc/tm.texi: Regenerate.
36236 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
36238 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
36239 earlier changes with respect to TYPE_USER_ALIGN.
36240 (vector_alignment_reachable_p): Likewise. Improve dumping.
36242 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36245 * config/arm/arm.md (xordi3): Force constant operand into a register
36248 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36250 * doc/invoke.texi (-fstore-merging): Correct default optimization
36251 levels at which it is enabled.
36252 (-O): Move -fstore-merging from list to...
36255 2017-01-25 Richard Biener <rguenther@suse.de>
36258 * omp-expand.c: Include debug.h.
36259 (expand_omp_taskreg): Make sure to generate early debug before
36260 outlining anything from a function.
36261 (expand_omp_target): Likewise.
36262 (grid_expand_target_grid_body): Likewise.
36264 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
36267 * asan.c (get_translation_unit_decl): New function.
36268 (asan_add_global): Extract modules file name from globals
36269 TRANSLATION_UNIT_DECL name.
36271 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
36274 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
36275 for long calls with APCS frame and VFP.
36277 2017-01-24 David Malcolm <dmalcolm@redhat.com>
36279 * cfg.c (original_copy_tables_initialized_p): New function.
36280 * cfg.h (original_copy_tables_initialized_p): New decl.
36281 * cfgrtl.c (relink_block_chain): Guard the call to
36282 free_original_copy_tables with a call to
36283 original_copy_tables_initialized_p.
36284 * cgraph.h (symtab_node::native_rtl_p): New decl.
36285 * cgraphunit.c (symtab_node::native_rtl_p): New function.
36286 (symtab_node::needed_p): Don't assert for early assembly output
36287 for __RTL functions.
36288 (cgraph_node::finalize_function): Set "force_output" for __RTL
36290 (cgraph_node::analyze): Bail out early for __RTL functions.
36291 (analyze_functions): Update assertion to support __RTL functions.
36292 (cgraph_node::expand): Bail out early for __RTL functions.
36293 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
36295 * function.h (struct function): Update comment for field
36297 * gimple-expr.c: Include "tree-pass.h".
36298 (gimple_has_body_p): Return false for __RTL functions.
36299 * Makefile.in (OBJS): Add run-rtl-passes.o.
36300 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
36302 (gcc::pass_manager::get_clean_slate): New accessor.
36303 * passes.c: Include "insn-addr.h".
36304 (should_skip_pass_p): Add logging. Update logic for running
36305 "expand" to be compatible with both __GIMPLE and __RTL. Guard
36306 property-provider override so it is only done for gimple passes.
36308 (skip_pass): New function.
36309 (execute_one_pass): Call skip_pass when skipping passes.
36310 * read-md.c (md_reader::read_char): Support filtering
36311 the input to a subset of line numbers.
36312 (md_reader::md_reader): Initialize fields
36313 m_first_line and m_last_line.
36314 (md_reader::read_file_fragment): New function.
36315 * read-md.h (md_reader::read_file_fragment): New decl.
36316 (md_reader::m_first_line): New field.
36317 (md_reader::m_last_line): New field.
36318 * read-rtl-function.c (function_reader::create_function): Only
36319 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
36320 curr_properties. Set DECL_INITIAL to a dummy block.
36321 (read_rtl_function_body_from_file_range): New function.
36322 * read-rtl-function.h (read_rtl_function_body_from_file_range):
36324 * run-rtl-passes.c: New file.
36325 * run-rtl-passes.h: New file.
36327 2017-01-24 Jeff Law <law@redhat.com>
36329 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
36332 2017-01-24 Bin Cheng <bin.cheng@arm.com>
36334 PR tree-optimization/79159
36335 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
36336 (record_nonwrapping_iv): Improve boundary using above function if no
36337 value range information.
36339 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
36340 Martin Jambor <mjambor@suse.cz>
36342 * brig-builtins.def: New file.
36343 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
36344 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
36345 (DEF_HSAIL_SAT_BUILTIN): Likewise.
36346 (DEF_HSAIL_INTR_BUILTIN): Likewise.
36347 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
36348 * builtin-types.def (BT_INT8): New.
36349 (BT_INT16): Likewise.
36350 (BT_UINT8): Likewise.
36351 (BT_UINT16): Likewise.
36352 (BT_FN_ULONG): Likewise.
36353 (BT_FN_UINT_INT): Likewise.
36354 (BT_FN_UINT_ULONG): Likewise.
36355 (BT_FN_UINT_LONG): Likewise.
36356 (BT_FN_UINT_PTR): Likewise.
36357 (BT_FN_ULONG_PTR): Likewise.
36358 (BT_FN_INT8_FLOAT): Likewise.
36359 (BT_FN_INT16_FLOAT): Likewise.
36360 (BT_FN_UINT32_FLOAT): Likewise.
36361 (BT_FN_UINT16_FLOAT): Likewise.
36362 (BT_FN_UINT8_FLOAT): Likewise.
36363 (BT_FN_UINT64_FLOAT): Likewise.
36364 (BT_FN_UINT16_UINT32): Likewise.
36365 (BT_FN_UINT32_UINT16): Likewise.
36366 (BT_FN_UINT16_UINT16_UINT16): Likewise.
36367 (BT_FN_INT_PTR_INT): Likewise.
36368 (BT_FN_UINT_PTR_UINT): Likewise.
36369 (BT_FN_LONG_PTR_LONG): Likewise.
36370 (BT_FN_ULONG_PTR_ULONG): Likewise.
36371 (BT_FN_VOID_UINT64_UINT64): Likewise.
36372 (BT_FN_UINT8_UINT8_UINT8): Likewise.
36373 (BT_FN_INT8_INT8_INT8): Likewise.
36374 (BT_FN_INT16_INT16_INT16): Likewise.
36375 (BT_FN_INT_INT_INT): Likewise.
36376 (BT_FN_UINT_FLOAT_UINT): Likewise.
36377 (BT_FN_FLOAT_UINT_UINT): Likewise.
36378 (BT_FN_ULONG_UINT_UINT): Likewise.
36379 (BT_FN_ULONG_UINT_PTR): Likewise.
36380 (BT_FN_ULONG_ULONG_ULONG): Likewise.
36381 (BT_FN_UINT_UINT_UINT): Likewise.
36382 (BT_FN_VOID_UINT_PTR): Likewise.
36383 (BT_FN_UINT_UINT_PTR: Likewise.
36384 (BT_FN_UINT32_UINT64_PTR): Likewise.
36385 (BT_FN_INT_INT_UINT_UINT): Likewise.
36386 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
36387 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
36388 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
36389 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
36390 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
36391 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
36392 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
36393 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
36394 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
36395 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
36396 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
36397 * doc/frontends.texi: List BRIG FE.
36398 * doc/install.texi (Testing): Add BRIG tesring requirements.
36399 * doc/invoke.texi (Overall Options): Mention BRIG.
36400 * doc/standards.texi (Standards): Doucment BRIG HSA version.
36402 2017-01-24 Richard Biener <rguenther@suse.de>
36404 PR translation/79208
36405 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
36407 2017-01-24 Martin Jambor <mjambor@suse.cz>
36410 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
36411 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
36412 and known_contexts.
36414 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
36416 PR middle-end/79123
36417 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
36418 casts from signed to unsigned really don't have a range.
36420 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
36422 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
36423 GMP_RNDx for compatiblity.
36425 2017-01-24 Martin Liska <mliska@suse.cz>
36428 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
36429 that would prevent us to call alloca with -1 as argument.
36431 2017-01-24 Jakub Jelinek <jakub@redhat.com>
36433 * dwarf2out.c (output_compilation_unit_header, output_file_names):
36434 Avoid -Wformat-security warning.
36436 2017-01-23 Andrew Pinski <apinski@cavium.com>
36438 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
36441 2017-01-23 Martin Sebor <msebor@redhat.com>
36443 PR middle-end/78703
36444 * gimple-ssa-sprintf.c (warn_level): New global.
36445 (format_integer): Use it here and throughout the rest of the file.
36446 Use the same switch to compute sign as base.
36447 (maybe_warn): New function.
36448 (format_directive): Factor out warnings into maybe_warn.
36449 Add debugging output. Use warn_level.
36450 (add_bytes): Use warn_level.
36451 (pass_sprintf_length::compute_format_length): Add debugging output.
36452 (try_substitute_return_value): Same.
36453 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
36455 PR middle-end/78703
36456 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
36457 (struct fmtresult, format_integer, format_floating): Adjust.
36458 (fmtresult::fmtresult): Set max correctly in two argument ctor.
36459 (get_string_length, format_string,format_directive): Same.
36460 (pass_sprintf_length::compute_format_length): Same.
36461 (try_substitute_return_value): Simplify slightly.
36463 PR middle-end/78703
36464 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
36465 (fmtresult::operator+=): Outlined.
36466 (struct fmtresult): Add ctors.
36467 (struct conversion_spec): Rename...
36468 (struct directive): ...to this. Add and remove data members.
36469 (directive::set_width, directive::set_precision): New functions.
36470 (format_percent): Use fmtresult ctor.
36471 (get_width_and_precision): Remove.
36472 (format_integer): Make naming changes. Avoid computing width and
36474 (format_floating): Same. Adjust indentation.
36475 (format_character, format_none): New functions.
36476 (format_string): Moved character handling to format_character.
36477 (format_directive): Remove arguments, change return type.
36478 (parse_directive): New function.
36479 (pass_sprintf_length::compute_format_length): Move directive
36480 parsing to parse_directive.
36482 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36484 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
36485 (assign_assembler_name_if_needed): ... this.
36486 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
36487 (assign_assembler_name_if_needed): ... this.
36488 (free_lang_data_in_cgraph): Adjust callers.
36489 * cgraphunit.c (cgraph_node::analyze): Likewise.
36490 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
36493 2017-01-23 Richard Biener <rguenther@suse.de>
36495 PR tree-optimization/79088
36496 PR tree-optimization/79188
36497 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
36498 resetting loop bounds after last path deletion. Reset loop
36499 bounds of the target loop, make code match the comments.
36500 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
36501 Make sure loops need no fixups.
36503 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
36505 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
36506 exponent support with double type for first argument.
36507 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
36508 type returned by __builtin_vec_extract_sig,
36509 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
36510 functions from "vector int" to "vector unsigned int" or from
36511 "vector long long int" to "vector unsigned long long int".
36512 Changed type returned by __builtin_vec_extract_exp,
36513 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
36514 functions from "vector int" to "vector unsigned int" or from
36515 "vector long long int" to "vector unsigned long long int".
36516 Changed return type of __builtin_vec_test_data_class,
36517 __builtin_vec_test_data_class_sp, and
36518 __builtin_vec_test_data_class_dp from "vector int" to
36519 "vector bool int" or from "vector long long int" to "vector bool
36520 long long int" and changed second argument type from "unsigned
36521 int" to "int". Added new overloaded function forms "vector float
36522 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
36523 "vector float __builtin_vec_insert_exp_sp (vector float, vector
36524 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
36525 double, vector unsigned long long int)" and "vector double
36526 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
36527 long int)". Changed return type of
36528 __builtin_scalar_test_data_class and
36529 __builtin_scalar_test_data_class_sp and
36530 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
36531 int" and changed second argument from "unsigned int" to "int".
36532 Changed type returned by __builtin_scalar_test_neg,
36533 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
36534 from "int" to "bool int". Added new overloaded function form
36535 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
36536 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
36537 exponent double-precision with floating point first argument.
36538 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
36539 documentation of scalar_test_data_class, scalar_test_neg,
36540 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
36541 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
36542 vec_test_data_class built-in functions to reflect refinements in
36543 their type signatures.
36545 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
36547 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
36549 (aarch64_elf_asm_destructor): Likewise.
36551 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
36553 PR rtl-optimization/78634
36554 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
36555 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
36556 * ifcvt.c (noce_try_cmove): Add missing cost check.
36558 PR rtl-optimization/71724
36559 * combine.c (if_then_else_cond): Look for situations where it is
36560 beneficial to undo the work of one of the recursive calls.
36562 2017-01-23 Bin Cheng <bin.cheng@arm.com>
36564 PR tree-optimization/70754
36565 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
36566 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
36567 combined stmt before it if not NULL.
36568 (combine_chains): Process refs reversely and compute dominance point
36571 2017-01-23 Martin Liska <mliska@suse.cz>
36573 PR tree-optimization/79196
36574 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
36575 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
36577 (strlen_optimize_stmt): Call the renamed function.
36579 2017-01-23 Michael Matz <matz@suse.de>
36581 PR tree-optimization/78384
36582 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
36584 2017-01-23 Richard Biener <rguenther@suse.de>
36586 PR tree-optimization/79186
36587 * tree-vrp.c (register_new_assert_for): Make sure we've seen
36588 both incoming edges before moving an assert.
36590 2017-01-23 Martin Jambor <mjambor@suse.cz>
36592 * ipa-prop.c (load_from_param_1): Removed.
36593 (load_from_unmodified_param): Bits from load_from_param_1 put back
36595 (load_from_param): Removed.
36596 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
36597 with stmt. Reverted back to use of load_from_unmodified_param.
36599 2017-01-23 Martin Jambor <mjambor@suse.cz>
36602 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
36603 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
36604 field a pointer to garbage collected vector, mark lattices and
36605 ipcp_orig_node with GTY((skip)).
36606 (ipa_get_param_count): Adjust to descriptors being a pointer.
36607 (ipa_get_param): Likewise.
36608 (ipa_get_type): Likewise.
36609 (ipa_get_param_move_cost): Likewise.
36610 (ipa_set_param_used): Likewise.
36611 (ipa_get_controlled_uses): Likewise.
36612 (ipa_set_controlled_uses): Likewise.
36613 (ipa_is_param_used): Likewise.
36614 (ipa_node_params_t): Move into garbage collector. New methods insert
36616 (ipa_node_params_sum): Annotate wth GTY(()).
36617 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
36619 (ipa_load_from_parm_agg): Adjust declaration.
36620 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
36621 * ipa-profile.c (ipa_profile): Likewise.
36622 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
36623 (ipa_populate_param_decls): Make descriptors parameter garbage
36625 (ipa_dump_param): Adjust to descriptors being a pointer.
36626 (ipa_alloc_node_params): Likewise.
36627 (ipa_initialize_node_params): Likewise.
36628 (load_from_param_1): Make descriptors parameter garbage collected.
36629 (load_from_unmodified_param): Likewise.
36630 (load_from_param): Likewise.
36631 (ipa_load_from_parm_agg): Likewise.
36632 (ipa_node_params::~ipa_node_params): Removed.
36633 (ipa_free_all_node_params): Remove call to delete operator.
36634 (ipa_node_params_t::insert): New.
36635 (ipa_node_params_t::remove): Likewise.
36636 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
36637 copy known_csts and known_contexts vectors.
36638 (ipa_read_node_info): Adjust to descriptors being a pointer.
36639 (ipcp_modif_dom_walker): Make m_descriptors field garbage
36641 (ipcp_transform_function): Make descriptors variable garbage
36644 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
36646 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
36647 * config/i386/avx512dqintrin.h: Ditto.
36648 * config/i386/avx512fintrin.h: Ditto.
36649 * config/i386/i386.c: Handle new builtins.
36650 * config/i386/i386-builtin.def: Add new builtins.
36651 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
36652 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
36654 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36655 Martin Liska <mliska@suse.cz>
36657 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
36658 * asan.c (asan_expand_poison_ifn): Support stores and use
36659 appropriate ASAN report function.
36660 * internal-fn.c (expand_ASAN_POISON_USE): New function.
36661 * internal-fn.def (ASAN_POISON_USE): Declare.
36662 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
36663 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
36664 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
36665 ASAN_POISON calls w/o LHS.
36666 * tree-ssa.c (execute_update_addresses_taken): Create clobber
36667 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
36668 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
36669 * gimplify.c (asan_poison_variables): Add attribute
36670 use_after_scope_memory to variables that really needs to live
36672 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
36673 having the attribute.
36675 2017-01-23 Martin Liska <mliska@suse.cz>
36677 * asan.c (create_asan_shadow_var): New function.
36678 (asan_expand_poison_ifn): Likewise.
36679 * asan.h (asan_expand_poison_ifn): New declaration.
36680 * internal-fn.c (expand_ASAN_POISON): Likewise.
36681 * internal-fn.def (ASAN_POISON): New builtin.
36682 * sanopt.c (pass_sanopt::execute): Expand
36683 asan_expand_poison_ifn.
36684 * tree-inline.c (copy_decl_for_dup_finish): Make function
36686 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
36687 * tree-ssa.c (is_asan_mark_p): New function.
36688 (execute_update_addresses_taken): Rewrite local variables
36689 (identified just by use-after-scope as addressable) into SSA.
36691 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
36693 * doc/install.texi (Specific): opensource.apple.com uses https
36694 now. Remove trailing slash.
36696 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
36698 * README.Portability: Remove note on an Irix compatibility issue.
36700 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
36702 * gcov.c (INCLUDE_ALGORITHM): Define.
36703 (INCLUDE_VECTOR): Define.
36704 No longer include <vector> and <algorithm> directly.
36706 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
36708 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
36710 * doc/invoke.texi (Code Gen Options): Ditto.
36712 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
36715 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
36717 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
36719 rtl-optimization/79125
36720 * cprop.c (local_cprop_pass): Handle cases where we make an
36721 unconditional trap.
36723 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
36727 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
36728 read from, for big endian.
36730 2017-01-20 Jiong Wang <jiong.wang@arm.com>
36732 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
36733 register pauth builtins for LP64 only.
36735 2017-01-20 Marek Polacek <polacek@redhat.com>
36738 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
36741 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
36743 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
36745 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
36746 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
36747 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
36749 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36752 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
36753 in big-endian mode when they are not a single duplicated value.
36755 2017-01-20 Richard Biener <rguenther@suse.de>
36757 * BASE-VER: Bump to 7.0.1.
36759 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
36761 * omp-low.c (omplow_simd_context): New struct. Use it...
36762 (lower_rec_simd_input_clauses): ...here and...
36763 (lower_rec_input_clauses): ...here to hold common data. Adjust all
36764 references to idx, lane, max_vf, is_simt.
36766 2017-01-20 Graham Markall <graham.markall@embecosm.com>
36768 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
36771 2017-01-20 Martin Jambor <mjambor@suse.cz>
36773 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
36774 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
36776 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
36777 (GTFILES): Rename hsa.c to hsa-common.c.
36778 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
36779 * hsa-dump.c: Likewise.
36780 * hsa-gen.c: Likewise.
36781 * hsa-regalloc.c: Likewise.
36782 * ipa-hsa.c: Likewise.
36783 * omp-expand.c: Likewise.
36784 * omp-low.c: Likewise.
36785 * toplev.c: Likewise.
36787 2017-01-20 Marek Polacek <polacek@redhat.com>
36790 * doc/invoke.texi: Document -Wduplicated-branches.
36791 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
36792 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
36793 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
36794 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
36795 return 0 only when not OEP_LEXICOGRAPHIC.
36796 (fold_build_cleanup_point_expr): Use the expression
36797 location when building CLEANUP_POINT_EXPR.
36798 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
36799 * tree.c (add_expr): Handle error_mark_node.
36801 2017-01-20 Martin Liska <mliska@suse.cz>
36804 * tree-profile.c (init_ic_make_global_vars): Do not call
36806 (gimple_init_gcov_profiler): Likewise.
36808 2017-01-20 Martin Liska <mliska@suse.cz>
36811 * cgraph.h (maybe_create_reference): Remove argument and
36813 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
36815 * ipa-cp.c (create_specialized_node): Likewise.
36816 * symtab.c (symtab_node::maybe_create_reference): Handle
36817 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
36819 2017-01-20 Martin Liska <mliska@suse.cz>
36821 * read-rtl-function.c (function_reader::create_function): Use
36822 build_decl instread of build_decl_stat.
36824 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
36826 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
36827 * config/i386/avx512dqintrin.h: Ditto.
36828 * config/i386/avx512fintrin.h: Ditto.
36829 * config/i386/i386-builtin-types.def: Add new types.
36830 * config/i386/i386.c: Handle new types.
36831 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
36832 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
36833 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
36834 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
36835 (__builtin_ia32_kshiftridi): New.
36836 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
36838 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
36842 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
36843 define to rs6000_init_stack_protect_guard.
36844 (rs6000_init_stack_protect_guard): New function.
36846 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36847 Yunqiang Su <yunqiang.su@imgtec.com>
36849 * config.gcc (supported_defaults): Add madd4.
36850 (with_madd4): Add validation.
36851 (all_defaults): Add madd4.
36852 * config/mips/mips.opt (mmadd4): New option.
36853 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
36855 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
36857 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
36858 (ISA_HAS_FUSED_MADD4): Likewise.
36859 * doc/invoke.texi (-mmadd4): Document the new option.
36860 * doc/install.texi (--with-madd4): Document the new option.
36862 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36864 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
36865 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
36866 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
36867 (aarch64_init_pauth_hint_builtins): New.
36868 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
36869 (aarch64_expand_builtin): Expand new builtins.
36871 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36873 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
36874 * combine-stack-adj.c (no_unhandled_cfa): Handle
36875 REG_CFA_TOGGLE_RA_MANGLE.
36876 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
36877 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
36878 info for return address signing.
36879 (aarch64_expand_epilogue): Likewise.
36881 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36883 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
36884 * config/aarch64/aarch64-protos.h
36885 (aarch64_return_address_signing_enabled): New declaration.
36886 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
36888 (aarch64_expand_prologue): Sign return address before it's pushed onto
36890 (aarch64_expand_epilogue): Authenticate return address fetched from
36892 (aarch64_override_options): Sanity check for ILP32 and ISA level.
36893 (aarch64_attributes): New function attributes for "sign-return-address".
36894 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
36895 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
36896 ("*do_return"): Generate combined instructions according to key index.
36897 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
36898 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
36900 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
36901 * config/aarch64/aarch64.opt (msign-return-address=): New.
36902 * doc/extend.texi (AArch64 Function Attributes): Documents
36903 "sign-return-address=".
36904 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
36906 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36908 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
36909 overall option summary.
36911 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36913 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
36914 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
36915 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
36916 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
36918 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
36920 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
36921 -mpower9-minmax by default for -mcpu=power9.
36922 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
36923 128-bit floating point.
36925 2017-01-20 Alan Modra <amodra@gmail.com>
36927 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
36928 optimizing for size.
36930 2017-01-20 Alan Modra <amodra@gmail.com>
36933 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
36934 for strcmp and strncmp from corresponding builtin decl.
36936 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36938 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
36939 instead of i386/rtems-64.h.
36940 * config/i386/rtems-64.h: Remove.
36942 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36946 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
36948 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
36950 2017-01-19 Tamar Christina <tamar.christina@arm.com>
36952 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
36953 Change int to HOST_WIDE_INT.
36954 * config/aarch64/aarch64-protos.h
36955 (aarch64_simd_gen_const_vector_dup): Likewise.
36956 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
36958 2017-01-19 David Malcolm <dmalcolm@redhat.com>
36960 * langhooks-def.h (lhd_type_for_size): New decl.
36961 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
36962 * langhooks.c (lhd_type_for_size): New function, taken from
36965 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
36967 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
36968 define_bypass for CR latency.
36969 (power9-cracked-alu): Update bypass latency and remove power9-branch.
36970 (power9-alu2): Add define_bypass for CR latency.
36972 (power9-mul): Update insn latency.
36973 (power9-mul-compare): Update insn latency, bypass latency and remove
36976 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36978 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
36980 * config/aarch64/aarch64.md
36981 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
36982 aarch64_nopcrelative_literal_loads.
36983 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
36985 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
36987 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
36988 TARGET_LOONGSON_3A.
36989 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
36991 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
36994 * config.gcc (supported_defaults): Add lxc1-sxc1.
36995 (with_lxc1_sxc1): Add validation.
36996 (all_defaults): Add lxc1-sxc1.
36997 * config/mips/mips.opt (mlxc1-sxc1): New option.
36998 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
37000 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
37001 __mips_no_lxc1_sxc1.
37002 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
37003 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
37004 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
37006 2017-01-19 Richard Biener <rguenther@suse.de>
37008 PR tree-optimization/72488
37009 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
37010 sure to restore SSA info.
37011 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
37013 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
37015 PR rtl-optimization/79121
37016 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
37017 of the inner type when shifting an extended value.
37019 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
37022 * symtab.c (symtab_node::equal_address_to): Fix comparing of
37023 interposable aliases.
37025 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
37028 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
37029 Use the evmergelohi instruction.
37030 (mov_si<mode>_e500_subreg4_2_le): Likewise.
37031 (mov_sitf_e500_subreg8_2_be): Likewise.
37032 (mov_sitf_e500_subreg12_2_le): Likewise.
37033 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
37034 (mov_si<mode>_e500_subreg4_2_be): Likewise.
37035 (mov_sitf_e500_subreg8_2_le): Likewise.
37036 (mov_sitf_e500_subreg12_2_be): Likewise.
37038 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37040 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
37041 attribute from vecsimple to vecperm.
37042 (altivec_vbpermq2): Likewise.
37044 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37047 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
37049 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37050 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
37051 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
37052 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
37053 case where N arg is SIZE_MAX.
37054 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
37055 (cmpstrsi): Add pattern.
37057 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
37059 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
37060 __builtin_vec_revb builtins.
37061 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
37062 built-in functions to support generation of the ISA 3.0 XXBR<x>
37063 vector byte reverse instructions.
37064 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
37065 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
37066 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
37067 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
37068 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
37069 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
37070 (P9V_BUILTIN_VEC_REVB): Likewise.
37071 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
37072 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
37073 (p9_xxbrq_v16qi): Likewise.
37074 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
37075 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
37076 (p9_xxbrh_v8hi): Likewise.
37077 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
37078 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
37079 vec_revb built-in functions.
37081 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
37083 PR rtl-optimization/78952
37084 * config/i386/i386.md (any_extract): New code iterator.
37085 (*insvqi_2): Use any_extract for source operand.
37086 (*insvqi_3): Use any_shiftrt for source operand.
37088 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
37090 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
37092 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
37094 2017-01-18 Matthias Klose <doko@ubuntu.com>
37096 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
37098 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37100 * config/rs6000/altivec.h (vec_bperm): Change #define.
37101 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
37102 (altivec_vbpermq2): New define_insn.
37103 (altivec_vbpermd): Likewise.
37104 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
37105 function interface.
37106 (VBPERMD): Likewise.
37107 (VBPERM): New polymorphic function interface.
37108 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
37109 Add entries for P9V_BUILTIN_VEC_VBPERM.
37110 * doc/extend.texi: Add interfaces for vec_bperm.
37112 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37114 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
37115 first letter of error messages.
37116 (s390_resolve_overloaded_builtin): Likewise.
37117 * config/s390/s390.c (s390_expand_builtin): Likewise.
37118 (s390_invalid_arg_for_unprototyped_fn): Likewise.
37119 (s390_valid_target_attribute_inner_p): Likewise.
37120 * config/s390/s390.md ("tabort"): Likewise.
37122 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
37124 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
37125 (ISA_AVOID_DIV_HILO): New macro.
37126 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
37127 (ISA_HAS_DDIV): Likewise.
37129 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37131 * doc/invoke.texi (fabi-version): Correct number of occurrences.
37133 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37135 * doc/invoke.texi (fabi-version): Spelling fix.
37137 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37140 * doc/invoke.texi (fabi-version): Mention mangling fix for
37143 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37146 * doc/invoke.texi (fabi-version): Document discriminator mangling.
37148 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
37151 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
37152 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
37154 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
37156 (stack_protect_test): Ditto.
37157 * config/rs6000/rs6000.opt (mstack-protector-guard=,
37158 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
37160 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
37161 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
37162 -mstack-protector-guard-offset=.
37163 (RS/6000 and PowerPC Options): Ditto.
37165 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37167 * config/i386/i386.h (MASK_CLASS_P): New define.
37168 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
37169 there are no registers from different register sets also when
37170 mask registers are used. Update function comment.
37171 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
37172 to (*k/*r) and (*k/*km) alternatives.
37174 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
37176 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
37177 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
37178 (EH_RETURN_HANDLER_RTX): New define.
37179 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
37180 Force frame pointer in EH return functions.
37181 (aarch64_expand_epilogue): Add barrier for eh_return.
37182 (aarch64_final_eh_return_addr): Remove.
37183 (aarch64_eh_return_handler_rtx): New function.
37184 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
37186 (aarch64_eh_return_handler_rtx): New prototype.
37188 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37190 * config/rs6000/altivec.h (vec_rlmi): New #define.
37191 (vec_vrlnm): Likewise.
37192 (vec_rlnm): Likewise.
37193 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
37194 (UNSPEC_VRLNM): Likewise.
37195 (VIlong): New mode iterator.
37196 (altivec_vrl<VI_char>mi): New define_insn.
37197 (altivec_vrl<VI_char>nm): Likewise.
37198 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
37200 (VRLDNM): Likewise.
37201 (RLNM): New polymorphic function entry.
37202 (VRLWMI): New monomorphic function entry.
37203 (VRLDMI): Likewise.
37204 (RLMI): New polymorphic function entry.
37205 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
37206 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
37207 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
37210 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37213 * dwarf2out.c (field_byte_offset): Restore the
37214 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
37215 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
37216 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
37219 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37222 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
37224 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37227 * dwarf2out.c (add_data_member_location_attribute): For constant
37228 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
37229 instead of DW_AT_data_member_location, DW_AT_bit_offset and
37230 DW_AT_byte_size attributes.
37232 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37234 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
37235 after forcing to constant memory when the code model is medium.
37237 2017-01-17 Julia Koval <julia.koval@intel.com>
37240 * config/i386/avx512fintrin.h
37241 (_mm512_i32gather_ps): Change __addr type to void const*.
37242 (_mm512_mask_i32gather_ps): Ditto.
37243 (_mm512_i32gather_pd): Ditto.
37244 (_mm512_mask_i32gather_pd): Ditto.
37245 (_mm512_i64gather_ps): Ditto.
37246 (_mm512_mask_i64gather_ps): Ditto.
37247 (_mm512_i64gather_pd): Ditto.
37248 (_mm512_mask_i64gather_pd): Ditto.
37249 (_mm512_i32gather_epi32): Ditto.
37250 (_mm512_mask_i32gather_epi32): Ditto.
37251 (_mm512_i32gather_epi64): Ditto.
37252 (_mm512_mask_i32gather_epi64): Ditto.
37253 (_mm512_i64gather_epi32): Ditto.
37254 (_mm512_mask_i64gather_epi32): Ditto.
37255 (_mm512_i64gather_epi64): Ditto.
37256 (_mm512_mask_i64gather_epi64): Ditto.
37257 (_mm512_i32scatter_ps): Change __addr type to void*.
37258 (_mm512_mask_i32scatter_ps): Ditto.
37259 (_mm512_i32scatter_pd): Ditto.
37260 (_mm512_mask_i32scatter_pd): Ditto.
37261 (_mm512_i64scatter_ps): Ditto.
37262 (_mm512_mask_i64scatter_ps): Ditto.
37263 (_mm512_i64scatter_pd): Ditto.
37264 (_mm512_mask_i64scatter_pd): Ditto.
37265 (_mm512_i32scatter_epi32): Ditto.
37266 (_mm512_mask_i32scatter_epi32): Ditto.
37267 (_mm512_i32scatter_epi64): Ditto.
37268 (_mm512_mask_i32scatter_epi64): Ditto.
37269 (_mm512_i64scatter_epi32): Ditto.
37270 (_mm512_mask_i64scatter_epi32): Ditto.
37271 (_mm512_i64scatter_epi64): Ditto.
37272 (_mm512_mask_i64scatter_epi64): Ditto.
37273 * config/i386/avx512pfintrin.h
37274 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
37275 (_mm512_mask_prefetch_i32gather_ps): Ditto.
37276 (_mm512_mask_prefetch_i64gather_pd): Ditto.
37277 (_mm512_mask_prefetch_i64gather_ps): Ditto.
37278 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
37279 (_mm512_prefetch_i32scatter_ps): Ditto.
37280 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
37281 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
37282 (_mm512_prefetch_i64scatter_pd): Ditto.
37283 (_mm512_prefetch_i64scatter_ps): Ditto.
37284 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
37285 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
37286 * config/i386/avx512vlintrin.h
37287 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
37288 (_mm_mmask_i32gather_ps): Ditto.
37289 (_mm256_mmask_i32gather_pd): Ditto.
37290 (_mm_mmask_i32gather_pd): Ditto.
37291 (_mm256_mmask_i64gather_ps): Ditto.
37292 (_mm_mmask_i64gather_ps): Ditto.
37293 (_mm256_mmask_i64gather_pd): Ditto.
37294 (_mm_mmask_i64gather_pd): Ditto.
37295 (_mm256_mmask_i32gather_epi32): Ditto.
37296 (_mm_mmask_i32gather_epi32): Ditto.
37297 (_mm256_mmask_i32gather_epi64): Ditto.
37298 (_mm_mmask_i32gather_epi64): Ditto.
37299 (_mm256_mmask_i64gather_epi32): Ditto.
37300 (_mm_mmask_i64gather_epi32): Ditto.
37301 (_mm256_mmask_i64gather_epi64): Ditto.
37302 (_mm_mmask_i64gather_epi64): Ditto.
37303 (_mm256_i32scatter_ps): Change __addr type to void*.
37304 (_mm256_mask_i32scatter_ps): Ditto.
37305 (_mm_i32scatter_ps): Ditto.
37306 (_mm_mask_i32scatter_ps): Ditto.
37307 (_mm256_i32scatter_pd): Ditto.
37308 (_mm256_mask_i32scatter_pd): Ditto.
37309 (_mm_i32scatter_pd): Ditto.
37310 (_mm_mask_i32scatter_pd): Ditto.
37311 (_mm256_i64scatter_ps): Ditto.
37312 (_mm256_mask_i64scatter_ps): Ditto.
37313 (_mm_i64scatter_ps): Ditto.
37314 (_mm_mask_i64scatter_ps): Ditto.
37315 (_mm256_i64scatter_pd): Ditto.
37316 (_mm256_mask_i64scatter_pd): Ditto.
37317 (_mm_i64scatter_pd): Ditto.
37318 (_mm_mask_i64scatter_pd): Ditto.
37319 (_mm256_i32scatter_epi32): Ditto.
37320 (_mm256_mask_i32scatter_epi32): Ditto.
37321 (_mm_i32scatter_epi32): Ditto.
37322 (_mm_mask_i32scatter_epi32): Ditto.
37323 (_mm256_i32scatter_epi64): Ditto.
37324 (_mm256_mask_i32scatter_epi64): Ditto.
37325 (_mm_i32scatter_epi64): Ditto.
37326 (_mm_mask_i32scatter_epi64): Ditto.
37327 (_mm256_i64scatter_epi32): Ditto.
37328 (_mm256_mask_i64scatter_epi32): Ditto.
37329 (_mm_i64scatter_epi32): Ditto.
37330 (_mm_mask_i64scatter_epi32): Ditto.
37331 (_mm256_i64scatter_epi64): Ditto.
37332 (_mm256_mask_i64scatter_epi64): Ditto.
37333 (_mm_i64scatter_epi64): Ditto.
37334 (_mm_mask_i64scatter_epi64): Ditto.
37335 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
37336 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
37337 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
37338 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
37339 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
37340 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
37341 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
37342 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
37343 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
37344 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
37345 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
37346 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
37347 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
37348 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
37349 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
37350 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
37351 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
37352 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
37353 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
37354 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
37355 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
37356 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
37357 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
37358 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
37359 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
37360 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
37361 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
37362 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
37363 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
37364 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
37365 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
37366 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
37367 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
37368 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
37369 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
37370 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
37371 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
37372 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
37373 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
37374 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
37375 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
37376 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
37377 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
37378 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
37379 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
37380 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
37381 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
37382 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
37383 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
37384 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
37385 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
37386 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
37387 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
37388 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
37389 definitions accordingly.
37391 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
37392 Kuan-Lin Chen <kuanlinchentw@gmail.com>
37395 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
37398 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
37401 * ira-conflicts.c (ira_build_conflicts): Update total conflict
37402 hard regs for inner regno.
37404 2017-01-17 Martin Liska <mliska@suse.cz>
37407 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
37408 assumption and add comment.
37410 2017-01-17 Nathan Sidwell <nathan@acm.org>
37412 * ipa-visibility.c (localize_node): New function, broken out of ...
37413 (function_and_variable_visibility): ... here. Call it.
37415 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
37417 PR middle-end/77445
37418 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
37419 correctly set frequency of oudgoing edge.
37420 (duplicate_thread_path): Fix profile updating.
37422 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37425 * configure.ac: Add GCC_BASE_VER.
37426 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
37427 version from BASE-VER file.
37428 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
37429 (gcc.o): Depend on $(BASEVER).
37430 * common.opt (dumpfullversion): New option.
37431 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
37432 * doc/invoke.texi: Document -dumpfullversion.
37433 * doc/install.texi: Document --with-gcc-major-version-only.
37434 * configure: Regenerated.
37436 2017-01-17 Richard Biener <rguenther@suse.de>
37438 PR tree-optimization/71433
37439 * tree-vrp.c (register_new_assert_for): Merge same asserts
37440 on all incoming edges.
37441 (process_assert_insertions_for): Handle insertions at the
37444 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
37446 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
37447 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
37449 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
37452 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
37455 2017-01-17 Alan Modra <amodra@gmail.com>
37458 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
37459 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
37460 symbolic stack limit when pic.
37462 2017-01-16 Martin Sebor <msebor@redhat.com>
37464 PR tree-optimization/78608
37465 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
37467 2017-01-16 Jeff Law <law@redhat.com>
37470 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
37471 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
37472 for several include directories that may be relative to sysroot.
37473 * config/i386/x-mingw32 (gplus_includedir): Define.
37474 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
37475 (native_system_includedir): Likewise.
37476 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
37477 override if TARGET_SYSTEM_ROOT is defined.
37478 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
37480 PR tree-optimization/79090
37481 PR tree-optimization/33562
37482 PR tree-optimization/61912
37483 PR tree-optimization/77485
37484 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
37485 and computed trims into the dump file.
37487 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37489 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
37491 2017-01-16 Jakub Jelinek <jakub@redhat.com>
37494 * gimplify.c (gimplify_init_constructor): If want_value and
37495 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
37499 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
37500 sequence. Formatting fixes.
37501 (doloop_optimize): Formatting fixes.
37504 * gcc.c (debug_level_greater_than_spec_func): New function.
37505 (static_spec_functions): Add debug-level-gt spec function.
37506 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
37508 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37509 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37510 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
37511 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
37512 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
37513 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
37515 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
37517 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
37518 QImode fixups to general and mask registers only.
37520 2017-01-16 Carl Love <cel@us.ibm.com>
37522 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
37523 for built-in functions
37524 vector signed char vec_nabs (vector signed char)
37525 vector signed short vec_nabs (vector signed short)
37526 vector signed int vec_nabs (vector signed int)
37527 vector signed long long vec_nabs (vector signed long long)
37528 vector float vec_nabs (vector float)
37529 vector double vec_nabs (vector double)
37530 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
37532 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
37533 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
37534 * doc/extend.texi: Update the documentation file for the new built-in
37537 2017-01-16 Martin Sebor <msebor@redhat.com>
37539 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
37542 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37544 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
37545 UNSPEC_VSX__XXSPLTD to require special splat handling.
37547 2017-01-16 David Malcolm <dmalcolm@redhat.com>
37550 * system.h: Poison strndup.
37552 2017-01-16 Alan Modra <amodra@gmail.com>
37555 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
37558 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
37560 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
37562 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
37564 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
37565 call recog here. Assert that INSN_CODE (insn) is non-negative.
37567 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37570 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
37572 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
37573 in the currently scheduled RTL fragment.
37575 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37577 PR rtl-optimization/78751
37578 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
37581 2017-01-14 Jeff Law <law@redhat.com>
37583 PR tree-optimization/79090
37584 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
37585 variable length stores.
37586 (compute_trims): Delete dead assignment to *trim_tail.
37587 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
37590 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
37592 PR rtl-optimization/78626
37593 PR rtl-optimization/78727
37594 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
37595 of a block, and split such blocks after everything else is finished.
37597 2017-01-14 Alan Modra <amodra@gmail.com>
37600 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
37601 target legitimate_combined_insn.
37602 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
37603 (rs6000_legitimate_combined_insn): New function.
37604 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
37606 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
37607 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
37608 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
37610 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
37612 * doc/frontends.texi (G++ and GCC): Remove references to Java.
37614 2017-01-13 Jeff Law <law@redhat.com>
37616 PR tree-optimization/33562
37617 PR tree-optimization/61912
37618 PR tree-optimization/77485
37619 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
37621 (delete_dead_assignment): Likewise.
37622 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
37623 statement to delete_dead_call and delete_dead_assignment.
37625 2017-01-13 David Malcolm <dmalcolm@redhat.com>
37628 * substring-locations.c (format_warning_va): Strengthen case 1 so
37629 that both endpoints of the substring must be within the format
37630 range for just the substring to be printed.
37632 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
37634 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
37635 * config/i386/i386.c (ix86_target_string): Add missing options
37636 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
37637 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
37638 flags_other and ix86_target_other to flags2_other. Display unknown
37640 (ix86_valid_target_attribute_inner_p): Add missing options and
37641 reorder options by implied ISAs, as in ix86_target_string.
37643 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
37645 * hash-table.h (hash_table::too_empty_p): New function.
37646 (hash_table::expand): Use it.
37647 (hash_table::traverse): Likewise.
37648 (hash_table::empty_slot): Use sizeof (value_type) instead of
37649 sizeof (PTR) to convert bytes to elements. Shrink the table
37650 if the current size is excessive for the current number of
37653 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
37655 * ira-costs.c (record_reg_classes): Break from the inner loop
37656 early once alt_fail is known to be true. Update outer loop
37657 handling accordingly.
37659 2017-01-13 Jeff Law <law@redhat.com>
37661 * tree-ssa-dse.c (decrement_count): New function.
37662 (increment_start_addr, maybe_trim_memstar_call): Likewise.
37663 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
37664 when we know the partially dead statement is a mem* function.
37666 PR tree-optimization/61912
37667 PR tree-optimization/77485
37668 * tree-ssa-dse.c: Include expr.h.
37669 (maybe_trim_constructor_store): New function.
37670 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
37672 PR tree-optimization/33562
37673 PR tree-optimization/61912
37674 PR tree-optimization/77485
37675 * doc/invoke.texi: Document new dse-max-object-size param.
37676 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
37677 * tree-ssa-dse.c: Include params.h.
37678 (dse_store_status): New enum.
37679 (initialize_ao_ref_for_dse): New, partially extracted from
37681 (valid_ao_ref_for_dse, normalize_ref): New.
37682 (setup_live_bytes_from_ref, compute_trims): Likewise.
37683 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
37684 (maybe_trim_partially_dead_store): Likewise.
37685 (maybe_trim_complex_store): Likewise.
37686 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
37687 Track what bytes live from the original store. Return tri-state
37688 for dead, partially dead or live.
37689 (dse_dom_walker): Add constructor, destructor and new private members.
37690 (delete_dead_call, delete_dead_assignment): New extracted from
37692 (dse_optimize_stmt): Make a member of dse_dom_walker.
37693 Use initialize_ao_ref_for_dse.
37695 PR tree-optimization/33562
37696 PR tree-optimization/61912
37697 PR tree-optimization/77485
37698 * sbitmap.h (bitmap_count_bits): Prototype.
37699 (bitmap_clear_range, bitmap_set_range): Likewise.
37700 * sbitmap.c (bitmap_clear_range): New function.
37701 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
37703 2017-01-13 Martin Liska <mliska@suse.cz>
37706 * function.c (set_cfun): Add new argument force.
37707 * function.h (set_cfun): Likewise.
37708 * ipa-inline-transform.c (inline_call): Use the function when
37709 strict alising from is dropped for function we inline to.
37711 2017-01-13 Richard Biener <rguenther@suse.de>
37713 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
37714 for dumping GIMPLE INTEGER_CSTs.
37716 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
37718 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
37719 to 201112L since C++17.
37721 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
37724 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
37725 if -fsanitize=kernel-address is present.
37727 2017-01-13 Richard Biener <rguenther@suse.de>
37729 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
37730 as _Literal ( type ) number in case usual suffixes do not
37731 preserve all information.
37733 2017-01-13 Richard Biener <rguenther@suse.de>
37735 PR tree-optimization/77283
37736 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
37737 and ssa-iterators.h.
37738 (is_feasible_trace): Implement a cost model based on joiner
37741 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
37744 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
37745 char or short to __float128/_Float128 directly.
37747 2017-01-12 Martin Sebor <msebor@redhat.com>
37749 to -Wformat-overflow.
37750 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
37751 (min_bytes_remaining): Same.
37752 (get_string_length): Same.
37753 (format_string): Same.
37754 (format_directive): Same.
37756 (pass_sprintf_length::handle_gimple_call): Same.
37758 2017-01-12 Jakub Jelinek <jakub@redhat.com>
37760 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
37761 info.nowrite calls with no lhs that can't throw. Return bool
37762 whether gsi_remove has been called or not.
37763 (pass_sprintf_length::handle_gimple_call): Return bool whether
37764 try_substitute_return_value called gsi_remove. Formatting fix.
37765 (pass_sprintf_length::execute): Don't use gsi_remove if
37766 handle_gimple_call returned true.
37769 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
37770 be removed due to side-effects, don't remove following barrier nor
37771 turn the successor edge into fallthru edge.
37773 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37776 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
37777 element-reversing loads and stores as not swappable.
37779 2017-01-12 Nathan Sidwell <nathan@acm.org>
37780 Nicolai Stange <nicstange@gmail.com>
37782 * combine.c (try_combine): Don't ignore result of overlap checking
37783 loop. Combine overlap & asm check into single loop.
37785 2017-01-12 Richard Biener <rguenther@suse.de>
37787 * tree-pretty-print.c (dump_generic_node): Provide -gimple
37788 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
37790 2017-01-12 Richard Biener <rguenther@suse.de>
37792 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
37793 and TS_TARGET_OPTION directly derive from TS_BASE.
37794 * tree-core.h (tree_optimization_option): Derive from tree_base.
37795 (tree_target_option): Likewise.
37797 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
37799 * config/i386/i386.c (memory_address_length): Increase len
37800 only when rip_relative_addr_p returns false.
37802 2017-01-11 Julia Koval <julia.koval@intel.com>
37804 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
37805 (OPTION_MASK_ISA_SGX_SET): New.
37806 (ix86_handle_option): Handle OPT_msgx.
37807 * config.gcc: Added sgxintrin.h.
37808 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
37809 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
37810 * config/i386/i386.c (ix86_target_string): Add -msgx.
37812 (ix86_option_override_internal): Handle new options.
37813 (ix86_valid_target_attribute_inner_p): Add sgx.
37814 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
37815 * config/i386/i386.opt: Add msgx.
37816 * config/i386/sgxintrin.h: New file.
37817 * config/i386/x86intrin.h: Add sgxintrin.h.
37819 2017-01-11 Jakub Jelinek <jakub@redhat.com>
37822 * fold-const.c (maybe_nonzero_address): Return 1 for function
37824 (tree_single_nonzero_warnv_p): Don't handle function local objects
37828 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
37831 2017-01-11 David Malcolm <dmalcolm@redhat.com>
37834 * opts.c: Include "spellcheck.h"
37835 (struct string_fragment): New struct.
37836 (struct edit_distance_traits<const string_fragment &>): New
37838 (get_closest_sanitizer_option): New function.
37839 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
37841 2017-01-11 Jakub Jelinek <jakub@redhat.com>
37843 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
37845 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
37846 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
37847 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
37848 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
37849 for initial die_offset if dwarf_split_debug_info.
37850 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
37851 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
37853 (output_skeleton_debug_sections): Formatting fix. Use
37854 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
37855 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
37857 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
37859 * config/arm/cortex-a53.md: Add bypasses for
37860 cortex_a53_r2f_cvt.
37861 (cortex_a53_r2f): Only use for transfers.
37862 (cortex_a53_f2r): Likewise.
37863 (cortex_a53_r2f_cvt): Add reservation for conversions.
37864 (cortex_a53_f2r_cvt): Likewise.
37866 2017-01-11 Tamar Christina <tamar.christina@arm.com>
37868 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
37869 to all inlined functions, change static to extern.
37871 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
37874 * config/arm/arm.c (legitimize_pic_address): Handle reference to
37876 (arm_assemble_integer): Likewise.
37878 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
37880 * config.gcc: Use new awk script to check CPU, FPU and architecture
37881 parameters for --with-... options.
37882 * config/arm/parsecpu.awk: New file
37883 * config/arm/arm-cpus.in: New file.
37884 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
37886 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
37888 * config/arm/t-arm: Update dependency rules.
37889 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
37890 of processing .def files.
37891 * config/arm/genopt.sh: Deleted.
37892 * config/arm/gentune.sh: Deleted.
37893 * config/arm/arm-cores.def: Deleted.
37894 * config/arm/arm-arches.def: Deleted.
37895 * config/arm/arm-fpus.def: Deleted.
37896 * config/arm/arm-tune.md: Regenerated.
37897 * config/arm/arm-tables.opt: Regenerated.
37898 * config/arm/arm-cpu.h: New generated file.
37899 * config/arm/arm-cpu-data.h: New generated file.
37900 * config/arm/arm-cpu-cdata.h: New generated file.
37902 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
37905 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
37907 (input_varpool_node): Unpack dynamically_initialized bit.
37909 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
37911 PR rtl-optimization/79032
37912 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
37913 the alignment of the adjusted memory reference against that of MODE,
37914 instead of the alignment of the original memory reference.
37916 2017-01-11 Martin Jambor <mjambor@suse.cz>
37918 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
37920 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
37921 decorated functions.
37923 2017-01-11 Richard Biener <rguenther@suse.de>
37925 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
37926 set range/nonnull info for PHI results. Do not set it on
37927 stmts marked for removal.
37929 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
37931 * expr.c (store_field): In the bitfield case, fetch the return value
37932 from the registers before applying a single big-endian adjustment.
37933 Always do a final load for a BLKmode value not larger than a word.
37935 2017-01-10 David Malcolm <dmalcolm@redhat.com>
37938 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
37939 that we correctly handle column numbers greater than
37940 LINE_MAP_MAX_COLUMN_NUMBER.
37942 2017-01-10 Martin Sebor <msebor@redhat.com>
37944 PR middle-end/78245
37945 * gimple-ssa-sprintf.c (get_destination_size): Call
37946 {init,fini}object_sizes.
37947 * tree-object-size.c (addr_object_size): Adjust.
37948 (pass_through_call): Adjust.
37949 (pass_object_sizes::execute): Adjust.
37950 * tree-object-size.h (fini_object_sizes): Declare.
37952 2017-01-10 Martin Sebor <msebor@redhat.com>
37954 PR tree-optimization/78775
37955 * builtins.c (get_size_range): Move...
37956 * calls.c: ...to here.
37957 (alloc_max_size): Accept zero argument.
37958 (operand_signed_p): Remove.
37959 (maybe_warn_alloc_args_overflow): Call get_size_range.
37960 * calls.h (get_size_range): Declare.
37962 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
37964 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
37965 from TI's devices.csv file as of September 2016.
37966 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
37968 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
37970 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
37971 * doc/invoke.texi: Likewise.
37972 * doc/md.texi: Likewise.
37973 * doc/objc.texi: Likewise.
37975 2017-01-10 Joshua Conner <joshconner@google.com>
37977 * config/arm/fuchsia-elf.h: New file.
37978 * config/fuchsia.h: New file.
37979 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
37980 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
37982 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
37984 2016-01-10 Richard Biener <rguenther@suse.de>
37986 PR tree-optimization/79034
37987 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
37988 Propagate out degenerate PHIs in the joiner.
37990 2017-01-10 Martin Liska <mliska@suse.cz>
37992 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
37993 (sort_congruence_classes_by_decl_uid): Likewise.
37994 (sort_congruence_class_groups_by_decl_uid): Likewise.
37995 (sem_item_optimizer::merge_classes): Sort class, groups in these
37996 classes and members in the groups by DECL_UID of declarations.
37997 This would make merge operations stable.
37999 2017-01-10 Martin Liska <mliska@suse.cz>
38001 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
38002 usage of m_classes_vec.
38003 (sem_item_optimizer::~sem_item_optimizer): Likewise.
38004 (sem_item_optimizer::get_group_by_hash): Likewise.
38005 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
38006 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
38007 (sem_item_optimizer::verify_classes): Likewise.
38008 (sem_item_optimizer::process_cong_reduction): Likewise.
38009 (sem_item_optimizer::dump_cong_classes): Likewise.
38010 (sem_item_optimizer::merge_classes): Likewise.
38011 * ipa-icf.h (congruence_class_hash): Rename from
38012 congruence_class_group_hash. Remove declaration of m_classes_vec.
38014 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
38016 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
38017 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
38018 * config.gcc: Add avx512vpopcntdqintrin.h.
38019 * config/i386/avx512vpopcntdqintrin.h: New.
38020 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
38021 * config/i386/i386-builtin-types.def: Add new types.
38022 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
38023 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
38024 __builtin_ia32_vpopcountq_v8di_mask): New.
38025 * config/i386/i386-c.c (ix86_target_macros_internal): Define
38026 __AVX512VPOPCNTDQ__.
38027 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
38028 (PTA_AVX512VPOPCNTDQ): Define.
38029 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
38030 TARGET_AVX512VPOPCNTDQ_P): Define.
38031 * config/i386/i386.opt: Add mavx512vpopcntdq.
38032 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
38033 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
38035 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38037 PR middle-end/77484
38038 * predict.def (PRED_CALL): Set to 67.
38040 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
38042 * expr.c (store_field): In the bitfield case, if the value comes from
38043 a function call and is of an aggregate type returned in registers, do
38044 not modify the field mode; extract the value in all cases if the mode
38045 is BLKmode and the size is not larger than a word.
38047 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
38050 * config/i386/cpuid.h: Fix undefined behavior.
38052 2017-01-04 Jeff Law <law@redhat.com>
38054 PR tree-optimization/79007
38055 PR tree-optimization/67955
38056 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
38057 conservative for pt.null when flag_non_call_exceptions is on.
38059 2017-01-09 Jakub Jelinek <jakub@redhat.com>
38061 PR translation/79019
38062 PR translation/79020
38063 * params.def (PARAM_INLINE_MIN_SPEEDUP,
38064 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
38065 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
38067 * config/avr/avr.opt (maccumulate-args): Likewise.
38068 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
38069 * common.opt (freport-bug): Likewise.
38070 * cif-code.def (CIF_FINAL_ERROR): Likewise.
38071 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
38072 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
38073 translatable string.
38074 * config/i386/i386.c (function_value_32): Likewise.
38075 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
38076 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
38078 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
38079 * common/config/msp430/msp430-common.c (msp430_handle_option):
38081 * symtab.c (symtab_node::verify_base): Likewise.
38082 * opts.c (set_debug_level): Likewise.
38083 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
38084 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
38085 missing whitespace to translatable strings.
38086 * config/avr/avr.md (bswapsi2): Fix typo in comment.
38087 * config/sh/superh.h: Likewise.
38088 * config/i386/xopintrin.h: Likewise.
38089 * config/i386/znver1.md: Likewise.
38090 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
38091 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
38092 * double-int.h (struct double_int): Likewise.
38093 * double-int.c (div_and_round_double): Likewise.
38094 * wide-int.cc: Likewise.
38095 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
38096 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
38097 * cfgcleanup.c (crossjumps_occured): Renamed to ...
38098 (crossjumps_occurred): ... this.
38099 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
38102 PR tree-optimization/78899
38103 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
38104 returning bool return struct loop *, NULL for failure and the new
38106 (versionable_outer_loop_p): Don't version outer loop if it has
38107 dont_vectorized bit set.
38108 (tree_if_conversion): When versioning outer loop, ensure
38109 tree_if_conversion is performed also on the inner loop of the
38110 non-vectorizable outer loop copy.
38111 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
38112 LOOP_VECTORIZED in inner loop of the scalar outer loop and
38113 prevent vectorization of it.
38114 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
38115 the outer loop vectorization of the non-scalar version is attempted
38116 before vectorization of the inner loop in scalar version. If
38117 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
38118 vectorization of its inner loop.
38119 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
38120 has 2 inner loops, rename also on edges from bb whose single pred
38121 is outer_loop->header. Fix typo in function comment.
38123 2017-01-09 Martin Sebor <msebor@redhat.com>
38126 * asan.c (asan_emit_stack_protection): Increase local buffer size
38127 to avoid snprintf truncation warning.
38129 2017-01-09 Andrew Pinski <apinski@cavium.com>
38131 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
38132 to reference thunderx2t99 for the tuning structure
38133 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
38135 (thunderx2t99_extra_costs): This.
38136 * config/aarch64/aarch64-tune.md: Regenerate.
38137 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
38138 (vulcan_addrcost_table): This.
38139 (vulcan_regmove_cost): Rename to ...
38140 (thunderx2t99_regmove_cost): This.
38141 (vulcan_vector_cost): Rename to ...
38142 (thunderx2t99_vector_cost): this.
38143 (vulcan_branch_cost): Rename to ...
38144 (thunderx2t99_branch_cost): This.
38145 (vulcan_tunings): Rename to ...
38146 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
38147 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
38149 2017-01-09 Martin Jambor <mjambor@suse.cz>
38153 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
38154 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
38155 (propagate_vr_accross_jump_function): Use the above function for all
38156 value range computations for pass-through jump functions and type
38157 converasion from explicit value range values.
38158 (ipcp_propagate_stage): Do not attempt to deduce types of formal
38159 parameters from TYPE_ARG_TYPES.
38160 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
38161 (ipa_write_node_info): Stream type of the actual argument.
38162 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
38164 2017-01-09 Martin Liska <mliska@suse.cz>
38167 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
38168 (lookup_compiler): Do not show error message with have_E.
38170 2017-01-09 Jakub Jelinek <jakub@redhat.com>
38172 PR tree-optimization/78938
38173 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
38174 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
38175 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
38176 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
38179 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38181 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
38184 2017-01-09 Richard Biener <rguenther@suse.de>
38186 PR tree-optimization/78997
38187 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
38188 name condition properly.
38190 2017-01-09 Richard Biener <rguenther@suse.de>
38193 * dwarf2out.c (is_cxx): New overload with context.
38194 (is_naming_typedef_decl): Use it.
38196 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38198 * invoke.texi (Option Summary): Correct spacing in option lists
38199 and add line breaks to fix over-long lines.
38201 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38203 PR middle-end/17660
38205 * extend.texi (Common Variable Attributes): Add xref to GCC
38206 Internals manual to explain mode attribute keywords.
38208 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38211 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
38212 and Preprocessor Options.
38213 (Options for Linking): Document -pthread here....
38214 (RS/6000 and PowerPC Options): ...not here.
38215 (Solaris 2 Options): ...or here.
38216 * doc/cppopts.texi: Document -pthread.
38218 2017-01-08 Martin Sebor <msebor@redhat.com>
38220 PR middle-end/77708
38221 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
38222 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
38223 New member functions.
38224 (format_directive): Used them.
38226 (pass_sprintf_length::handle_gimple_call): Same.
38227 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
38228 to avoid truncation for any argument.
38229 (extract_affine_mul): Same.
38230 * tree.c (get_file_function_name): Same.
38232 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38234 PR middle-end/77484
38235 * predict.def (PRED_INDIR_CALL): Set to 86.
38237 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38239 PR preprocessor/54124
38240 * doc/cppopts.texi: Reformat -d subtable to list the full name
38241 of the options. Add cross-reference to the docs for the general
38242 compiler -d options.
38243 * doc/invoke.texi (Developer Options): Add cross-reference to the
38244 preprocessor-specific -d option documentation.
38246 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38248 PR preprocessor/13498
38249 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
38250 redudant material, and reflect new command-line options.
38251 (System Headers): Likewise.
38253 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38255 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
38256 -isystem, and -idirafter. Copy-edit.
38257 * doc/cppopts.texi: Copy-edit. Remove contradiction about
38258 default for -ftrack-macro-expansion. Delete obsolete and
38259 badly-formatted implementation details about -fdebug-cpp output.
38260 * doc/cppwarnopts.texi: Copy-edit.
38262 2017-01-07 David Malcolm <dmalcolm@redhat.com>
38265 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
38266 that the transition from a max line width >= 1<<10 to narrower
38267 lines works correctly.
38269 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
38271 * doc/options.texi (PerFunction): New.
38272 * opt-functions.awk (switch_flags): Map both Optimization and
38273 PerFunction to CL_OPTIMIZATION.
38274 * opth-gen.awk: Test for PerFunction flag along with
38276 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
38277 it only when the latter is present. Skip those that don't in
38278 the hash function generator.
38279 * common.opt (fvar-tracking): Mark as PerFunction instead of
38281 (fvar-tracking-assignments): Likewise.
38282 (fvar-tracking-assignments-toggle): Likewise.
38283 (fvar-tracking-uninit): Likewise.
38285 2017-01-07 Jakub Jelinek <jakub@redhat.com>
38287 PR translation/79018
38288 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
38291 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
38294 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
38295 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
38296 TARGET_LONG_JUMP_TABLE_OFFSETS.
38297 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
38298 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
38299 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
38300 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
38301 * config/m68k/m68k.md (tablejump expander): Likewise.
38302 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
38303 TARGET_LONG_JUMP_TABLE_OFFSETS.
38304 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
38305 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
38307 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
38308 David Holsgrove <david.holsgrove@xilinx.com>
38310 * common/config/microblaze/microblaze-common.c
38311 (TARGET_EXCEPT_UNWIND_INFO): Remove.
38312 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
38314 * config/microblaze/microblaze.c (microblaze_must_save_register)
38315 (microblaze_expand_epilogue, microblaze_return_addr): Handle
38317 (microblaze_eh_return): New function.
38318 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
38319 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
38320 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
38321 * config/microblaze/microblaze.md (eh_return): New pattern.
38323 2017-01-06 Jakub Jelinek <jakub@redhat.com>
38325 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
38326 GCC_DIAGNOSTIC_STRINGIFY): Define.
38328 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
38330 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38332 * config/arm/arm.md (<mcrr>): New.
38334 * config/arm/arm.c (arm_arch5te): New.
38335 (arm_option_override): Set arm_arch5te.
38336 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
38338 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
38339 (arm_mcrr_qualifiers): ... this. New.
38340 (MRRC_QUALIFIERS): Define to...
38341 (arm_mrrc_qualifiers): ... this. New.
38342 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
38344 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
38345 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
38346 (MRRCI, mrrc, MRRC): New.
38347 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
38348 VUNSPEC_MRRC2): New.
38350 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38352 * config/arm/arm.md (<mcr>): New.
38354 * config/arm/arm.c (arm_coproc_builtin_available): Add
38355 support for mcr, mrc, mcr2 and mrc2.
38356 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
38357 (arm_mcr_qualifiers): ... this. New.
38358 (MRC_QUALIFIERS): Define to ...
38359 (arm_mrc_qualifiers): ... this. New.
38360 (MCR_QUALIFIERS): Define to ...
38361 (arm_mcr_qualifiers): ... this. New.
38362 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
38364 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
38365 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
38366 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
38367 VUNSPEC_MRC2): New.
38369 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38371 * config/arm/arm.md (*ldc): New.
38375 * config/arm/arm.c (arm_coproc_builtin_available): Add
38376 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
38377 (arm_coproc_ldc_stc_legitimate_address): New.
38378 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
38379 'qualifier_const_pointer'.
38380 (LDC_QUALIFIERS): Define to...
38381 (arm_ldc_qualifiers): ... this. New.
38382 (STC_QUALIFIERS): Define to...
38383 (arm_stc_qualifiers): ... this. New.
38384 * config/arm/arm-protos.h
38385 (arm_coproc_ldc_stc_legitimate_address): New.
38386 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
38387 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
38388 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
38389 stc2, stcl, stc2l): New.
38390 * config/arm/constraints.md (Uz): New.
38391 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
38392 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
38393 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
38394 VUNSPEC_STC2L): New.
38396 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38398 * config/arm/arm.md (<cdp>): New.
38399 * config/arm/arm.c (neon_const_bounds): Rename this ...
38400 (arm_const_bounds): ... this.
38401 (arm_coproc_builtin_available): New.
38402 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
38403 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
38404 (CDP_QUALIFIERS): Define to...
38405 (arm_cdp_qualifiers): ... this. New.
38407 (arm_expand_builtin_args): Add case for 6 arguments.
38408 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
38409 (arm_const_bounds): ... this.
38410 (arm_coproc_builtin_available): New.
38411 * config/arm/arm_acle.h (__arm_cdp): New.
38413 * config/arm/arm_acle_builtins.def (cdp): New.
38415 * config/arm/iterators.md (CDPI,CDP,cdp): New.
38416 * config/arm/neon.md: Rename all 'neon_const_bounds' to
38417 'arm_const_bounds'.
38418 * config/arm/types.md (coproc): New.
38419 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
38420 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
38421 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
38422 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
38424 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38426 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
38427 (UBINOP_QUALIFIERS): New.
38429 (acle_builtin_data): New. Change comment.
38430 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
38431 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
38432 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
38433 arm_acle_builtins.def.
38434 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
38435 (arm_init_acle_builtins): New.
38436 (CRC32_BUILTIN): Remove.
38437 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
38438 crc32cb, crc32ch and crc32cw.
38439 (arm_init_crc32_builtins): Remove.
38440 (arm_init_builtins): Use arm_init_acle_builtins rather
38441 than arm_init_crc32_builtins.
38442 (arm_expand_acle_builtin): New.
38443 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
38444 * config/arm/arm_acle_builtins.def: New.
38446 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38448 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
38449 (arm_builtin_datum): ... this.
38450 (arm_init_neon_builtin): Rename to ...
38451 (arm_init_builtin): ... this. Add a new parameters PREFIX
38452 and USE_SIG_IN_NAME.
38453 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
38454 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
38455 'arm_builtin_datum'.
38456 (arm_init_vfp_builtins): Likewise.
38457 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
38458 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
38459 (arm_expand_neon_args): Rename to ...
38460 (arm_expand_builtin_args): ... this. Rename builtin_arg
38461 enum values and differentiate between ARG_BUILTIN_MEMORY
38462 and ARG_BUILTIN_NEON_MEMORY.
38463 (arm_expand_neon_builtin_1): Rename to ...
38464 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
38465 values, arm_expand_builtin_args and add bool parameter NEON.
38466 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
38467 (arm_expand_vfp_builtin): Likewise.
38468 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
38470 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38472 PR middle-end/77484
38473 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
38474 * predict.c (tree_estimate_probability_bb): Reverse direction of
38475 polymorphic call predictor.
38477 2017-01-06 David Malcolm <dmalcolm@redhat.com>
38479 * passes.c (execute_one_pass): Split out pass-skipping logic into...
38480 (determine_pass_name_match): ...this new function and...
38481 (should_skip_pass_p): ...this new function.
38483 2017-01-06 Nathan Sidwell <nathan@acm.org>
38485 * ipa-visibility.c (function_and_variable_visibility): Reformat
38486 comments and long lines. Remove extrneous if.
38487 * symtab.c (symtab_node::make_decl_local): Fix code format.
38488 (symtab_node::set_section_for_node): Fix comment typo.
38490 2017-01-06 Martin Liska <mliska@suse.cz>
38493 * lra-constraints.c: Rename invariant to lra_invariant.
38494 * predict.c (set_even_probabilities): Initialize e to NULL.
38496 2017-01-05 Martin Sebor <msebor@redhat.com>
38498 PR tree-optimization/78910
38499 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
38500 (format_integer): Correct off-by-one error in the handling
38501 of precision with negative numbers in signed conversions..
38503 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
38505 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
38507 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38509 PR tree-optimization/71016
38510 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
38511 factor_out_conditional_conversion. Formatting fix.
38512 (factor_out_conditional_conversion): Add cond_stmt argument.
38513 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
38514 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
38517 2017-01-05 David Malcolm <dmalcolm@redhat.com>
38519 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
38520 read-rtl-function.o, and selftest-rtl.o.
38521 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
38522 (selftest::aarch64_test_loading_full_dump): New function.
38523 (selftest::aarch64_run_selftests): New function.
38524 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
38525 selftest::aarch64_run_selftests.
38526 * config/i386/i386.c
38527 (selftest::ix86_test_loading_dump_fragment_1): New function.
38528 (selftest::ix86_test_loading_call_insn): New function.
38529 (selftest::ix86_test_loading_full_dump): New function.
38530 (selftest::ix86_test_loading_unspec): New function.
38531 (selftest::ix86_run_selftests): Call the new functions.
38532 * emit-rtl.c (maybe_set_max_label_num): New function.
38533 * emit-rtl.h (maybe_set_max_label_num): New decl.
38534 * function.c (instantiate_decls): Guard call to
38535 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
38536 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
38538 * gensupport.c (gen_reader::gen_reader): Pass "false"
38539 for new "compact" param of rtx_reader.
38540 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
38541 rather than an empty string for NULL strings.
38542 * read-md.c: Potentially include config.h rather than bconfig.h.
38543 Wrap include of errors.h with #ifdef GENERATOR_FILE.
38544 (have_error): New global, copied from errors.c.
38545 (md_reader::read_name): Rename to...
38546 (md_reader::read_name_1): ...this, adding "out_loc" param,
38547 and converting "missing name or number" to returning false, rather
38549 (md_reader::read_name): Reimplement in terms of read_name_1.
38550 (md_reader::read_name_or_nil): New function.
38551 (md_reader::read_string): Handle "(nil)" by returning NULL.
38552 (md_reader::md_reader): Add new param "compact".
38553 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
38554 (md_reader::read_file): New method.
38555 * read-md.h (md_reader::md_reader): Add new param "compact".
38556 (md_reader::read_file): New method.
38557 (md_reader::is_compact): New accessor.
38558 (md_reader::read_name): Convert return type from void to file_location.
38559 (md_reader::read_name_or_nil): New decl.
38560 (md_reader::read_name_1): New decl.
38561 (md_reader::m_compact): New field.
38562 (noop_reader::noop_reader): Pass "false" for new "compact" param
38564 (rtx_reader::rtx_reader): Add new "compact" param.
38565 (rtx_reader::read_rtx_operand): Make virtual and convert return
38566 type from void to rtx.
38567 (rtx_reader::read_until): New decl.
38568 (rtx_reader::handle_any_trailing_information): New virtual function.
38569 (rtx_reader::postprocess): New virtual function.
38570 (rtx_reader::finalize_string): New virtual function.
38571 (rtx_reader::m_in_call_function_usage): New field.
38572 (rtx_reader::m_reuse_rtx_by_id): New field.
38573 * read-rtl-function.c: New file.
38574 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
38575 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
38576 (selftest::verify_three_block_rtl_cfg): New decl.
38577 * read-rtl-function.h: New file.
38578 * read-rtl.c: Potentially include config.h rather than bconfig.h.
38579 For host, include function.h, memmodel.h, and emit-rtl.h.
38580 (one_time_initialization): New function.
38581 (struct compact_insn_name): New struct.
38582 (compact_insn_names): New array.
38583 (find_code): Handle insn codes in compact dumps.
38584 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
38585 (bind_subst_iter_and_attr): Likewise.
38586 (add_condition_to_string): Likewise.
38587 (add_condition_to_rtx): Likewise.
38588 (apply_attribute_uses): Likewise.
38589 (add_current_iterators): Likewise.
38590 (apply_iterators): Likewise.
38591 (initialize_iterators): Guard usage of apply_subst_iterator with
38592 #ifdef GENERATOR_FILE.
38593 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
38594 (md_reader::read_mapping): Likewise.
38595 (add_define_attr_for_define_subst): Likewise.
38596 (add_define_subst_attr): Likewise.
38597 (read_subst_mapping): Likewise.
38598 (check_code_iterator): Likewise.
38599 (rtx_reader::read_rtx): Likewise. Move one-time initialization
38601 (one_time_initialization): New function.
38602 (rtx_reader::read_until): New method.
38603 (read_flags): New function.
38604 (parse_reg_note_name): New function.
38605 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
38606 Handle reuse_rtx ids.
38607 Wrap iterator lookup within #ifdef GENERATOR_FILE.
38608 Add parsing support for RTL dumps, mirroring the special-cases in
38609 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
38610 values, and calling handle_any_trailing_information.
38611 (rtx_reader::read_rtx_operand): Convert return type from void
38612 to rtx, returning return_rtx. Handle case 'e'. Call
38613 finalize_string on XSTR and XTMPL fields.
38614 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
38615 "(nil)" values were omitted. Call the postprocess vfunc on the
38617 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
38618 class ctor. Initialize m_in_call_function_usage. Call
38619 one_time_initialization.
38620 * rtl-tests.c (selftest::test_uncond_jump): Call
38621 set_new_first_and_last_insn.
38622 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
38623 * selftest-rtl.c: New file.
38624 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
38625 (selftest::get_insn_by_uid): New decl.
38626 * selftest-run-tests.c (selftest::run_tests): Call
38627 read_rtl_function_c_tests.
38628 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
38629 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
38632 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
38634 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
38635 operands in a special way. Assert that pos+len <= mode precision.
38637 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38639 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
38640 3 argument Alias with unlimited for the negative form.
38641 (fno-vect-cost-model): Removed.
38643 2017-01-05 Martin Liska <mliska@suse.cz>
38645 * hsa-gen.c (gen_hsa_divmod): New function.
38646 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
38648 2017-01-05 Martin Liska <mliska@suse.cz>
38651 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
38654 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38656 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
38657 small constant length operands.
38659 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38661 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
38662 between loop iterations.
38664 2017-01-05 Martin Liska <mliska@suse.cz>
38667 * gimplify.c (gimplify_decl_expr): Compare to
38668 asan_poisoned_variables instread of checking flags.
38669 (gimplify_target_expr): Likewise.
38670 (gimplify_expr): Likewise.
38671 (gimplify_function_tree): Conditionally initialize
38672 asan_poisoned_variables.
38674 2017-01-04 Jeff Law <law@redhat.com>
38676 PR tree-optimizatin/78812
38677 * rtl.h (contains_mem_rtx_p): Prototype.
38678 * ifcvt.c (containts_mem_rtx_p): Move from here to...
38679 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
38680 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
38681 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
38682 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
38684 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38686 * input.c (assert_char_at_range): Default-initialize actual_range.
38688 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38690 * df-scan.c (df_ref_create_structure): Make regno unsigned,
38691 to match the caller.
38693 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38695 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
38696 insns after final jump in test to emit dummy move.
38698 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38700 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
38701 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
38703 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38705 * multiple_target.c (create_dispatcher_calls): Init e_next.
38706 * tree-ssa-loop-split.c (split_loop): Init border.
38707 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
38710 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
38715 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
38716 (altivec_register_operand): Do not return true if the operand
38717 contains a SUBREG mixing SImode and SFmode.
38718 (vsx_register_operand): Likewise.
38719 (vsx_reg_sfsubreg_ok): New predicate.
38720 (vfloat_operand): Do not return true if the operand contains a
38721 SUBREG mixing SImode and SFmode.
38722 (vint_operand): Likewise.
38723 (vlogical_operand): Likewise.
38724 (gpc_reg_operand): Likewise.
38725 (int_reg_operand): Likewise.
38726 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
38727 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
38728 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
38730 (rs6000_emit_move_si_sf_subreg): New helper function.
38731 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
38732 fixup SUBREGs involving SImode and SFmode.
38733 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
38734 numbers for the new peephole2 optimization.
38735 (peephole2 for SFmode unions): New peephole2 to optimize cases in
38736 the GLIBC math library that do AND/IOR/XOR operations on single
38737 precision floating point.
38738 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
38739 target macros to say whether we need to avoid SUBREGs mixing
38741 (TARGET_ALLOW_SF_SUBREG): Likewise.
38742 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
38743 (UNSPEC_SI_FROM_SF): Likewise.
38744 (iorxor): Change spacing.
38745 (and_ior_xor): New iterator for AND, IOR, and XOR.
38746 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
38747 (movdi_from_sf_zero_ext): Likewise.
38748 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
38749 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
38750 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
38751 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
38752 (fms<mode>4): Likewise.
38753 (fnma<mode>4): Likewise.
38754 (fnms<mode>4): Likewise.
38755 (nfma<mode>4): Likewise.
38756 (nfms<mode>4): Likewise.
38758 2017-01-04 Marek Polacek <polacek@redhat.com>
38761 * doc/invoke.texi: Document -Wpointer-compare.
38763 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38765 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
38768 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
38769 descriptions for -gdwarf-5 and emit them as uleb128 instead of
38772 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
38775 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
38776 documentation of the powerpc_popcntb_ok attribute.
38777 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
38778 code to issue warning messages if a requested CPU configuration is
38779 not supported by the binary (assembler and loader) toolchain.
38780 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
38781 made to define a built-in function that has been disabled.
38782 (paired_init_builtins): Add assertion to prevent ICE if attempt is
38783 made to define a built-in function that has been disabled.
38784 (altivec_init_builtins): Add comment explaining why definition
38785 of the DST built-in functions is not preceded by an assertion
38786 check. Add assertions to prevent ICE if attempts are made to
38787 define an altivec predicate or an abs* built-in function that has
38789 (htm_init_builtins): Add comment explaining why definition of the
38790 htm built-in functions is not preceded by an assertion check.
38792 2017-01-04 Jeff Law <law@redhat.com>
38794 PR tree-optimizatin/67955
38795 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
38796 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
38797 the points-to solution does not include pt_null. Use DECL_PT_UID
38800 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
38802 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
38803 Use gen_int_mode instead of gen_lopwart for const_int operands.
38805 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38807 PR tree-optimization/71563
38808 * match.pd: Simplify X << Y into X if Y is known to be 0 or
38809 out of range value - has low bits known to be zero.
38811 2017-01-04 Alan Modra <amodra@gmail.com>
38813 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
38814 * configure: Regenerate.
38815 * config.in: Regenerate.
38817 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38820 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
38821 a substring of the message, but strcmp with the whole message. Ifdef
38822 ENABLE_NLS, translate the message first using dgettext.
38824 2017-01-03 Jeff Law <law@redhat.com>
38826 PR tree-optimizatin/78856
38827 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
38828 (mark_threaded_blocks): Remove code to truncate thread paths that
38829 cross multiple loop headers. Instead invalidate the cached loop
38830 iteration information and handle case of a thread path walking
38831 into an irreducible region.
38833 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
38836 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
38837 assertions. Add support for doing the signbit if the IEEE 128-bit
38838 floating point value is in a GPR.
38839 * config/rs6000/rs6000.md (Fsignbit): Delete.
38840 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
38841 Update the length attribute if the value is in a GPR.
38842 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
38843 the sign or zero extension instruction, since the value is always 0/1.
38844 (signbit<mode>2_dm2): Delete using <Fsignbit>.
38847 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
38848 extracting SImode to a GPR register so that we can generate a
38849 store, limit the vector to be in a traditional Altivec register
38850 for the vextuwrx instruction.
38852 2017-01-03 Ian Lance Taylor <iant@google.com>
38854 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
38856 2017-01-03 Martin Sebor <msebor@redhat.com>
38858 PR tree-optimization/78696
38859 * gimple-ssa-sprintf.c (format_floating): Correct handling of
38860 precision. Use MPFR for %f for greater fidelity. Correct handling
38862 (pass_sprintf_length::compute_format_length): Set width and precision
38863 specified by asrerisk to void_node for vararg functions.
38864 (try_substitute_return_value): Adjust dump output.
38866 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
38868 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
38870 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
38872 * doc/invoke.texi (SPARC options): Document -mlra as the default.
38873 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
38874 -mlra/-mno-lra was passed to the compiler.
38876 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
38878 PR rtl-optimization/65618
38879 * emit-rtl.c (try_split): Move initialization of "before" and
38880 "after" to just before the call to emit_insn_after_setloc.
38882 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
38884 * doc/md.texi (Standard Names): Remove reference to Java frontend.
38886 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
38888 * dwarf2out.c (gen_enumeration_type_die): When
38889 -gno-strict-dwarf, add a DW_AT_encoding attribute.
38891 2017-01-03 Jakub Jelinek <jakub@redhat.com>
38893 PR tree-optimization/78965
38894 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
38895 Change first argument from const call_info & to call_info &. For %n
38896 set info.nowrite to false.
38898 PR middle-end/78901
38899 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
38900 possibly throwing calls.
38902 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
38903 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
38904 and fns handling, rather than in a separate case SSA_NAME.
38906 2017-01-02 Jeff Law <law@redhat.com>
38908 * config/darwin-driver.c (darwin_driver_init): Const-correctness
38909 fixes for first_period and second_period variables.
38911 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
38914 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
38915 (*insvqi_1): New insn pattern.
38916 (*insvqi_1_mem_rex64): Ditto.
38917 (*insvqi_2): Ditto.
38918 (*insvqi_3): Rename from *insvqi.
38920 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
38922 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
38924 * doc/cfg.texi (Edges): Remove reference to Java.
38925 (Maintaining the CFG): Ditto.
38927 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38929 PR middle-end/77674
38930 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
38931 transparent aliases.
38933 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38935 PR middle-end/77484
38936 * predict.def (PRED_CALL): Update hitrate.
38937 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
38938 * predict.c (tree_estimate_probability_bb): Split CALL predictor
38939 into direct/indirect/polymorphic variants.
38941 2017-01-01 Jakub Jelinek <jakub@redhat.com>
38943 Update copyright years.
38945 * gcc.c (process_command): Update copyright notice dates.
38946 * gcov-dump.c (print_version): Ditto.
38947 * gcov.c (print_version): Ditto.
38948 * gcov-tool.c (print_version): Ditto.
38949 * gengtype.c (create_file): Ditto.
38950 * doc/cpp.texi: Bump @copying's copyright year.
38951 * doc/cppinternals.texi: Ditto.
38952 * doc/gcc.texi: Ditto.
38953 * doc/gccint.texi: Ditto.
38954 * doc/gcov.texi: Ditto.
38955 * doc/install.texi: Ditto.
38956 * doc/invoke.texi: Ditto.
38958 Copyright (C) 2017 Free Software Foundation, Inc.
38960 Copying and distribution of this file, with or without modification,
38961 are permitted in any medium without royalty provided the copyright
38962 notice and this notice are preserved.