1 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
4 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
5 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
6 * config/sh/sh.c: Skip debug insns besides notes.
7 * config/sh/sh.md: Likewise.
8 * config/sh/sh_treg_combine.cc: Likewise.
9 * config/sh/sync.md: Likewise.
11 2017-12-14 Tom de Vries <tom@codesourcery.com>
13 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
14 Add item for weak_undefined.
16 2017-12-14 Richard Biener <rguenther@suse.de>
18 PR tree-optimization/67842
19 * sese.h (bb_in_region): Remove #if 0'ed code.
21 2017-12-14 Richard Biener <rguenther@suse.de>
23 PR tree-optimization/83326
24 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
25 may_be_zero parameter and handle it by not marking the first
26 peeled copy as not exiting the loop.
27 (try_peel_loop): Likewise.
28 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
29 to handle the case of constant or zero iterations and perform
30 loop header copying on-the-fly.
32 2017-12-14 Richard Biener <rguenther@suse.de>
34 PR tree-optimization/83418
35 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
36 Instead of asserting we don't get unfolded comparisons deal with
39 2017-12-14 Jakub Jelinek <jakub@redhat.com>
42 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
43 basic blocks. Assert debug bind insns don't appear outside of bbs,
44 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
47 PR tree-optimization/83198
48 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
49 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
50 value if arg is a REAL_CST with incompatible type.
52 2017-12-14 Sudakshina Das <sudi.das@arm.com>
53 Bin Cheng <bin.cheng@arm.com>
56 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
58 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
61 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
63 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
64 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
66 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
67 Alan Hayward <alan.hayward@arm.com>
68 David Sherwood <david.sherwood@arm.com>
70 * poly-int.h: New file.
71 * poly-int-types.h: Likewise.
72 * coretypes.h: Include them.
73 (POLY_INT_CONVERSION): Define.
74 * target.def (estimated_poly_value): New hook.
75 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
76 * doc/tm.texi: Regenerate.
77 * doc/poly-int.texi: New file.
78 * doc/gccint.texi: Include it.
79 * doc/rtl.texi: Describe restrictions on subreg modes.
80 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
81 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
82 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
83 * targhooks.h (default_estimated_poly_value): Declare.
84 * targhooks.c (default_estimated_poly_value): New function.
85 * target.h (estimated_poly_value): Likewise.
86 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
87 (wi::unary_traits): Delete.
88 (wi::binary_traits::signed_shift_result_type): Define for
89 offset_int << HOST_WIDE_INT, etc.
90 (generic_wide_int::operator <<=): Define for all types and use
91 wi::lshift instead of <<.
92 (wi::hwi_with_prec): Add a default constructor.
93 (wi::ints_for): New class.
94 (operator <<): Define for all wide-int types.
95 (operator /): New function.
96 (operator %): Likewise.
97 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
98 (ASSERT_MAYBE_NE_AT): New macros.
100 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
101 Dominik Vogt <vogt@linux.vnet.ibm.com>
104 * emit-rtl.c (init_emit): Remove ??? comment.
105 * explow.c (get_dynamic_stack_size): Take known alignment of stack
106 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
107 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
108 alignment of 3 virtual registers to BITS_PER_WORD.
110 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
112 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
114 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
115 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
117 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
120 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
122 2017-12-13 Jakub Jelinek <jakub@redhat.com>
124 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
125 with the exception of debug begin stmt markers appear before
129 * final.c (rest_of_handle_final): Call variable_tracking_main only
130 if !flag_var_tracking.
132 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
133 Jakub Jelinek <jakub@redhat.com>
137 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
138 labels debug stmts that can only appear after labels.
140 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
142 PR rtl-optimization/82398
143 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
144 EXPR_USEFULNESS in priority comparison.
146 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
148 PR rtl-optimization/83393
149 * combine.c (move_deaths): If reg_stat points to a too new insn in
150 last_death, do not use it: find the proper insn instead.
152 2017-12-12 Jeff Law <law@redhat.com>
154 PR tree-optimization/83298
155 PR tree-optimization/83362
156 PR tree-optimization/83383
157 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
158 push_value_range a public interface. Add new argument to
159 record_ranges_from_stmt.
160 * gimple-ssa-evrp-analyze.c
161 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
162 Update comments. Handle recording temporary equivalences.
163 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
164 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
165 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
166 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
167 gimple-ssa-evrp-analyze.h.
168 (record_temporary_equivalences_from_phis): Add new argument. When
169 the PHI arg is an SSA_NAME, set the result's range to the range
171 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
173 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
174 Pass it down to children as needed.
175 (thread_outgoing_edges): Likewise.
176 (thread_across_edge): Likewise. Push/pop range state as needed.
177 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
179 2017-12-12 Julia Koval <julia.koval@intel.com>
181 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
182 (PTA_CANNONLAKE): Remove PTA_CLWB.
184 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
187 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
188 (vcondv2div2df): Likewise.
189 (vconduv2dfv2di): Likewise.
191 2017-12-12 Jakub Jelinek <jakub@redhat.com>
193 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
194 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
195 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
196 ATTR_MATHFN_FPROUNDING_ERRNO.
198 2017-12-12 Richard Biener <rguenther@suse.de>
200 PR tree-optimization/83385
201 * graphite-scop-detection.c (get_order, order): Remove.
202 (bb_to_rpo): New global.
204 (build_scops): Sort pbbs in RPO order.
206 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
208 * combine.c (simplify_set): Do not transform subregs to zero_extends
209 if the destination mode is a vector mode.
211 2017-12-12 Jakub Jelinek <jakub@redhat.com>
213 PR tree-optimization/83359
214 * tree-cfg.h (fold_loop_internal_call): Declare.
215 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
216 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
217 (find_loop_dist_alias): New function.
218 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
219 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
222 PR tree-optimization/80631
223 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
224 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
225 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
226 hardcoding zero as the value if COND_EXPR is never true. For
227 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
228 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
229 hardcoding MAX_EXPR as the reduction operation.
230 (is_nonwrapping_integer_induction): Allow negative step.
231 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
232 vect_create_epilog_for_reduction, if no value is suitable, don't
233 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
235 2017-12-12 Richard Biener <rguenther@suse.de>
237 PR tree-optimization/81889
238 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
239 range info from the non-wrapping IV instead of just the range
242 2017-12-12 Julia Koval <julia.koval@intel.com>
244 * config.gcc: Add vaesintrin.h.
245 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
247 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
248 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
250 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
251 * config/i386/immintrin.h: Include vaesintrin.h.
252 * config/i386/sse.md (vaesdec_<mode>): New pattern.
253 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
254 _mm_aesdec_epi128): New intrinsics.
256 2017-12-12 Julia Koval <julia.koval@intel.com>
258 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
259 OPTION_MASK_ISA_VAES_UNSET): New.
260 (ix86_handle_option): Handle -mvaes.
261 * config/i386/cpuid.h: Define bit_VAES.
262 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
263 * config/i386/i386-c.c (__VAES__): New.
264 * config/i386/i386.c (ix86_target_string): Add -mvaes.
265 (ix86_valid_target_attribute_inner_p): Ditto.
266 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
267 * config/i386/i386.opt: Add -mvaes.
268 * doc/invoke.texi: Ditto.
270 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
272 * debug.h (gcc_debug_hooks): Add inline_entry.
273 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
274 * debug.c (do_nothing_debug_hooks): Likewise.
275 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
276 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
277 (dwarf2_lineno_debug_hooks): Likewise.
279 * common.opt (gstatement-frontiers): New, setting
280 debug_nonbind_markers_p.
281 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
282 * toplev.c (process_options): Autodetect value for debug statement
284 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
285 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
287 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
288 markers. Integrate source bind into debug stmt expand loop.
289 (pass_expand::execute): Check debug marker limit. Avoid deep
290 TER and expand debug locations for debug bind insns only.
291 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
293 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
294 * final.c (reemit_insn_block_notes): Take current block from
295 nonbind markers. Declare note where it's first set.
296 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
297 begin stmt markers if enabled.
298 (notice_source_line): Handle nonbind markers. Fail if their
299 location is unknown or that of builtins.
300 (rest_of_handle_final): Convert begin stmt markers to notes if
301 var-tracking didn't run.
302 (rest_of_clean_state): Skip begin stmt markers.
303 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
305 * function.c (allocate_struct_function): Set begin_stmt_markers.
306 * function.h (struct function): Add debug_marker_count counter
307 and debug_nonbind_markers flag.
308 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
309 * gimple-low.c (lower_function_body): Adjust
310 debug_nonbind_markers.
311 (lower_stmt): Drop or skip gimple debug stmts.
312 (lower_try_catch): Skip debug stmts.
313 * gimple.c (gimple_build_debug_begin_stmt): New.
314 (gimple_copy): Increment debug_marker_count if copying one.
315 * gimple.h (gimple_build_debug_begin_stmt): Declare.
316 * gimplify.c (rexpr_location): New.
317 (rexpr_has_location): New.
318 (warn_switch_unreachable_r): Handle gimple debug stmts.
319 (shortcut_cond_r): Call expr_location.
321 (find_goto_label): New.
322 (shortcut_cond_expr): Call expr_has_location, expr_location, and
324 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
326 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
327 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
328 (LANG_HOOKS_INITIALIZER): ... this.
329 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
330 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
332 (update_ebb_live_info): Skip debug insn markers.
333 * lra.c (debug_insn_static_data): Rename to...
334 (debug_bind_static_data): ... this.
335 (debug_marker_static_data): New.
336 (lra_set_insn_recog_data): Select one of the above depending
338 (lra_update_isn_regno_info): Don't assume debug insns have
340 (push_insns): Skip debug insns.
341 * lto-streamer-in.c (input_function): Drop debug stmts
342 depending on active options. Adjust debug_nonbind_markers.
343 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
344 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
345 begin stmt marker notes.
346 (print_insn): Likewise.
347 * recog.c (extract_insn): Recognize rtl for debug markers.
348 * rtl.def (DEBUG_MARKER): New.
349 * tree-inline.c: Include params.h.
350 (remap_gimple_stmt): Handle nonbind markers.
351 (maybe_move_debug_stmts_to_successors): Likewise.
352 (copy_debug_stmt): Likewise.
353 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
354 markers regardless of no side effects.
355 (tsi_link_before): Don't update container's side effects when adding
357 (tsi_link_after): Likewise.
358 (expr_first): Skip begin stmt markers.
359 (expr_last): Likewise.
360 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
361 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
362 Disregard nonbind markers.
363 * tree.c (make_node_stat): Don't set side effects for begin stmt
365 (build1_stat): Likewise.
366 * tree.def (DEBUG_BEGIN_STMT): New.
367 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
368 * var-tracking.c (delete_debug_insns): Renamed to...
369 (delete_vta_debug_insns): ... this.
370 (reemit_marker_as_note): New.
371 (vt_initialize): Reemit markers.
372 (delete_vta_debug_insns): Likewise.
373 (vt_debug_insns_local): Reemit or delete markers.
374 (variable_tracking_main_1): Likewise.
375 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
376 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
377 (gimple_debug_nonbind_marker_p): New.
378 (gimple_build_debug_bind): Adjust.
379 (gimple_build_debug_begin_stmt): New.
380 * doc/invoke.texi (max-debug-marker-count): New param.
381 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
382 (debug_parameter_ref, debug_marker): New.
383 (NOTE_INSN_BEGIN_STMT): New.
384 (DEBUG_INSN): Describe begin stmt markers.
386 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
387 without debug insns, we wouldn't, but clean up debug insns
388 after a control flow insn nevertheless.
390 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
391 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
392 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
393 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
394 (rtl_tidy_fallthru_edge): Likewise.
395 (rtl_verify_fallthru): Likewise.
396 (rtl_verify_bb_layout): Likewise.
397 (skip_insns_after_block): Likewise.
398 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
399 * dwarf2out.c: Include print-rtl.h.
400 (dwarf2out_next_real_insn): New.
401 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
402 Dump debug binds in asm comments.
403 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
404 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
405 callers to use gsi_start_nondebug_bb instead.
406 (gsi_after_labels): Skip gimple debug stmts.
407 (gsi_start_nondebug): New.
408 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
409 (proper_loop_form_for_interchange): Adjust.
410 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
411 * gimple.h (gimple_seq_last_nondebug_stmt): New.
412 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
413 (collect_fallthrough_labels): Likewise.
414 (should_warn_for_implicit_fallthrough): Likewise.
415 (warn_implicit_fallthrough_r): Likewise.
416 (expand_FALLTHROUGH_r): Likewise.
417 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
418 (graphite_copy_stmts_from_block): Skip nonbind markers.
419 * haifa-sched.c (sched_extend_bb): Skip debug insns.
420 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
421 * jump.c (clean_barriers): Skip debug insns.
422 * omp-expand.c (expand_parallel_call): Skip debug insns.
423 (expand_task_call): Likewise.
424 (remove_exit_barrier): Likewise.
425 (expand_omp_taskreg): Likewise.
426 (expand_omp_for_init_counts): Likewise.
427 (expand_omp_for_generic): Likewise.
428 (expand_omp_for_static_nochunk): Likewise.
429 (expand_omp_for_static_chunk): Likewise.
430 (expand_omp_simd): Likewise.
431 (expand_omp_taskloop_for_outer): Likewise.
432 (expand_omp_taskloop_for_inner): Likewise.
433 (expand_oacc_for): Likewise.
434 (expand_omp_sections): Likewise.
435 (expand_omp_single): Likewise.
436 (expand_omp_synch): Likewise.
437 (expand_omp_atomic_load): Likewise.
438 (expand_omp_atomic_store): Likewise.
439 (expand_omp_atomic_fetch_op): Likewise.
440 (expand_omp_atomic_pipeline): Likewise.
441 (expand_omp_atomic_mutex): Likewise.
442 (expand_omp_target): Likewise.
443 (grid_expand_omp_for_loop): Likewise.
444 (grid_expand_target_grid_body): Likewise.
445 (build_omp_regions_1): Likewise.
446 * omp-low.c (check_combined_parallel): Skip debug stmts.
447 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
448 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
449 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
451 * tree-cfg.c (make_blobs_1): Skip debug stmts.
452 (make_edges): Likewise.
453 (cleanup_dead_labels): Likewise.
454 (gimple_can_merge_blocks_p): Likewise.
455 (stmt_starts_bb_p): Likewise.
456 (gimple_block_label): Likewise.
457 (gimple_redirect_edge_and_branch): Likewise.
458 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
460 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
462 * tree-pretty-print (print_declaration): Omit initializer in slim
464 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
466 (eliminate_unnecessary_stmts): Stabilize block removal order.
467 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
468 * var-tracking.c (get_first_insn): New.
469 (vt_emit_notes): Call it.
470 (vt_initialize): Walk any insns before the first BB.
471 (delete_debug_insns): Likewise.
473 * gimple.h (enum gimple_debug_subcode): Add
474 GIMPLE_DEBUG_BEGIN_STMT.
475 (gimple_debug_begin_stmt_p): New.
476 (gimple_debug_nonbind_marker_p): New.
477 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
478 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
479 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
480 * insn-notes.def (BEGIN_STMT): New.
481 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
482 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
483 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
484 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
485 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
486 (INSN_DEBUG_MARKER_KIND): New.
487 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
488 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
489 (INSN_VAR_LOCATION_PTR): New.
490 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
492 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
493 deps only if debug bind insns are enabled.
494 (pass_expand::execute): Avoid deep TER and expand
495 debug locations for debug bind insns only.
496 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
497 debug stmts special handling down to debug bind stmts.
498 * combine.c (try_combine): Narrow debug insns special handling
499 down to debug bind insns.
500 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
501 Narrow debug insns preexisting special handling down to debug
503 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
504 handling down to debug bind insns.
505 * function.c (instantiate_virtual_regs): Skip debug markers,
506 adjust handling of debug binds.
507 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
508 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
509 * haifa-sched.c (schedule_insn): Narrow special handling of debug
510 insns to debug bind insns.
511 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
512 special handling of debug stmts to debug bind stmts.
513 * ipa-split.c (split_function): Likewise.
514 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
515 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
517 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
518 * regrename.c (build_def_use): Likewise.
519 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
520 (pass_cprop_hardreg): Narrow special casing of debug insns to
522 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
523 * reload1.c (reload): Likewise.
524 * sese.c (sese_insert_phis_for_liveouts): Narrow special
525 casing of debug stmts to debug bind stmts.
526 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
527 * ssa-iterators.h (num_imm_uses): Likewise.
528 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
529 debug stmts to debug bind stmts.
530 * tree-inline.c (tree_function_versioning): Narrow special casing
531 of debug stmts to debug bind stmts.
532 * tree-loop-distribution.c (generate_loops_for_partition):
533 Narrow special casing of debug stmts to debug bind stmts.
534 * tree-sra.c (analyze_access_subtree): Narrow special casing
535 of debug stmts to debug bind stmts.
536 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
537 stmts to debug bind stmts.
538 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
539 casing of debug stmts to debug bind stmts.
540 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
541 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
542 casing of debug stmts to debug bind stmts.
543 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
545 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
546 debug stmts to debug bind stmts.
547 (gimple_replace_ssa_lhs): Likewise.
548 (insert_debug_temp_for_var_def): Likewise.
549 (insert_debug_temps_for_defs): Likewise.
550 (reset_debug_uses): Likewise.
551 * tree-ssanames.c (release_ssa_name_fn): Likewise.
552 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
553 (adjust_debug_stmts): Likewise.
554 (adjust_phi_and_debug_stmts): Likewise.
555 (vect_do_peeling): Likewise.
556 * tree-vect-loop.c (vect_transform_loop): Likewise.
557 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
558 * var-tracking.c (adjust_mems): Narrow special casing of debug
559 insns to debug bind insns.
560 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
561 (compute_bb_dataflow, vt_find_locations): Likewise.
562 (vt_expand_loc, emit_notes_for_changes): Likewise.
563 (vt_init_cfa_base): Likewise.
564 (vt_emit_notes): Likewise.
565 (vt_initialize): Likewise.
566 (vt_finalize): Likewise.
568 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
569 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
570 (next_nonnote_nondebug_insn_bb): New.
571 (prev_nonnote_nondebug_insn_bb): New.
572 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
573 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
574 (prev_nonnote_nondebug_insn_bb): Declare.
575 (next_nonnote_nondebug_insn_bb): Declare.
576 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
577 * cfgrtl.c (get_last_bb_insn): Likewise.
578 * lra.c (push_insns): Likewise.
580 2017-12-11 David Malcolm <dmalcolm@redhat.com>
583 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
585 * selftest.c (selftest::test_runner::test_runner): New ctor.
586 (selftest::test_runner::~test_runner): New dtor.
587 * selftest.h (class selftest::test_runner): New class.
589 2017-12-11 Carl Love <cel@us.ibm.com>
591 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
592 vec_extract_fp32_from_shortl]): Add #defines.
593 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
594 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
595 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
596 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
597 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
598 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
599 * doc/extend.texi: Add documentation for the added builtins.
601 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
603 PR rtl-optimization/80693
604 PR rtl-optimization/81019
605 PR rtl-optimization/81020
606 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
607 are not mentioned in i3. Place the REG_UNUSED note on i2,
608 possibly modified to REG_DEAD, if it did not originate in i3.
610 2017-12-11 Jakub Jelinek <jakub@redhat.com>
612 * recog.c (store_data_bypass_p_1): New function.
613 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
614 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
616 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
618 PR rtl-optimization/83361
619 * ifcvt.c (if_convert): Call fixup_partitions.
621 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
623 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
624 early folding of splat_u{8,16,32}.
626 2017-12-11 Jakub Jelinek <jakub@redhat.com>
628 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
629 output_operand_lossage first argument with capital letter.
630 (aarch64_override_options): Don't start error and sorry first argument
633 2017-12-11 Andi Kleen <ak@linux.intel.com>
635 PR gcov-profile/83355
636 * auto-profile.c (string_table::get_index_by_decl): Don't
637 recurse when abstract origin points to itself.
639 2017-12-11 Bin Cheng <bin.cheng@arm.com>
641 PR tree-optimization/83320
642 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
643 (prune_datarefs_not_in_loop): Ditto.
645 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
647 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
649 2017-12-10 Jakub Jelinek <jakub@redhat.com>
651 PR tree-optimization/83337
652 * gimple-loop-interchange.cc (compute_access_stride): Handle
653 bitfield DRs properly.
655 2017-12-09 Jakub Jelinek <jakub@redhat.com>
657 PR tree-optimization/83338
658 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
659 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
660 vector of unsigned integers to vector of signed integers.
662 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
664 PR rtl-optimization/83317
665 * lra-constraints.c (process_address_1): Add insn code check.
667 2017-12-08 Michael Matz <matz@suse.de>
669 Fix PR tree-optimization/83323
670 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
671 head-controlled loops and loop BBs.
672 * common.opt (funroll-and-jam): Remove, instead ...
673 (floop-unroll-and-jam): ... reuse this option.
674 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
675 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
676 (-floop-unroll-and-jam): ... this option.
678 2017-12-08 Jakub Jelinek <jakub@redhat.com>
680 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
681 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
682 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
684 2017-12-08 Richard Biener <rguenther@suse.de>
687 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
688 handle accesses outside of zero-sized vars.
690 2017-12-08 Martin Jambor <mjambor@suse.cz>
692 PR tree-optimization/83141
693 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
694 test for MEM_REFs implicitely changing types with padding. Remove
696 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
698 2017-12-08 Jakub Jelinek <jakub@redhat.com>
700 * config/arc/arc.c (arc_attribute_table): Add exclusions to
702 * config/avr/avr.c (avr_attribute_table): Likewise.
703 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
704 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
705 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
706 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
707 of struct attribute_spec.
708 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
710 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
713 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
715 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
718 * profile-count.c (profile_count::from_gcov_type): Move from
719 profile-count.h; handle overflow.
720 * profile-count.h (profile_count::from_gcov_type): Move offline.
722 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
724 PR rtl-optimization/83304
725 * combine.c (move_deaths): If we do not know where a register died,
728 2017-12-08 Richard Biener <rguenther@suse.de>
730 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
731 Provide -fopt-info-loop feedback when we interchange in a nest.
733 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
735 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
736 for armv6 ARM CPU IDs.
738 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
740 * common/config/arm/arm-common.c: Include <algorithm>.
741 (INCLUDE_VECTOR): Define.
742 (compare_opt_names): New function.
743 (arm_rewrite_selected_arch): Only strip out extensions that can be
744 expressed through -mfpu. Sort the remaining extensions
747 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
749 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
750 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
751 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
752 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
754 2017-06-08 Tristan Gingold <gindold@adacore.com>
757 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
758 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
760 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
762 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
763 of struct attribute_spec.
765 2017-12-08 Julia Koval <julia.koval@intel.com>
767 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
768 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
769 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
770 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
771 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
772 _mm_maskz_dpwssds_epi32): Ditto.
774 2017-12-08 Richard Biener <rguenther@suse.de>
776 PR tree-optimization/81303
777 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
778 conditions try to create a comparison vector type matching
779 the data vector type.
780 (vectorizable_condition): Adjust.
781 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
782 Leave invariant conditions alone in case we can vectorize those.
784 2017-12-08 Julia Koval <julia.koval@intel.com>
786 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
787 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
788 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
789 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
790 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
792 2017-12-08 Julia Koval <julia.koval@intel.com>
794 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
795 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
796 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
797 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
798 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
799 _mm_maskz_dpbusds_epi32): New intrinsics.
801 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
803 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
805 (ld<bh>io_signed): Likewise.
806 (st<bhw>io): Likewise.
807 * config/nios2/predicates.md (ldstio_memory_operand): Allow
808 SMALL_INT12 constant integer operand.
810 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
812 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
814 (nios2_large_constant_allowed): ...this. Adjust uses.
815 (nios2_plus_symbolic_constant_p): Rename to...
816 (nios2_plus_large_constant_p): ...this. Adjust uses.
817 (nios2_legitimate_address_p): Correct CONST_INT handling.
818 (nios2_symbolic_memory_operand_p): Rename to...
819 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
820 (nios2_large_constant_p): Check for large constant integers too.
821 (nios2_split_large_constant): Handle constant integers.
822 (nios2_split_symbolic_memory_operand): Rename to...
823 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
824 (nios2_legitimize_constant_address): Handle constant integers.
825 (r0rel_constant_p): Handle small constant integers.
826 (nios2_print_operand_address): Handle r0-relative integer addresses.
827 * config/nios2/nios2-protos.h: Adjust for renamed functions.
828 * config/nios2/nios2.md: Adjust for renamed functions.
830 2017-12-07 Andrew Waterman <andrew@sifive.com>
832 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
833 (TARGET_HAVE_SRODATA_SECTION): New define.
834 (riscv_select_section): New function.
836 2017-12-08 Joseph Myers <joseph@codesourcery.com>
837 Alexander Monakov <amonakov@ispras.ru>
838 Jakub Jelinek <jakub@redhat.com>
841 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
843 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
845 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
846 a bool success value. Don't call output_operand_lossage here.
847 (aarch64_print_ldpstp_address): Return a bool success value.
848 (aarch64_print_operand_address): Call output_addr_const if
849 aarch64_print_address_internal fails.
850 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
851 'y'; call output_operand_lossage instead. Call output_operand_lossage
852 if aarch64_print_ldpstp_address fails.
854 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
856 * tree-vector-builder.h
857 (tree_vector_builder::binary_encoded_nelts): Declare.
858 * tree-vector-builder.c
859 (tree_vector_builder::binary_encoded_nelts): New function.
860 * fold-const.c (negate_expr_p): Likewise.
861 (operand_equal_p, fold_checksum_tree): Likewise.
862 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
863 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
864 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
865 (uniform_vector_p): Likewise.
866 * varasm.c (const_hash_1, compare_constant): Likewise.
867 * tree-ssa-ccp.c: Include tree-vector-builder.h.
868 (valid_lattice_transition): Operate directly on the VECTOR_CST
870 * ipa-icf.c: Include tree-vector-builder.h.
871 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
872 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
874 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
876 * tree.c (build_vector): Delete.
877 * tree.h (build_vector): Make static and move into the self-testing
880 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
882 * vector-builder.h (vector_builder::derived): New const overload.
883 (vector_builder::elt): New function.
884 * tree-vector-builder.h (tree_vector_builder::type): New function.
885 (tree_vector_builder::apply_step): Declare.
886 * tree-vector-builder.c (tree_vector_builder::apply_step): New
888 * gimple-fold.h (tree_vector_builder): Declare.
889 (gimple_build_vector): Take a tree_vector_builder instead of a
890 type and vector of elements.
891 * gimple-fold.c (gimple_build_vector): Likewise.
892 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
894 (get_initial_defs_for_reduction): Likewise.
895 (vectorizable_induction): Likewise.
897 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
899 * tree-vector-builder.h
900 (tree_vector_builder::new_binary_operation): Declare.
901 * tree-vector-builder.c
902 (tree_vector_builder::new_binary_operation): New function.
903 * fold-const.c (fold_relational_const): Use it.
904 (const_binop): Likewise. Check that both input vectors have
905 the same number of elements, thus excluding things like WIDEN_SUM.
906 Check whether it is possible to operate directly on the encodings
909 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
911 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
912 new_unary_operation, operating only on the encoded elements.
913 (const_unop): Likewise.
914 (exact_inverse): Likewise.
915 (distributes_over_addition_p): New function.
916 (const_binop): Use tree_vector_builder and new_unary_operation
917 for combinations of VECTOR_CST and INTEGER_CST. Operate only
918 on the encoded elements unless the encoding is strided and the
919 operation does not distribute over addition.
920 (fold_convert_const): Use tree_vector_builder and
921 new_unary_operation. Operate only on the encoded elements
922 for truncating integer conversions, or for non-stepped encodings.
924 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
926 * config/sparc/sparc.c: Include tree-vector-builder.h.
927 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
928 * expmed.c: Include tree-vector-builder.h.
929 (make_tree): Use tree_vector_builder instead of build_vector.
930 * fold-const.c: Include tree-vector-builder.h.
931 (const_binop): Use tree_vector_builder instead of build_vector.
932 (const_unop): Likewise.
933 (native_interpret_vector): Likewise.
934 (fold_vec_perm): Likewise.
935 (fold_ternary_loc): Likewise.
936 * gimple-fold.c: Include tree-vector-builder.h.
937 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
939 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
940 (simplify_vector_constructor): Use tree_vector_builder instead
942 * tree-vect-generic.c: Include tree-vector-builder.h.
943 (add_rshift): Use tree_vector_builder instead of build_vector.
944 (expand_vector_divmod): Likewise.
945 (optimize_vector_constructor): Likewise.
946 * tree-vect-loop.c: Include tree-vector-builder.h.
947 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
948 of build_vector. Explicitly use a stepped encoding for
950 * tree-vect-slp.c: Include tree-vector-builder.h.
951 (vect_get_constant_vectors): Use tree_vector_builder instead
953 (vect_transform_slp_perm_load): Likewise.
954 (vect_schedule_slp_instance): Likewise.
955 * tree-vect-stmts.c: Include tree-vector-builder.h.
956 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
957 (vect_gen_perm_mask_any): Likewise.
958 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
959 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
961 (build_vector_from_val): Likewise. Explicitly use a duplicate
964 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
966 * doc/generic.texi (VECTOR_CST): Describe new representation of
968 * vector-builder.h: New file.
969 * tree-vector-builder.h: Likewise.
970 * tree-vector-builder.c: Likewise.
971 * Makefile.in (OBJS): Add tree-vector-builder.o.
972 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
973 * tree-core.h (tree_base): Add a vector_cst field to the u union.
974 (tree_vector): Change the number of elements to
975 vector_cst_encoded_nelts.
976 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
977 (VECTOR_CST_ELTS): Delete.
978 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
979 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
980 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
981 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
982 (VECTOR_CST_ENCODED_ELT): Likewise.
983 (vector_cst_encoded_nelts): New function.
984 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
985 VECTOR_CST_NELTS_PER_PATTERN as arguments.
986 (vector_cst_int_elt, vector_cst_elt): Declare.
987 * tree.c: Include tree-vector-builder.h.
988 (tree_code_size): Abort if passed VECTOR_CST.
989 (tree_size): Update for new VECTOR_CST layout.
990 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
991 VECTOR_CST_NELTS_PER_PATTERN as arguments.
992 (build_vector): Use tree_vector_builder.
993 (vector_cst_int_elt, vector_cst_elt): New functions.
994 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
995 encoded elements and then create the vector in the canonical form.
996 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
997 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
998 (tree_c_tests): Call test_vector_cst_patterns.
999 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
1001 (hash_tree): Likewise.
1002 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
1003 (streamer_write_tree_header): Likewise.
1004 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
1005 (streamer_alloc_tree): Likewise. Update call to make_vector.
1006 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
1008 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1010 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
1011 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
1012 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
1014 2017-12-07 Bin Cheng <bin.cheng@arm.com>
1015 Richard Biener <rguenther@suse.de>
1017 PR tree-optimization/81303
1018 * Makefile.in (gimple-loop-interchange.o): New object file.
1019 * common.opt (floop-interchange): Reuse the option from graphite.
1020 * doc/invoke.texi (-floop-interchange): Ditto. New document for
1021 -floop-interchange and mention it for -O3.
1022 * opts.c (default_options_table): Enable -floop-interchange at -O3.
1023 * gimple-loop-interchange.cc: New file.
1024 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
1025 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
1026 * passes.def (pass_linterchange): New pass.
1027 * timevar.def (TV_LINTERCHANGE): New time var.
1028 * tree-pass.h (make_pass_linterchange): New declaration.
1029 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
1030 interchange. Record IV before/after increment in new parameters.
1031 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
1032 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
1034 (check_reduction_path): ...New function here.
1035 * tree-vectorizer.h (check_reduction_path): New declaration.
1037 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
1040 PR rtl-optimization/80818
1041 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
1042 always early clobbered.
1043 * lra-lives.c (process_bb_lives): Check input hard regs for early
1044 clobbered non-operand hard reg.
1046 2017-12-07 Jakub Jelinek <jakub@redhat.com>
1049 * tree-cfg.c (verify_gimple_assign_binary): Don't require
1050 types_compatible_p, just that TYPE_MODE is the same.
1052 2017-12-07 Martin Sebor <msebor@redhat.com>
1055 * attribs.c (empty_attribute_table): Initialize new member of
1056 struct attribute_spec.
1057 (decl_attributes): Add argument. Handle mutually exclusive
1058 combinations of attributes.
1059 (selftests::test_attribute_exclusions): New function.
1060 (selftests::attribute_c_tests): Ditto.
1061 * attribs.h (decl_attributes): Add default argument.
1062 * selftest.h (attribute_c_tests): Declare.
1063 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
1064 * tree-core.h (attribute_spec::exclusions, exclude): New type and
1066 * doc/extend.texi (Common Function Attributes): Update const and pure.
1067 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
1068 of struct attribute_spec.
1069 * config/arc/arc.c (arc_attribute_table): Same.
1070 * config/arm/arm.c (arm_attribute_table): Same.
1071 * config/avr/avr.c ( avr_attribute_table): Same.
1072 * config/bfin/bfin.c (bfin_attribute_table): Same.
1073 * config/cr16/cr16.c (cr16_attribute_table): Same.
1074 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
1075 * config/h8300/h8300.c (h8300_attribute_table): Same.
1076 * config/i386/i386.c (ix86_attribute_table): Same.
1077 * config/ia64/ia64.c (ia64_attribute_table): Same.
1078 * config/m32c/m32c.c (m32c_attribute_table): Same.
1079 * config/m32r/m32r.c (m32r_attribute_table): Same.
1080 * config/m68k/m68k.c (m68k_attribute_table): Same.
1081 * config/mcore/mcore.c (mcore_attribute_table): Same.
1082 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
1083 * config/mips/mips.c (mips_attribute_table): Same.
1084 * config/msp430/msp430.c (msp430_attribute_table): Same.
1085 * config/nds32/nds32.c (nds32_attribute_table): Same.
1086 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
1087 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
1088 * config/rl78/rl78.c (rl78__attribute_table): Same.
1089 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
1090 * onfig/rx/rx.c (rx_attribute_table): Same.
1091 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
1092 * config/sh/sh.c (sh_attribute_table): Same.
1093 * config/sparc/sparc.c (sparc_attribute_table): Same.
1094 * config/spu/spu.c (spu_attribute_table): Same.
1095 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
1096 * config/v850/v850.c (v850_attribute_table): Same.
1097 * config/visium/visium.c (visium_attribute_table): Same.
1099 2017-12-07 Tamar Christina <tamar.christina@arm.com>
1102 * config/arm/arm.c (INCLUDE_STRING): Define.
1103 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
1104 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
1107 2017-12-07 Michael Matz <matz@suse.de>
1109 Add unroll and jam pass
1111 * gimple-loop-jam.c: New file.
1112 * Makefile.in (OBJS): Add gimple-loop-jam.o.
1113 * common.opt (funroll-and-jam): New option.
1114 * opts.c (default_options_table): Add unroll-and-jam at -O3.
1115 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
1116 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
1117 * passes.def: Add pass_loop_jam.
1118 * timevar.def (TV_LOOP_JAM): Add.
1119 * tree-pass.h (make_pass_loop_jam): Declare.
1120 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
1121 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
1122 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
1123 to flow_loop_tree_node_add.
1124 (duplicate_subloops, copy_loops_to): Append to sibling list.
1125 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
1126 * doc/invoke.texi (-funroll-and-jam): Document new option.
1127 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
1129 2017-12-07 Richard Biener <rguenther@suse.de>
1131 PR tree-optimization/83296
1132 PR tree-optimization/67769
1133 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
1134 flow sensitive info in an unrelated BB.
1135 (value_replacement): Use reset_flow_sensitive_info.
1136 (minmax_replacement): Reset flow sensitive info on the def
1137 we move. Do not reset flow sensitive info in the whole BB
1138 we move the stmt to.
1139 (abs_replacement): Likewise.
1141 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1144 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1145 rs6000_cpu to the given -mcpu=, or to the default processor.
1147 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1149 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
1150 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
1151 instead of rs6000_cpu_attr.
1152 (rs6000_adjust_cost): Ditto.
1153 (is_microcoded_insn): Ditto.
1154 (rs6000_adjust_priority): Ditto.
1155 (rs6000_issue_rate): Ditto.
1156 (rs6000_use_sched_lookahead): Ditto.
1157 (rs6000_use_sched_lookahead_guard): Ditto.
1158 (rs6000_sched_reorder): Ditto.
1159 (force_new_group): Ditto.
1160 * config/rs6000/rs6000.md (cpu attribute): Ditto.
1161 (group_ending_nop): Ditto.
1163 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1165 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
1166 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
1167 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
1168 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
1170 (rs6000_reassociation_width): Ditto.
1171 (rs6000_emit_epilogue): Ditto.
1172 (rs6000_adjust_cost): Ditto.
1173 (is_microcoded_insn): Ditto.
1174 (is_cracked_insn): Ditto.
1175 (rs6000_adjust_priority): Ditto.
1176 (rs6000_sched_reorder): Ditto.
1177 (rs6000_sched_reorder2): Ditto.
1178 (insn_must_be_first_in_group): Ditto.
1179 (insn_must_be_last_in_group): Ditto.
1180 (rs6000_register_move_cost): Ditto.
1181 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
1184 2017-12-07 Julia Koval <julia.koval@intel.com>
1186 * config.gcc: Add vaesintrin.h.
1187 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
1189 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
1190 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
1192 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1193 * config/i386/immintrin.h: Include vaesintrin.h.
1194 * config/i386/sse.md (vaesdec_<mode>): New pattern.
1195 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
1196 _mm_aesdec_epi128): New intrinsics.
1198 2017-12-06 David Malcolm <dmalcolm@redhat.com>
1200 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
1201 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
1202 c-family/c-spellcheck.cc.
1203 (best_macro_match::best_macro_match): Likewise.
1205 (struct edit_distance_traits<cpp_hashnode *>): Move to
1206 c-family/c-spellcheck.h.
1207 (class best_macro_match): Likewise.
1209 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1211 PR tree-optimization/83293
1212 * gimple-ssa-strength-reduction.c (insert_initializers): Use
1213 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
1214 might insert into empty bb.
1217 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
1218 simplify for plus with :c added, and pointer_plus without that.
1219 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
1220 with undefined overflow and the conversion is not widening,
1221 perform negation in utype and only convert to type afterwards.
1222 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
1223 simplify for plus with :c added, and pointer_plus without that.
1224 If type is integral with undefined overflow and the conversion is
1225 not widening, perform minus in utype and only convert to type
1226 afterwards. Move the last pointer_diff_expr simplify into the
1229 2017-12-06 Martin Sebor <msebor@redhat.com>
1231 PR tree-optimization/82646
1232 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
1233 strncpy, not maxlen.
1235 2017-12-06 Martin Sebor <msebor@redhat.com>
1237 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
1240 PR tree-optimization/83075
1241 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
1242 strncat/strncpy don't change length of source string.
1244 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
1247 2017-11-29 Martin Aberg <maberg@gaisler.com>
1249 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1250 to prevent b2bst errata sequence.
1251 (sqrtdf2_fix): Likewise.
1253 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1255 PR tree-optimization/81945
1256 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
1257 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
1258 to dest_cfun has orig_loop_num set, either remap it to the new
1259 loop number if the loop got moved too, or clear it.
1261 2017-12-05 Steve Ellcey <sellcey@cavium.com>
1263 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
1265 (thunderx2t99_nothing): New insn reservation.
1266 (thunderx2t99_mrs): New insn reservation.
1267 (thunderx2t99_multiple): New insn reservation.
1268 (thunderx2t99_alu_basi): Add bfx to reservation.
1269 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
1271 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1274 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
1276 2017-12-05 Bin Cheng <bin.cheng@arm.com>
1278 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
1279 tree-ssa-pre.c::remove_dead_inserted_code.
1280 * tree-ssa-dce.h: New file.
1281 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
1282 (remove_dead_inserted_code): Move and rename to function
1283 tree-ssa-dce.c::simple_dce_from_worklist.
1284 (pass_pre::execute): Update use.
1286 2017-12-05 Richard Biener <rguenther@suse.de>
1288 PR tree-optimization/83277
1289 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
1290 to code-gen liveout vars.
1292 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
1294 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
1295 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
1296 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
1299 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
1301 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
1303 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
1304 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
1307 2017-12-05 Richard Biener <rguenther@suse.de>
1309 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
1310 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
1311 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
1312 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
1314 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1316 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
1318 2017-12-05 Martin Liska <mliska@suse.cz>
1319 Jakub Jelinek <jakub@redhat.com>
1321 * doc/invoke.texi: Document the options.
1322 * flag-types.h (enum sanitize_code): Add
1323 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1324 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
1325 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1326 * opts.c: Define new sanitizer options.
1327 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
1328 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
1330 2017-12-05 Julia Koval <julia.koval@intel.com>
1332 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
1333 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
1334 (ix86_handle_option): Handle -mavx512vnni.
1335 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
1336 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
1337 * config/i386/i386-c (__AVX512VNNI__): New.
1338 * config/i386/i386.c (ix86_target_string): Handle new option.
1339 (ix86_valid_target_attribute_inner_p): Handle new option.
1340 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
1341 * config/i386/i386.opt (mavx512vnni): New option.
1343 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1346 * athlon.md: Disable for generic.
1347 * haswell.md: Enable for generic.
1348 * i386.c (ix86_sched_init_global): Add core hooks for generic.
1349 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
1351 (ix86_adjust_cost): Move generic to haswell path.
1353 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
1355 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
1356 instead of MEM_P in a couple more places. Fix formatting issues.
1358 2017-12-04 Jim Wilson <jimw@sifive.com>
1360 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
1361 instead of GP_REG_LAST-1.
1362 (riscv_adjust_libcall_cfi_prologue): Likewise.
1363 (riscv_adjust_libcall_cri_epilogue): Likewise.
1364 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
1367 2017-12-04 Luis Machado <luis.machado@linaro.org>
1369 * ipa-pure-const.c (check_decl): Add missing newline.
1370 (state_from_flags): Likewise.
1372 2017-12-04 Jeff Law <law@redhat.com>
1374 PR tree-optimizatin/78496
1375 * gimple-ssa-evrp-analyze.h
1376 (evrp_range_analyzer::get_vr_values): Simplify.
1377 * gimple-ssa-evrp-analyze.c: Corresponding changes.
1378 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
1379 and gimple-ssa-evrp-analyze.h.
1380 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
1381 (simplify_stmt_for_jump_threading): Copy a blob of code from
1382 tree-vrp.c to use ranges to simplify statements.
1383 (dom_opt_dom_walker::before_dom_children): Call
1384 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
1385 (dom_opt_dom_walker::after_dom_children): Similarly for
1386 evrp_range_analyzer::leave.
1387 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
1390 * gimple-ssa-evrp-analyze.c
1391 (evrp_range_analyzer::extract_range_from_stmt): Always use
1392 vr_values::update_value_range so preexisting range info is
1393 medged with new range info, even if the new range is VR_VARYING.
1395 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1397 * combine.c: Adjust comment.
1398 (use_crosses_set_p): Delete.
1399 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
1400 (try_combine): Ditto.
1402 2017-12-04 Richard Biener <rguenther@suse.de>
1404 PR tree-optimization/83255
1405 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
1406 Re-add zero-iteration check.
1408 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1410 PR rtl-optimization/83245
1411 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
1412 hard registers as earlyclobber, also if not in an asm.
1414 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1418 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1421 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1422 rs6000_cpu based on cpu_index, not tune_index.
1424 2017-12-04 Richard Biener <rguenther@suse.de>
1426 PR tree-optimization/83238
1427 * graphite-scop-detection.c (scop_detection::merge_sese): Make
1428 code match comment, rejecting invalid SESE regions.
1430 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
1432 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
1433 require base operand is a REG_POINTER prior to reload on targets
1434 with non-equivalent space registers.
1436 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1438 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
1439 (update_specialized_profile): Fix updating of counts.
1440 (perhaps_add_new_callers): Likewise.
1442 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1445 * x86-tune.def: Remove obsolette FIXMEs.
1446 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
1447 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
1448 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
1450 (X86_TUNE_PAD_RETURNS): Disable for generic.
1452 2017-12-02 Jakub Jelinek <jakub@redhat.com>
1454 PR tree-optimization/83170
1455 PR tree-optimization/83241
1456 * gimple-ssa-store-merging.c
1457 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
1458 gimple_vuse (ins_stmt) in case it has changed.
1459 (imm_store_chain_info::output_merged_store): Likewise.
1461 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
1465 * tree-cfg.c (pass_warn_function_return::execute): Handle
1466 __builtin_ubsan_handle_missing_return like __builtin_unreachable
1467 with BUILTINS_LOCATION.
1471 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
1472 is BLKmode for vector field with vector raw mode, use TYPE_MODE
1473 instead of DECL_MODE.
1475 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
1476 last argument to rtx pointer.
1477 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
1478 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
1479 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
1480 depending on the chosen ISAs.
1481 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
1482 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
1483 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
1485 * config/i386/sse.md (mov<mode>_internal): Likewise.
1486 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
1488 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1490 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
1492 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1494 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
1495 parameters from prototype.
1496 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
1497 parameters. Don't print a comment.
1498 (emit_fusion_gpr_load): Adjust.
1499 (emit_fusion_load_store): Adjust.
1500 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
1501 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
1502 comment on the second line.
1504 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1507 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1508 rs6000_cpu based on cpu_index, not tune_index.
1510 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1512 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
1513 which_alternative instead of which_alternative + 1.
1514 (output_asm_insn): Print an extra tab if the template is short.
1516 2017-12-01 Jim Wilson <jimw@sifive.com>
1518 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
1520 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
1521 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
1523 * doc/tm.texi: Regenerate.
1525 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
1528 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
1529 whether we can allocate pseudos before trying to fix an address.
1530 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
1531 memory address is indexed or indirect.
1532 (floatuns_<mode>si2_hw2): Likewise.
1534 2017-12-01 Jason Merrill <jason@redhat.com>
1536 * Makefile.in (TAGS): Add c-family/*.cc.
1538 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
1540 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
1541 (call_value_insn): Likewise.
1542 (sibcall_insn): Likewise.
1543 (sibcall_value_insn): Likewise.
1544 (movsi_aarch64): Likewise.
1545 (movdi_aarch64): Likewise.
1546 (add_losym_): Likewise.
1547 (ldr_got_small_): Likewise.
1548 (ldr_got_small_sidi): Likewise.
1549 (ldr_got_small_28k_): Likewise.
1550 (ldr_got_small_28k_sidi): Likewise.
1551 * config/aarch64/aarch64.c (aarch64_print_address_internal):
1552 Move output_addr_const to symbolic case. Add error check.
1554 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1556 * config/s390/predicates.md (plus16_Q_operand): New predicate.
1557 * config/s390/s390.md: Disable MVC merging peephole if it would
1558 disable operand forwarding.
1559 (new peephole2): Split MVCs if it would turn them into up to 2
1562 2017-12-01 Richard Biener <rguenther@suse.de>
1564 PR tree-optimization/83232
1565 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
1566 detection of same access. Instead of breaking the group here
1567 do not consider the duplicate. Add comment explaining real fix.
1569 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
1571 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
1573 2017-12-01 Sudakshina Das <sudi.das@arm.com>
1575 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
1577 2017-12-01 Jakub Jelinek <jakub@redhat.com>
1579 * function.h (struct function): Remove cilk_frame_decl,
1580 is_cilk_function and calls_cilk_spawn fields.
1581 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
1583 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
1584 cilk_elemental field.
1585 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
1586 * target.def: Adjust comment.
1587 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
1588 Don't test cilk_elemental.
1590 PR tree-optimization/83233
1591 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
1592 bswap_stat name for the struct.
1595 * tree.h (SWITCH_BREAK_LABEL_P): Define.
1596 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
1597 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
1598 SWITCH_BREAK_LABEL_P set on the label.
1599 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
1600 added for default case if it was missing and not all cases covered.
1601 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
1602 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
1604 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
1605 to avoid -Wimplicit-fallthrough warning.
1606 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
1607 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
1609 PR tree-optimization/83221
1610 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
1612 (init_reassoc): Formatting fix.
1615 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
1616 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
1620 * tree-cfg.c: Include asan.h.
1621 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
1623 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1625 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
1627 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1629 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
1630 vpcmpeqd instruction.
1632 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1634 * config/i386/i386.c (standard_sse_constant_opcode): Fix
1635 registers type for 128bit mode.
1637 2017-11-30 Jakub Jelinek <jakub@redhat.com>
1639 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
1641 * gengtype-state.c (read_state_pair): Likewise.
1642 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
1643 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
1644 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
1645 * ifcvt.c (noce_try_store_flag_constants): Likewise.
1646 * tree-ssa-ccp.c (ccp_finalize): Likewise.
1647 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
1648 * builtins.c (fold_builtin_3): Likewise.
1649 * graphite-scop-detection.c
1650 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
1651 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
1654 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
1657 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
1658 parameter. Return true if ignore_decl_rtl_set_p is true and other
1659 conditions are satisfied.
1660 * asan.h (asan_protect_global): Add new parameter.
1661 * varasm.c (categorize_decl_for_section): Pass true as second parameter
1662 to asan_protect_global calls.
1664 2017-11-30 Jim Wilson <jimw@sifive.com>
1666 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
1667 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
1668 -mstrict-align, add info on default value. Delete redundant lines for
1669 -mabi. Add missing -mexplicit-relocs docs.
1671 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1673 * config/arc/arc.md (trap): New pattern.
1675 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1677 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
1678 instruction to end into a delay slot.
1679 * config/arc/arc.md (cond_delay_insn): Check if the instruction
1680 can be placed into a delay slot against reg_note.
1682 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1684 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
1685 labels number of usages.
1687 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1689 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
1691 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
1693 2017-11-30 Jakub Jelinek <jakub@redhat.com>
1696 * internal-fn.c (expand_mul_overflow): Optimize unsigned
1697 multiplication by power of 2 constant into two shifts + comparison.
1699 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
1702 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
1704 2017-11-30 Richard Biener <rguenther@suse.de>
1706 PR tree-optimization/83202
1707 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
1708 allow_peel argument and guard peeling.
1709 (canonicalize_loop_induction_variables): Likewise.
1710 (canonicalize_induction_variables): Pass false.
1711 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
1712 peeling from cunrolli.
1714 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1716 * combine.c (try_combine): Print a message to dump file whenever
1717 I0, I1, or I2 cannot be combined into I3.
1719 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1721 PR rtl-optimization/83156
1722 PR rtl-optimization/82621
1723 * combine.c (try_combine): Don't split an I2 if one of the dests is
1724 set again before I3. Allow unused dests.
1726 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1728 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
1730 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
1732 PR rtl-optimization/80818
1733 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
1734 recursively. Use insn code for clobber.
1735 (lra_set_insn_recog_data): Pass the new arg to
1736 collect_non_operand_hard_regs.
1737 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
1739 (lra_update_insn_regno_info): Pass insn to
1740 add_regs_to_insn_regno_info.
1742 2017-11-29 Jim Wilson <jimw@sifive.com>
1743 Andrew Waterman <andrew@sifive.com>
1745 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
1746 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
1747 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
1748 use SINGLE_SHIFT_COST.
1749 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
1750 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
1752 2017-11-29 Julia Koval <julia.koval@intel.com>
1754 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
1755 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
1756 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
1757 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
1758 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
1759 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
1760 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
1761 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
1762 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
1763 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
1764 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
1765 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
1766 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
1767 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
1768 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
1769 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
1770 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
1771 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
1772 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
1773 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
1774 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
1775 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
1776 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
1777 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
1778 __builtin_ia32_vpshldv_v2di_mask,
1779 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
1780 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
1781 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
1783 2017-11-29 Julia Koval <julia.koval@intel.com>
1785 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
1786 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
1787 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
1788 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
1789 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
1790 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
1791 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
1792 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
1793 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
1794 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
1795 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
1796 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
1797 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
1798 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
1799 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
1800 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
1801 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
1802 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
1803 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1804 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
1805 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
1807 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1809 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
1810 movsi_pic_gotdata_op instruction as a load for the UT699 errata
1813 2017-11-29 Martin Aberg <maberg@gaisler.com>
1815 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1816 to prevent b2bst errata sequence.
1817 (sqrtdf2_fix): Likewise.
1819 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1821 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
1822 (div_sqrt_insn_p): New function.
1823 (sparc_do_work_around_errata): Insert NOP instructions to
1824 prevent sequences that could trigger the TN-0013 errata for
1825 certain LEON3 processors.
1826 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
1827 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
1828 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
1829 (in_branch_delay): Prevent div and sqrt in delay slot if
1831 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
1833 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1835 * config/sparc/sparc.c (atomic_insn_p): New function.
1836 (sparc_do_work_around_errata): Insert NOP instructions to
1837 prevent sequences that could trigger the TN-0010 errata for
1839 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
1840 instruction referable in atomic_insns_p.
1842 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1844 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
1845 (atomic_compare_and_swap_leon3_1): Likewise.
1848 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1850 * config/sparc/sparc.c (fpop_insn_p): New function.
1851 (sparc_do_work_around_errata): Insert NOP instructions to
1852 prevent sequences that could trigger the TN-0012 errata for
1854 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
1855 * config/sparc/sparc.md (fix_gr712rc): New attribute.
1856 (in_branch_annul_delay): Prevent floating-point instructions
1857 in delay slot of annulled integer branch.
1859 2017-11-29 Richard Biener <rguenther@suse.de>
1861 PR tree-optimization/83202
1862 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
1864 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
1865 nodes vectorized to the same stmts multiple times.
1866 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
1867 (vect_analyze_slp_instance): Adjust.
1868 (scalar_stmts_to_slp_tree_map_t): New typedef.
1869 (vect_schedule_slp_instance): Add a map recording the SLP node
1870 representing the vectorized stmts for a set of scalar stmts.
1871 Avoid code-generating redundancies.
1872 (vect_schedule_slp): Allocate map and pass it down.
1874 2017-11-29 Nathan Sidwell <nathan@acm.org>
1877 * tree.c (build_complex_type): Fix canonicalization. Only fill in
1880 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
1882 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
1883 cases for printing LDP/STP memory addresses.
1884 (aarch64_print_address_internal): Renamed from
1885 aarch64_print_operand_address, added parameter, add Pmode check.
1886 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
1887 (aarch64_print_operand_address): Indirect to
1888 aarch64_print_address_internal.
1889 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
1890 'y' operand output specifier.
1892 2017-11-29 Jakub Jelinek <jakub@redhat.com>
1895 * tree.c (build_simple_mem_ref_loc): Handle
1896 get_addr_base_and_unit_offset returning a MEM_REF.
1899 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
1902 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
1905 2017-11-28 David Malcolm <dmalcolm@redhat.com>
1907 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
1909 (layout::move_to_column): Likewise.
1911 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1914 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
1915 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
1917 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1918 Martin Jambor <mjambor@suse.cz>
1921 * tree.h (expr_type_first_operand_type_p): Declare
1922 * tree.c (expr_type_first_operand_type_p): New function.
1923 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
1924 (ipa_value_from_jfunc): Adjust declaration.
1925 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
1926 Use it as result type for arithmetics, unless it is NULL in which case
1927 be more conservative.
1928 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
1929 ipa_get_jf_pass_through_result.
1930 (propagate_vals_across_pass_through): Likewise.
1931 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
1932 is to propagate_vals_across_pass_through.
1933 (propagate_constants_across_call): Pass PARM_TYPE to
1934 propagate_scalar_across_jump_function.
1935 (find_more_scalar_values_for_callers_subset): Pass parameter type to
1936 ipa_value_from_jfunc.
1937 (cgraph_edge_brings_all_scalars_for_node): Likewise.
1938 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
1939 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
1940 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
1941 target_type, pass it to ipa_value_from_jfunc.
1942 (update_indirect_edges_after_inlining): Pass parameter type to
1943 try_make_edge_direct_simple_call.
1945 2017-11-28 Jeff Law <law@redhat.com>
1947 * gimple-ssa-evrp-analyze.c
1948 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
1949 refine ranges if scev_initialized_p returns true.
1950 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
1952 2017-11-28 Julia Koval <julia.koval@intel.com>
1954 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
1955 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
1956 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
1957 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
1958 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
1959 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
1960 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
1961 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
1962 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
1963 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
1964 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
1965 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
1966 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
1967 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
1968 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
1969 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
1970 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
1971 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
1972 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
1973 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
1974 __builtin_ia32_vpshrd_v2di_mask): New builtins.
1975 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
1977 2017-11-28 Julia Koval <julia.koval@intel.com>
1979 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
1980 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
1981 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
1982 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
1983 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
1984 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
1985 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
1986 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
1987 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
1988 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
1989 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
1990 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
1991 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
1992 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
1993 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
1994 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
1995 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
1996 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
1997 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
1998 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
1999 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
2000 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
2001 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
2002 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
2003 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
2004 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
2005 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
2006 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
2007 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
2008 __builtin_ia32_vpshld_v2di_mask): New builtins.
2009 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2010 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
2012 2017-11-28 Richard Biener <rguenther@suse.de>
2014 PR tree-optimization/80776
2015 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
2017 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
2019 (evrp_range_analyzer::record_ranges_from_incoming_edges):
2020 If the incoming edge is an effective fallthru because the other
2021 edge only reaches a __builtin_unreachable () then record ranges
2022 derived from the controlling condition in SSA info.
2023 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
2024 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
2026 2017-11-28 Olivier Hainque <hainque@adacore.com>
2028 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
2029 on mingw build hosts.
2031 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2034 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
2036 2017-11-28 Richard Biener <rguenther@suse.de>
2039 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
2040 copies generated from memcpy use a character array as reference
2043 2017-11-28 Julia Koval <julia.koval@intel.com>
2044 Sebastian Peryt <sebastian.peryt@intel.com>
2046 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
2047 c-family/c-cilkplus.o, c-family/array-notation-common.o,
2048 cilk-common.o, cilk.h, cilk-common.c): Remove.
2050 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
2051 * builtins.c (is_builtin_name): Remove cilkplus condition.
2052 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
2053 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
2054 cilk-builtins.def, cilkplus.def): Remove.
2055 * cif-code.def (CILK_SPAWN): Remove.
2056 * cilk-builtins.def: Delete.
2057 * cilk-common.c: Ditto.
2059 * cilkplus.def: Ditto.
2060 * config/darwin.h (fcilkplus): Delete.
2061 * cppbuiltin.c: Ditto.
2062 * doc/extend.texi: Remove cilkplus doc.
2063 * doc/generic.texi: Ditto.
2064 * doc/invoke.texi: Ditto.
2065 * doc/passes.texi: Ditto.
2066 * gcc.c (fcilkplus): Remove.
2067 * gengtype.c (cilk.h): Remove.
2068 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
2070 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
2072 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
2074 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
2075 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
2076 cilkplus conditions.
2077 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
2078 inline_read_section): Ditto.
2079 * ipa-inline-analysis.c (cilk.h): Remove.
2080 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
2081 * lto-wrapper.c (merge_and_complain, append_compiler_options,
2082 append_linker_options): Remove condition for fcilkplus.
2083 * lto/lto-lang.c (cilk.h): Remove.
2084 (lto_init): Remove condition for fcilkplus.
2085 * omp-expand.c (expand_cilk_for_call): Delete.
2086 (expand_omp_taskreg, expand_omp_for_static_chunk,
2087 expand_omp_for): Remove cilkplus
2089 (expand_cilk_for): Delete.
2090 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
2091 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
2092 execute_lower_omp, diagnose_sb_0): Ditto.
2093 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
2094 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2095 * tree-nested.c: Ditto.
2096 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
2097 (dump_generic_node): Ditto.
2098 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2099 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
2101 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
2103 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2105 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
2106 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
2107 for vector mode and !TARGET_SIMD.
2109 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2111 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
2113 * tree.h (SWITCH_LABELS): Remove.
2114 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
2115 assert SWITCH_BODY is non-NULL.
2116 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
2118 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
2120 PR tree-optimization/80788
2121 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
2122 has TREE_OVERFLOW set, call drop_tree_overflow.
2124 2017-11-28 Richard Biener <rguenther@suse.de>
2126 PR tree-optimization/83158
2127 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
2129 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
2132 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
2133 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
2135 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
2137 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
2138 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
2140 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2143 * varasm.c (compare_constant): Take the mode of the constants into
2144 account when comparing floating point constants.
2146 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
2148 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
2149 from explicit instantiation of debug_helper.
2150 * vec.h (DEFINE_DEBUG_VEC): Ditto.
2152 2017-11-27 Richard Biener <rguenther@suse.de>
2154 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
2157 2017-11-27 Richard Biener <rguenther@suse.de>
2159 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
2160 (type_hash_canon): Also clear int_cst_hash_table entry for
2162 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
2164 2017-11-27 Tamar Christina <tamar.christina@arm.com>
2166 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2167 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2168 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2170 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2172 * hash-map.h (gt_cleare_cache): Avoid UB.
2174 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
2176 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
2177 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
2178 (decide_unroll_runtime_iterations): Likewise.
2179 (decide_unroll_stupid): Likewise.
2181 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2184 * config/i386/i386.md: Add a loop with incssp.
2186 2017-11-27 Martin Jambor <mjambor@suse.cz>
2188 PR tree-optimization/81248
2189 * tree-sra.c (splice_param_accesses): Remove size check.
2190 (decide_one_param_reduction): Fix size check.
2191 * gimple-pretty-print.c (dump_profile): Silence warning.
2192 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
2194 2017-11-27 Jakub Jelinek <jakub@redhat.com>
2197 * dbxout.c (lastlineno): New variable.
2198 (dbx_debug_hooks): Use dbxout_switch_text_section as
2199 switch_text_section debug hook.
2200 (dbxout_function_end): Switch to current_function_section
2201 rather than function_section. If crtl->has_bb_partition,
2202 output just one N_FUN, depending on in_cold_section_p.
2203 (dbxout_source_line): Remember last lineno in lastlineno.
2204 (dbxout_switch_text_section): New function.
2205 (dbxout_function_decl): Adjust dbxout_block caller.
2206 (dbx_block_with_cold_children): New function.
2207 (dbxout_block): Return true if any LBRAC/RBRAC have been
2208 emitted. Use dbx_block_with_cold_children at depth == 0
2209 in second partition. Add PARENT_BLOCKNUM argument, pass
2210 it optionally adjusted to children. Output LBRAC/RBRAC
2211 around recursive call only if the block is in the current
2212 partition, if not and anything was output, emit empty
2214 * final.c (final_scan_insn): Compute cold_function_name
2215 before calling switch_text_section debug hook. Call
2216 that hook even if dwarf2out_do_frame if not emitting
2220 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
2221 TREE_READONLY decls.
2223 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2225 PR rtl-optimization/82488
2226 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
2228 2017-11-26 Julia Koval <julia.koval@intel.com>
2230 * config/i386/i386.c (processor_target_table): Add skylake_cost for
2232 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
2235 2017-11-26 Julia Koval <julia.koval@intel.com>
2237 * config/i386/driver-i386.c (host_detect_local_cpu):
2238 Detect skylake-avx512.
2240 2017-11-26 Julia Koval <julia.koval@intel.com>
2242 * config.gcc: Add -march=cannonlake.
2243 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
2244 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
2245 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
2246 (PTA_CANNONLAKE): New.
2247 (processor_target_table): Add cannonlake.
2248 (ix86_option_override_internal): Ditto.
2249 (fold_builtin_cpu): Ditto.
2250 (get_builtin_code_for_version): Handle cannonlake.
2251 (M_INTEL_COREI7_CANNONLAKE): New.
2252 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
2253 * doc/invoke.texi: Add -march=cannonlake.
2255 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
2257 * plugin.c (add_new_plugin): Use platform-specific library extensions.
2258 (try_init_one_plugin): Alternative implementation for MinGW.
2259 * Makefile.in (plugin_implib): New.
2260 (gengtype-lex.c): Fix broken AIX workaround.
2261 * configure: Regenerate.
2262 * doc/plugins.texi: Document support for MinGW.
2264 2017-11-25 Jakub Jelinek <jakub@redhat.com>
2266 PR rtl-optimization/81553
2267 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
2268 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
2269 is a shift where C1 has different mode than the whole shift, use C1's
2270 mode for MULT rather than the shift's mode.
2273 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
2274 builtins not enabled in the currently selected ISA.
2276 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
2278 PR tree-optimization/71026
2279 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
2280 (insert_reciprocals): Change to insert reciprocals before a division
2281 by a square and to insert the square of a reciprocal.
2282 (execute_cse_reciprocals_1): Change to consider division by a square.
2283 (register_division_in): Add importance parameter.
2285 2017-11-24 Richard Biener <rguenther@suse.de>
2287 PR tree-optimization/82402
2288 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
2289 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
2291 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
2293 * match.pd (0-ptr): New transformation.
2295 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
2298 * ipa-inline.c (inline_small_functions): Set current badnes correctly
2299 when skipping checking.
2301 2017-11-24 Richard Biener <rguenther@suse.de>
2303 PR tree-optimization/83128
2304 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
2305 (vn_reference_lookup_3): Likewise.
2307 2017-11-24 Jakub Jelinek <jakub@redhat.com>
2310 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
2311 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
2314 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
2315 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
2316 check for the latter with gimple_call_builtin_p. Do not handle
2317 BUILT_IN_STPNCPY_CHK which is not a pass through call.
2319 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
2321 * config/arm/arm_neon.h: Fix pragma GCC push_options before
2324 2017-11-23 Julia Koval <julia.koval@intel.com>
2326 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
2327 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
2328 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
2329 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
2330 _mm512_maskz_expandloadu_epi16): New intrinsics.
2331 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
2332 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
2333 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
2334 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
2335 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
2336 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
2337 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
2338 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
2339 _mm256_maskz_expandloadu_epi8): New intrinsics.
2340 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
2341 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
2342 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
2343 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
2344 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
2345 * config/i386/sse.md (VI248_VLBW): New iterator.
2346 (expand<mode>_mask, expand<mode>_maskz): New patterns.
2348 2017-11-23 Julia Koval <julia.koval@intel.com>
2350 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
2351 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
2352 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
2353 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
2354 _mm512_mask_compressstoreu_epi16): New.
2355 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
2356 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
2357 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
2358 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
2359 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
2360 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
2361 _mm256_mask_compressstoreu_epi8): New.
2362 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
2363 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
2364 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
2365 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
2366 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
2367 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
2368 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
2369 __builtin_ia32_compresshi128_mask,
2370 __builtin_ia32_compressstoreuqi512_mask,
2371 __builtin_ia32_compressstoreuhi512_mask,
2372 __builtin_ia32_compressstoreuqi256_mask,
2373 __builtin_ia32_compressstoreuqi128_mask,
2374 __builtin_ia32_compressstoreuhi256_mask,
2375 __builtin_ia32_compressstoreuhi128_mask): New builtins.
2376 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
2378 (ix86_expand_special_args_builtin): Handle new types.
2379 (s4fma_expand): Handle new builtin array.
2380 * config/i386/immintrin.h: Include new headers.
2381 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
2382 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
2384 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2386 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
2387 fixes. Declare temp and g variables at the top in order to avoid
2388 {} in most of the cases.
2390 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
2392 * match.pd (ptr-0): New transformation.
2394 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2396 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
2397 (struct addr_mode_cost_table): New.
2398 (struct tune_params): Add field addr_mode_costs.
2399 * config/arm/arm.c (generic_addr_mode_costs): New.
2400 (arm_slowmul_tune): Initialise addr_mode_costs field.
2401 (arm_fastmul_tune): Likewise.
2402 (arm_strongarm_tune): Likewise.
2403 (arm_xscale_tune): Likewise.
2404 (arm_9e_tune): Likewise.
2405 (arm_marvell_pj4_tune): Likewise.
2406 (arm_v6t2_tune): Likewise.
2407 (arm_cortex_tune): Likewise.
2408 (arm_cortex_a8_tune): Likewise.
2409 (arm_cortex_a7_tune): Likewise.
2410 (arm_cortex_a15_tune): Likewise.
2411 (arm_cortex_a35_tune): Likewise.
2412 (arm_cortex_a53_tune): Likewise.
2413 (arm_cortex_a57_tune): Likewise.
2414 (arm_exynosm1_tune): Likewise.
2415 (arm_xgene1_tune): Likewise.
2416 (arm_cortex_a5_tune): Likewise.
2417 (arm_cortex_a9_tune): Likewise.
2418 (arm_cortex_a12_tune): Likewise.
2419 (arm_cortex_a73_tune): Likewise.
2420 (arm_v7m_tune): Likewise.
2421 (arm_cortex_m7_tune): Likewise.
2422 (arm_v6m_tune): Likewise.
2423 (arm_fa726te_tune): Likewise.
2424 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
2426 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2428 * config/arm/arm.c (arm_mem_costs): New function.
2429 (arm_rtx_costs_internal): Use arm_mem_costs.
2431 2017-11-23 Mark Wielaard <mark@klomp.org>
2433 * dwarf2out.c (init_sections_and_labels): Use generation to create
2434 unique ranges_section_label and ranges_base_label. Return generation.
2435 (output_rnglists): Add generation argument. Use generation to create
2436 unique ranges labels.
2437 (dwarf2out_finish): Get generation from init_sections_and_labels
2438 and pass generation to output_rnglists.
2440 2017-11-23 Mike Stump <mikestump@comcast.net>
2441 Eric Botcazou <ebotcazou@adacore.com>
2443 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
2444 * cfgloop.h (struct loop): Add unroll field.
2445 * function.h (struct function): Add has_unroll bitfield.
2446 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
2447 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
2448 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
2449 (pass_rtl_unroll_loops::gate): Likewise.
2450 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
2451 for which loop->unroll==1.
2452 (decide_unroll_constant_iterations): Use note for consistency and
2453 take loop->unroll into account. Return early if loop->unroll is set.
2454 Fix thinko in existing test.
2455 (decide_unroll_runtime_iterations): Use note for consistency and
2456 take loop->unroll into account.
2457 (decide_unroll_stupid): Likewise.
2458 * lto-streamer-in.c (input_cfg): Read loop->unroll.
2459 * lto-streamer-out.c (output_cfg): Write loop->unroll.
2460 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
2462 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
2463 (print_loop): Print loop->unroll if set.
2464 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
2465 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
2466 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
2468 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
2469 loop->unroll is set and smaller than the trip count. Otherwise bypass
2470 entirely the heuristics if loop->unroll is set. Remove dead note.
2471 Fix off-by-one bug in other note.
2472 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
2473 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
2475 (tree_unroll_loops_completely): Make static.
2476 (pass_complete_unroll::execute): Use correct type for variable.
2477 (pass_complete_unrolli::execute): Fix formatting.
2478 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
2480 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
2482 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
2483 enable when TARGET_PREFER_AVX128 is set.
2485 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2487 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
2489 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
2490 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
2492 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2494 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
2495 frequencies. Fix estimation of aggregate parameters.
2497 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2499 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
2502 2017-11-23 Tom de Vries <tom@codesourcery.com>
2504 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
2506 2017-11-23 Tom de Vries <tom@codesourcery.com>
2508 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
2511 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
2514 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
2515 sibcall_value_pcrel_fdpic): Use local variable instead of
2517 (calli_tbr_rel): Add missing operand 2.
2518 (call_valuei_tbr_rel): Add missing operand 3.
2520 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2523 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
2524 bitpos/bitsize covering the whole destination, use store_expr only if
2525 the complex mode is the same. Otherwise, use expand_normal and if
2526 it returns CONCAT, subreg each part separately instead of trying to
2527 subreg the whole result.
2529 2017-11-23 Richard Biener <rguenther@suse.de>
2531 PR tree-optimization/23094
2532 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
2533 come here from walking over backedges in the first iteration.
2534 (vn_reference_lookup_3): Skip clobbers that store the same value.
2536 2017-11-23 Richard Biener <rguenther@suse.de>
2538 PR tree-optimization/81403
2539 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
2540 a block we need a leader relative to.
2541 (phi_translate_1): For nary processing require a leader from
2542 get_representative_for given we run expression simplification
2543 using match-and-simplify. Remove previous fix.
2545 2017-11-22 Jeff Law <law@redhat.com>
2547 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
2548 Use new method allocate_value_range rather than accessing the
2549 vrp_value_range_pool data member directly.
2550 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
2551 to use extract_range_from_stmt method to avoid need for
2552 extract_range_from_assignment method.
2553 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
2554 method rather than setting values_propgated data member directly.
2555 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
2556 and values propagated data members and extract_range_from_assignment
2557 method. Reorder private data members to conform to standards.
2558 Add new methods set_lattice_propagation_complete and
2559 allocate_value_range.
2561 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
2563 PR rtl-optimization/83030
2564 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
2565 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
2566 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
2567 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
2568 CROSSING_JUMP_P flag is already set before setting it.
2569 * cfgrtl.c (fixup_partition_crossing): Likewise.
2570 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
2573 2017-11-22 Jakub Jelinek <jakub@redhat.com>
2575 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
2576 Handle the case where both arguments are using gen_const_vec_series.
2578 2017-11-22 David Malcolm <dmalcolm@redhat.com>
2581 * pretty-print.c (pp_format): Move quoting implementation to
2582 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
2583 to pass address of "quote" local.
2584 (pp_begin_quote): New function.
2585 (pp_end_quote): New function.
2586 * pretty-print.h (printer_fn): Convert penultimate param from bool
2588 (pp_begin_quote): New decl.
2589 (pp_end_quote): New decl.
2590 * tree-diagnostic.c (default_tree_printer): Convert penultimate
2591 param from bool to bool *.
2592 * tree-diagnostic.h (default_tree_printer): Likewise.
2594 2017-11-22 Jeff Law <law@redhat.com>
2596 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
2597 Initialize vr_values.
2598 (evrp_range_analyzer::try_find_new_range): Call methods attached to
2599 vr_values via vr_values class instance rather than delegators.
2600 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
2601 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
2602 (evrp_range_analyzer::push_value_range): Likewise.
2603 (evrp_range_analyzer::pop_value_range): Likewise.
2604 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
2605 most delegators. Those remaining are exposed as public interfaces.
2606 Make vr_values a pointer and private.
2607 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
2609 (evrp_range_analyzer::get_vr_value): New method.
2610 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
2611 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
2612 (class evrp_dom_walker): Attach evrp_folder class, initialize
2613 it in the ctor. Remove temporary delegators.
2614 (evrp_dom_walker::before_dom_children): Call methods in attached
2615 evrp_range_analyzer class via class instance pointer. Use
2616 free value_range_constant_singleton to remove need for
2617 op_with_constant_singleton_value delegator method. Do not
2618 create a vrp_prop class instance for every call! Narrow
2619 scope of a couple variables.
2620 (evrp_dom_walker::cleanup): Call methods in attached
2621 evrp_range_analyzer class via class instance pointer.
2622 * vr-values.h (class vr_values): Privatize many methods and
2625 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2627 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
2628 * cfgexpand.c (expand_debug_expr): Remove handling for them.
2629 * expr.c (expand_expr_real_2): Likewise.
2630 * fold-const.c (const_unop): Likewise.
2631 * optabs-tree.c (optab_for_tree_code): Likewise.
2632 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2633 * tree-inline.c (estimate_operator_cost): Likewise.
2634 * tree-pretty-print.c (dump_generic_node): Likewise.
2635 (op_code_prio): Likewise.
2636 (op_symbol_code): Likewise.
2637 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
2638 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
2639 * internal-fn.c (direct_internal_fn_optab): New function.
2640 (direct_internal_fn_array, direct_internal_fn_supported_p
2641 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
2642 * fold-const-call.c (fold_const_reduction): New function.
2643 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
2645 * tree-vect-loop.c: Include internal-fn.h.
2646 (reduction_code_for_scalar_code): Rename to...
2647 (reduction_fn_for_scalar_code): ...this and return an internal
2649 (vect_model_reduction_cost): Take an internal_fn rather than
2651 (vect_create_epilog_for_reduction): Likewise. Build calls rather
2653 (vectorizable_reduction): Use internal functions rather than tree
2654 codes for the reduction operation. Update calls to the functions
2656 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
2657 Use calls to internal functions rather than REDUC tree codes.
2658 * config/aarch64/aarch64-simd.md: Update comment accordingly.
2660 2017-11-22 Olivier Hainque <hainque@adacore.com>
2662 * config/vxworks.c (vxworks_override_options): Pick default
2663 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
2664 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
2665 DWARF_GNAT_ENCODINGS_DEFAULT.
2666 * config/vxworksae.h: Likewise.
2668 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
2670 PR tree-optimization/83104
2671 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
2674 2017-11-22 Marek Polacek <polacek@redhat.com>
2675 H.J. Lu <hongjiu.lu@intel.com>
2676 Jason Merrill <jason@redhat.com>
2681 * calls.c (initialize_argument_information): Call
2682 warn_parameter_passing_abi target hook.
2683 (store_one_arg): Use 0 for empty record size. Don't push 0 size
2684 argument onto stack.
2685 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
2686 * common.opt: Update -fabi-version description.
2687 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
2688 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
2690 (ix86_is_empty_record): New function.
2691 (ix86_warn_parameter_passing_abi): New function.
2692 (TARGET_EMPTY_RECORD_P): Redefine.
2693 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
2694 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
2695 * doc/tm.texi: Regenerated.
2696 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
2697 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
2698 * dwarf2out.c (get_ultimate_context): Move to tree.c.
2699 * explow.c (hard_function_value): Call arg_int_size_in_bytes
2700 instead of int_size_in_bytes.
2701 * expr.c (copy_blkmode_to_reg): Likewise.
2702 * function.c (aggregate_value_p): Return 0 for empty types.
2703 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
2705 (locate_and_pad_parm): Call arg size_in_bytes instead
2707 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
2708 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
2709 * target.def (empty_record_p, warn_parameter_passing_abi): New target
2711 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
2712 (std_gimplify_va_arg_expr): Skip empty records. Call
2713 arg_size_in_bytes instead size_in_bytes.
2714 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
2715 * tree-core.h (tree_type_common): Add empty_flag.
2716 (tree_decl_common): Update comments.
2717 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
2719 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
2720 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
2722 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
2723 * tree.c (default_is_empty_type): New function.
2724 (default_is_empty_record): New function.
2725 (arg_int_size_in_bytes): New function.
2726 (arg_size_in_bytes): New function.
2727 (get_ultimate_context): New function.
2728 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
2729 TRANSLATION_UNIT_WARN_EMPTY_P.
2730 (default_is_empty_record, arg_int_size_in_bytes,
2731 arg_size_in_bytes, get_ultimate_context): Declare.
2733 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2735 * config/arm/arm.c (cmse_clear_registers): New function.
2736 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
2737 code by call to cmse_clear_registers.
2738 (cmse_nonsecure_entry_clear_before_return): Likewise.
2740 2017-11-22 Tamar Christina <tamar.christina@arm.com>
2742 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
2743 (vdot_s32, vdotq_s32): New.
2744 (vdot_lane_u32, vdotq_lane_u32): New.
2745 (vdot_lane_s32, vdotq_lane_s32): New.
2748 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2751 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
2752 for unsigned values with fewer HWIs than the precision.
2753 (test_overflow): New function.
2754 (wide_int_cc_tests): Call it.
2756 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2758 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
2759 has the same mode class as Pmode.
2761 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2763 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
2764 padding_bits_to_clear_ptr.
2765 (cmse_nonsecure_entry_clear_before_return): Likewise.
2767 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2769 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
2770 auto_sbitap instead of integer bitfield to control register needing
2773 2017-11-22 Jakub Jelinek <jakub@redhat.com>
2775 PR tree-optimization/83044
2776 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
2777 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
2778 that need the upper bound. Subtract offset from
2779 get_addr_base_and_unit_offset only if positive and subtract it
2780 before division by eltsize rather than after it.
2783 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
2784 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
2785 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
2789 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
2790 expand_binop with *mul_widen_optab, make sure at least one of the
2791 operands doesn't have VOIDmode.
2794 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
2796 PR rtl-optimization/82044
2797 PR tree-optimization/82042
2798 * dse.c (record_store): Check for overflow.
2799 (check_mem_read_rtx): Properly check for overflow if width == -1, call
2800 add_wild_read instead of clear_rhs_from_active_local_stores on
2801 overflow and log it into dump_file.
2803 2017-11-22 Richard Biener <rguenther@suse.de>
2805 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
2806 fake edges to exit when looking for a place to insert.
2807 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
2809 (insert_into_preds_of_block): Commit edge insertion immediately,
2810 assert that doesn't require new BBs.
2811 (fini_pre): Release expressions.
2812 (pass_pre::execute): Shuffle things around a bit, if the fn
2813 is too large do not compute AVAIL either as this is really the
2816 2017-11-22 Richard Biener <rguenther@suse.de>
2818 PR tree-optimization/83089
2819 * tree-if-conv.c (pass_if_conversion::execute): If anything
2820 changed reset SCEV and free the number of iteration estimates.
2822 2017-11-21 Martin Sebor <msebor@redhat.com>
2824 PR tree-optimization/82945
2825 * calls.h (warn_nonstring_bound): Remove unused function.
2827 2017-11-21 Martin Sebor <msebor@redhat.com>
2829 PR tree-optimization/82945
2830 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
2831 * calls.h (maybe_warn_nonstring_arg): Declare new function.
2832 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
2834 (initialize_argument_information): Call maybe_warn_nonstring_arg.
2835 * calls.h (get_attr_nonstring_decl): Declare new function.
2836 * doc/extend.texi (attribute nonstring): Update.
2837 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
2838 get_attr_nonstring_decl and handle it.
2839 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
2840 detection of nul-termination.
2841 (strlen_to_stridx): Change to a pointer.
2842 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
2843 (pass_strlen::execute): Same.
2845 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
2847 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
2848 for the new option -mprefer-vector-width=[none|128|256|512].
2849 * config/i386/i386.c (ix86_target_string): remove old style options
2850 -mprefer-avx256 and make -mprefer-avx128 as alias.
2851 (ix86_option_override_internal): Apply defaults for the
2852 -mprefer-vector-width=[128|256] option.
2853 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
2854 Implement macros to work with -mprefer-vector-width=.
2855 * config/i386/i386.opt: Implemented option
2856 -mprefer-vector-width=[none|128|256|512].
2857 * doc/invoke.texi: Documentation for
2858 -mprefer-vector-width=[none|128|256|512].
2860 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
2862 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
2864 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
2866 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
2868 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
2870 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
2872 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
2873 * expr.c (expand_expr_real_2): Likewise.
2874 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
2875 fold_binary_loc): Likewise.
2876 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
2877 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
2878 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
2879 MINUS_EXPR transformations.
2880 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
2881 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
2882 * tree-inline.c (estimate_operator_cost): Likewise.
2883 * tree-pretty-print.c (dump_generic_node, op_code_prio,
2884 op_symbol_code): Likewise.
2885 * tree-vect-stmts.c (vectorizable_operation): Likewise.
2886 * vr-values.c (extract_range_from_binary_expr): Likewise.
2887 * varasm.c (initializer_constant_valid_p_1): Likewise.
2888 * tree.def: New tree code POINTER_DIFF_EXPR.
2890 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
2892 * config/i386/i386.md (*bswap<mode>2_movbe): Add
2893 integer suffix to movbe mnemonic.
2894 (*bswaphi2_movbe): Ditto.
2895 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
2897 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2900 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
2901 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
2902 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
2904 2017-11-21 Martin Liska <mliska@suse.cz>
2906 * tree-inline.c (expand_call_inline): Remove not needed
2909 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
2910 Jakub Jelinek <jakub@redhat.com>
2913 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
2914 Remove static keyword from f variable.
2916 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2918 PR tree-optimization/83086
2919 * gimple-ssa-store-merging.c
2920 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
2921 rather than n.base_addr.
2923 2017-11-21 Martin Liska <mliska@suse.cz>
2925 PR rtl-optimization/82044
2926 PR tree-optimization/82042
2927 * dse.c (check_mem_read_rtx): Check for overflow.
2929 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2931 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
2934 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2936 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
2937 and its corresponding call arg location note.
2939 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2940 Andrew Burgess <andrew.burgess@embecosm.com>
2942 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
2944 (arc_return_slot_offset): Likewise.
2945 (arc_eh_return_address_location): New declaration.
2946 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
2947 (MUST_SAVE_REGISTER): Add exception handler case.
2948 (MUST_SAVE_RETURN_ADDR): Likewise.
2949 (arc_frame_pointer_required): Likewise.
2950 (arc_frame_pointer_needed): New function.
2951 (arc_compute_frame_size): Changed.
2952 (arc_expand_prologue): Likewise.
2953 (arc_expand_epilogue): Likewise.
2954 (arc_initial_elimination_offset): Likewise.
2955 (arc_return_slot_offset): Delete.
2956 (arc_eh_return_address_location): New function.
2957 (arc_builtin_setjmp_frame_value): Likewise.
2958 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
2959 (EH_RETURN_STACKADJ_RTX): Define.
2960 (EH_RETURN_HANDLER_RTX): Likewise.
2961 * config/arc/arc.md (eh_return): Delete.
2963 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2965 * print-tree.h (debug_vec_tree): Remove prototype.
2966 * gdbinit.in (pvt): Remove macro.
2968 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2970 PR tree-optimization/83047
2971 * gimple-ssa-store-merging.c
2972 (imm_store_chain_info::output_merged_store): If the loads with the
2973 same vuse are in different basic blocks, for load_gsi pick a load
2974 location that is dominated by the other loads.
2977 * config/i386/i386.c (ix86_memmodel_check): Start
2978 -Winvalid-memory-model diagnostics with lowercase letter.
2981 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
2982 set in_cold_section_p to the partition containing loc_list->first.
2983 When seeing loc_list->last_before_switch node, update secname and
2984 perform range_across_switch second partition handling only after that.
2987 * run-rtl-passes.c: Include debug.h.
2988 (run_rtl_passes): Call debug_hooks->assembly_start.
2989 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
2993 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
2994 OPTAB_DIRECT in calls to expand_simple_binop.
2996 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2999 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
3001 2017-11-20 David Malcolm <dmalcolm@redhat.com>
3004 * spellcheck.h (best_match::blithely_get_best_candidate): New
3007 2017-11-20 Jakub Jelinek <jakub@redhat.com>
3009 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
3010 with lower case letter.
3012 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
3014 * config/i386/i386.md (bswaphi2): New expander.
3015 (*bswaphi2_movbe): New insn pattern.
3016 (bswaphi -> rorhi pepehole2): New peephole pattern.
3018 2017-11-20 Jeff Law <law@redhat.com>
3020 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
3021 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
3022 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
3023 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
3024 gimple-ssa-evrp-analyze.h.
3026 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
3027 set BB_VISITED here.
3028 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
3030 2017-11-20 Bin Cheng <bin.cheng@arm.com>
3032 * tree-predcom.c: Add general comment on Store-Store chains.
3033 (split_data_refs_to_components): Postpone clearing eliminate_store_p
3035 (get_chain_last_ref_at): Rename into...
3036 (get_chain_last_write_at): ...this.
3037 (get_chain_last_write_before_load): New function.
3038 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
3039 CT_STORE_STORE when write reference is added.
3040 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
3041 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
3042 (initialize_root_vars_store_elim_1): Ditto.
3043 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
3044 definition is created.
3045 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
3046 chain by replacing it with dominant stored value.
3048 2017-11-20 Bin Cheng <bin.cheng@arm.com>
3050 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
3052 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
3054 * vr-values.c (extract_range_from_binary_expr): Use a full range
3057 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
3059 * config/arm/arm.md (R4_REGNUM): Define constant.
3060 (nonsecure_call_internal): Remove r4 clobber.
3061 (nonsecure_call_value_internal): Likewise.
3062 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
3063 clobber and resequence match_operands.
3064 (nonsecure_call_value_reg_thumb1_v5): Likewise.
3065 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
3066 (nonsecure_call_value_reg_thumb2): Likewise.
3068 2017-11-20 Jakub Jelinek <jakub@redhat.com>
3070 PR tree-optimization/78821
3071 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
3072 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
3073 to the address of the base rather than the base itself.
3074 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
3075 (find_bswap_or_nop_finalize): New function.
3076 (find_bswap_or_nop): Use it.
3077 (bswap_replace): Return a tree rather than bool, change first
3078 argument from gimple * to gimple_stmt_iterator, allow inserting
3079 into an empty sequence, allow ins_stmt to be NULL - then emit
3080 all stmts into gsi. Fix up MEM_REF address gimplification.
3081 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
3082 (struct store_immediate_info): Add N and INS_STMT non-static
3084 (store_immediate_info::store_immediate_info): Initialize them
3085 from newly added ctor args.
3086 (merged_store_group::apply_stores): Formatting fixes. Sort by
3088 (stmts_may_clobber_ref_p): For stores call also
3089 refs_anti_dependent_p.
3090 (gather_bswap_load_refs): New function.
3091 (imm_store_chain_info::try_coalesce_bswap): New method.
3092 (imm_store_chain_info::coalesce_immediate_stores): Use it.
3093 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
3094 (imm_store_chain_info::output_merged_store): Fail if number of
3095 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
3096 and NOP_EXPR rhs_code.
3097 (pass_store_merging::process_store): Compute n and ins_stmt, if
3098 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
3099 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
3101 (pass_store_merging::execute): Calculate dominators.
3103 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
3104 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
3105 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
3106 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3107 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
3108 class pass_optimize_bswap, bswap_replace,
3109 pass_optimize_bswap::execute): Moved to ...
3110 * gimple-ssa-store-merging.c: ... this file.
3111 Include optabs-tree.h.
3112 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
3113 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3114 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
3115 anonymous namespace, remove static keywords.
3116 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
3117 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
3119 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
3122 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
3124 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
3126 * vec.h (debug_helper): New function.
3127 (DEFINE_DEBUG_VEC): New macro.
3128 * hash-set.h (debug_helper): New function.
3129 (DEFINE_DEBUG_HASH_SET): New macro.
3130 * cfg.c (debug_slim (edge)): New function.
3131 Call DEFINE_DEBUG_VEC for edges.
3132 Call DEFINE_DEBUG_HASH_SET for edges.
3133 * cfghooks.c (debug_slim (basic_block)): New function.
3134 Call DEFINE_DEBUG_VEC for basic blocks.
3135 Call DEFINE_DEBUG_HASH_SET for basic blocks.
3136 * print-tree.c (debug_slim): New function to handle trees.
3137 Call DEFINE_DEBUG_VEC for trees.
3138 Call DEFINE_DEBUG_HASH_SET for trees.
3139 (debug (vec<tree, va_gc>) &): Remove.
3140 (debug (<vec<tree, va_gc>) *): Remove.
3141 * print-rtl.c (debug_slim): New function to handle const_rtx.
3142 Call DEFINE_DEBUG_VEC for rtx_def.
3143 Call DEFINE_DEBUG_VEC for rtx_insn.
3144 Call DEFINE_DEBUG_HASH_SET for rtx_def.
3145 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
3146 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
3147 (debug (vec<rtx_insn *> *ptr): Remove.
3148 (debug_insn_vector): Remove.
3149 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
3151 2017-11-20 Tom de Vries <tom@codesourcery.com>
3153 PR rtl-optimization/82020
3154 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
3155 IF_THEN_ELSE condition.
3157 2017-11-19 Jeff Law <law@redhat.com>
3159 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
3160 of degenerates resulting from ignoring an edge.
3162 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3165 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
3167 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3170 * profile-count.c (profile_count::to_sreal_scale): Fix return value
3171 for uninitialied counts.
3173 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3176 * tree-inline.c (estimate_num_insns): Set to 1 at least.
3178 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3181 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
3183 2017-11-19 Tom de Vries <tom@codesourcery.com>
3185 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
3188 2017-11-19 Tom de Vries <tom@codesourcery.com>
3190 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
3193 2017-11-19 Tom de Vries <tom@codesourcery.com>
3195 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
3198 2017-11-19 Tom de Vries <tom@codesourcery.com>
3200 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
3202 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
3204 2017-11-19 Tom de Vries <tom@codesourcery.com>
3206 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
3208 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
3209 semicolon after MCORE_EXPORT_NAME call.
3211 2017-11-19 Tom de Vries <tom@codesourcery.com>
3214 * vmsdbgout.c (vmsdbgout_early_finish): New function.
3215 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
3217 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
3219 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
3220 (cgraph_edge::clone): Cleanup updating of profile.
3221 * ipa-cp.c (update_profiling_info): Likewise.
3222 * ipa-inline-transform.c (inline_transform): Likewise.
3223 * ipa-inline.c (inline_small_functions): Add missing space to dump.
3224 * ipa-split.c (execute_split_functions): Do not split when function
3226 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
3227 * profile-count.c (profile_count::dump): Add global0.
3228 (profile_count::to_cgraph_frequency): Do not ICE when entry is
3230 (profile_count::to_sreal_scale): Likewise.
3231 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
3232 (profile_count::combine_with_ipa_count): New function.
3233 * profile-count.h (profile_guessed_global0adjusted): New.
3234 (profile_count::adjusted_zero): New.
3235 (profile_count::global0adjusted): New.
3236 (profile_count::combine_with_ipa_count): New.
3237 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
3238 correct profile of return block of split functions.
3239 (copy_cfg_body): Remove unused profile_count.
3240 (copy_body): Likewise.
3241 (expand_call_inline): Update.
3242 (tree_function_versioning): Update.
3244 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
3246 * hash-set.h (hash_set::empty): New.
3247 * tree-ssa-threadbackward.h: Delete.
3248 * tree-ssa-threadbackward.c (class thread_jumps): New.
3249 Move max_threaded_paths into class.
3250 (fsm_find_thread_path): Remove arguments that are now in class.
3251 (profitable_jump_thread_path): Rename to...
3252 (thread_jumps::profitable_jump_thread_path): ...this.
3253 (convert_and_register_jump_thread_path): Rename to...
3254 (thread_jumps::convert_and_register_current_path): ...this.
3255 (check_subpath_and_update_thread_path): Rename to...
3256 (thread_jumps::check_subpath_and_update_thread_path): ...this.
3257 (register_jump_thread_path_if_profitable): Rename to...
3258 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
3259 (handle_phi): Rename to...
3260 (thread_jumps::handle_phi): ...this.
3261 (handle_assignment): Rename to...
3262 (thread_jumps::handle_assignment): ...this.
3263 (fsm_find_control_statement_thread_paths): Rename to...
3264 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
3265 (find_jump_threads_backwards): Rename to...
3266 (thread_jumps::find_jump_threads_backwards): ...this.
3267 Initialize path local data.
3268 (pass_thread_jumps::execute): Call find_jump_threads_backwards
3269 from within thread_jumps class.
3270 (pass_early_thread_jumps::execute): Same.
3272 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3274 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
3276 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3278 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
3279 to canonical location.
3281 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3284 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
3286 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
3288 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
3290 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
3291 rcrt1.o%s/grcrt1.o%s for -static-pie.
3293 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3295 * i386.c (ix86_multiplication_cost, ix86_division_cost,
3296 ix86_shift_rotate_cost): Break out from ...
3297 (ix86_rtx_costs): ... here.
3298 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
3301 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3303 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
3304 when entry block was promoted unlikely.
3305 (estimate_bb_frequencies): Increase frequency scale.
3306 * profile-count.h (profile_count): Export precision info.
3308 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3310 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
3311 disturbin profile of entry block.
3313 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3315 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
3318 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3320 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
3321 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
3323 2017-11-17 Jeff Law <law@redhat.com>
3325 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
3326 from evrp_dom_walker class. Various methods moved into new class.
3327 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
3328 (evrp_range_analyzer::enter): New method.
3329 (evrp_range_analyzer::leave): New method.
3330 (evrp_dom_walker): Remove delegators no longer needed by this class.
3331 Replace vr_values data member with evrp_range_analyzer
3333 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
3334 method extracted from evrp_dom_walker::before_dom_children.
3335 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
3336 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
3338 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
3339 Add private copy constructor and move assignment operators.
3340 Privatize methods and class data where trivially possible.
3341 (evrp_dom_walker::cleanup): New function, extracted from
3342 execute_early_vrp. Simplify access to class data.
3344 * vr-values.h (get_output_for_vrp): Prototype.
3345 * vr-values.c (get_output_for_vrp): New function extracted from
3346 vrp_visit_assignment_or_call and extract_range_from_stmt.
3347 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
3349 2017-11-17 Luis Machado <luis.machado@linaro.org>
3351 * config/aarch64/aarch64.c
3352 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
3353 (qdf24xx_tunings) <autoprefetcher_model>: Set to
3354 tune_params::AUTOPREFETCHER_WEAK.
3356 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3359 * config/arm/arm.c (arm_valid_target_attribute_rec):
3360 Parse "arch=" and "+<ext>".
3361 (arm_valid_target_attribute_tree): Re-init global options.
3362 (arm_option_override): Make non-static.
3363 (arm_options_perform_arch_sanity_checks): Make errors fatal.
3364 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
3365 (__ARM_FEATURE_CRC32): Support undef.
3366 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
3367 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
3369 2017-11-17 David Malcolm <dmalcolm@redhat.com>
3371 * gdbinit.in (break-on-diagnostic): New command.
3373 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3375 * config/i386/i386.c (ix86_expand_epilogue): Change simple
3376 return to indirect jump for EH return if control-flow
3377 protection is enabled. Change explicit 'false' argument in
3378 pro_epilogue_adjust_stack with a value of
3380 * config/i386/i386.md (simple_return_indirect_internal):
3381 Remove SImode restriction to support 64-bit.
3383 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3385 * combine.c (added_notes_insn): New.
3386 (try_combine): Handle added_notes_insn like added_links_insn.
3387 Rewrite return value code.
3388 (distribute_notes): Set added_notes_insn to the earliest insn we added
3391 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3393 PR rtl-optimization/82621
3394 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
3395 dest of one of those SETs is unused.
3397 2017-11-17 Richard Biener <rguenther@suse.de>
3400 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
3401 * tree-pretty-print.c (dump_generic_node): Handle
3402 annot_expr_parallel_kind.
3403 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
3404 * gimplify.c (gimple_boolify): Likewise.
3406 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3408 * config.gcc (extra_headers): Add cet.h for x86 targets.
3409 * config/i386/cet.h: New file.
3410 * doc/install.texi: Add --enable-cet/--disable-cet.
3412 2017-11-17 Richard Biener <rguenther@suse.de>
3414 PR tree-optimization/83017
3415 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
3416 (gen_parallel_loop): Properly count iterations.
3417 (parallelize_loops): Handle loop->can_be_parallel independent
3418 of flag_loop_parallelize_all. Make static profitability test match
3420 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
3421 * invoke.texi (parloops-min-per-thread): Document.
3423 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
3425 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
3426 upstreaming review comments.
3428 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3430 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
3431 with fast unaligned access.
3432 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
3434 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
3436 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
3437 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
3439 2017-11-17 Richard Biener <rguenther@suse.de>
3441 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
3442 folding of references.
3444 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
3447 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
3448 of replacing call to strncmp with corresponding call to strcmp when
3451 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
3453 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
3454 option prefer-avx256 for skylake-avx512 configuration.
3455 * config/i386/i386.c (ix86_option_override_internal): Ditto.
3456 (get_builtin_code_for_version): Ditto.
3458 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3459 Monk Chiang <sh.chiang04@gmail.com>
3461 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
3462 (FIXED_REGISTERS): Reserve more register numbers.
3463 (CALL_USED_REGISTERS): Likewise.
3464 (REG_ALLOC_ORDER): Likewise.
3465 (REG_CLASS_CONTENTS): Likewise.
3466 (REGISTER_NAMES): Likewise.
3468 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3469 Kito Cheng <kito.cheng@gmail.com>
3471 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
3473 * config/nds32/iterators.md: Add vector mode iterators and attributes.
3475 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
3477 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
3478 parameter list for vec_splats.
3480 2017-11-16 Joseph Myers <joseph@codesourcery.com>
3482 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
3484 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
3486 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
3489 * doc/install.texi: Document incompatibility of Perl >=5.6.26
3490 with the required version of automake 1.11.6.
3492 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
3494 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
3496 (DU_C2_3_power9): Correct reservation combinations.
3497 (FP_DIV_power9, VEC_DIV_power9): New.
3498 (power9-alu): Split out rotate/shift...
3499 (power9-rot): ...to here, correct dispatch resource.
3500 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
3502 (power9-fp): Correct latency.
3503 (power9-sdiv): Add div/sqrt resource.
3504 (power9-ddiv): Correct latency, add div/sqrt resource.
3505 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
3506 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
3508 (power9-qpdiv, power9-qpmul): Adjust resource usage.
3510 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
3512 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
3513 switch statement mapping KF built-ins to TF built-ins if we don't
3514 have the proper ISA 3.0 assembler support.
3516 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3518 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
3519 (lower_emutls_function_body): Do not compute it.
3521 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3523 * ipa-split.c (split_bb_info): Turn time to sreal.
3524 (split_point): Likewise.
3525 (dump_split_point): Likewise.
3526 (fine_split_points): Likewise.
3527 (execute_split_functions): Only zero split_bbs; turn time to sreals.
3529 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3531 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
3534 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3536 * predict.c (combine_predictions_for_bb): Preserve zero predicted
3538 (expensive_function_p): Remove useless assert.
3539 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
3541 2017-11-16 Martin Sebor <msebor@redhat.com>
3543 PR tree-optimization/82588
3544 PR tree-optimization/82583
3545 * tree-vrp.c (check_array_ref): Handle flexible array members,
3546 string literals, and inner indices.
3547 (search_for_addr_array): Add detail to diagnostics.
3549 2017-11-16 Nathan Sidwell <nathan@acm.org>
3553 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
3554 (SET_DECL_ASSEMBLER_NAME): Forward to
3555 overwrite_decl_assembler_name.
3556 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
3557 (overwrite_decl_assembler_name): Declare.
3558 * tree.c (overwrite_decl_assembler_name): New.
3559 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
3560 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
3561 (LANG_HOOKS_INITIALIZER): Add it.
3562 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
3563 * langhooks.c (lhd_set_decl_assembler_name): Use
3564 SET_DECL_ASSEMBLER_NAME.
3565 (lhd_overwrite_decl_assembler_name): Default implementation.
3567 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
3568 Jackson Woodruff <jackson.woodruff@arm.com>
3570 PR tree-optimization/71026
3571 * match.pd: Canonicalize constant multiplies in division.
3573 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
3575 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
3576 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
3577 Remove OPT_fomit_frame_pointer entry.
3578 * common/config/arc/arc-common.c: Likewise.
3579 * common/config/arm/arm-common.c: Likewise.
3580 * common/config/avr/avr-common.c: Likewise.
3581 * common/config/c6x/c6x-common.c: Likewise.
3582 * common/config/cr16/cr16-common.c: Likewise.
3583 * common/config/cris/cris-common.c: Likewise.
3584 * common/config/epiphany/epiphany-common.c: Likewise.
3585 * common/config/fr30/fr30-common.c: Likewise.
3586 * common/config/frv/frv-common.c: Likewise.
3587 * common/config/ia64/ia64-common.c: Likewise.
3588 * common/config/iq2000/iq2000-common.c: Likewise.
3589 * common/config/lm32/lm32-common.c: Likewise.
3590 * common/config/m32r/m32r-common.c: Likewise.
3591 * common/config/mcore/mcore-common.c: Likewise.
3592 * common/config/microblaze/microblaze-common.c: Likewise.
3593 * common/config/mips/mips-common.c: Likewise.
3594 * common/config/mmix/mmix-common.c: Likewise.
3595 * common/config/mn10300/mn10300-common.c: Likewise.
3596 * common/config/nios2/nios2-common.c: Likewise.
3597 * common/config/pa/pa-common.c: Likewise.
3598 * common/config/pdp11/pdp11-common.c: Likewise.
3599 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
3600 * common/config/riscv/riscv-common.c: Likewise.
3601 * common/config/rs6000/rs6000-common.c: Likewise.
3602 * common/config/rx/rx-common.c: Likewise.
3603 * common/config/s390/s390-common.c: Likewise.
3604 * common/config/sh/sh-common.c: Likewise.
3605 * common/config/sparc/sparc-common.c: Likewise.
3606 * common/config/tilegx/tilegx-common.c: Likewise.
3607 * common/config/tilepro/tilepro-common.c: Likewise.
3608 * common/config/v850/v850-common.c: Likewise.
3609 * common/config/visium/visium-common.c: Likewise.
3610 * common/config/xstormy16/xstormy16-common.c: Likewise.
3611 * common/config/xtensa/xtensa-common.c: Likewise.
3612 * invoke.texi (-fomit-frame-pointer): Update documentation.
3614 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3616 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
3618 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3620 * cfg.c (scale_bbs_frequencies_int,
3621 cale_bbs_frequencies_gcov_type): Remove.
3622 * cfg.h (scale_bbs_frequencies_int,
3623 cale_bbs_frequencies_gcov_type): Remove.
3625 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3627 * tree-ssa-loop-manip.c
3628 (scale_dominated_blocks_in_loop): Update to profile counts.
3629 (tree_transform_and_unroll_loop): Likewise.
3631 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3633 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
3634 scale_bbs_frequencies_int.
3636 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3638 * final.c (compute_alignments): Use counts rather than frequencies.
3640 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3642 * cfgloopanal.c: Include sreal.h
3643 (average_num_loop_insns): Use counts and sreal for accounting.
3645 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3647 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
3650 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3651 Kito Cheng <kito.cheng@gmail.com>
3653 * config/nds32/constraints.md: Provide more constraints.
3654 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
3655 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
3656 support constraints usage.
3658 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3660 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
3662 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3663 Kito Cheng <kito.cheng@gmail.com>
3665 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
3666 * config/nds32/nds32.opt: Refine the layout.
3667 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
3668 TARGET_EXT_STRING): Support new options.
3669 * config/nds32/nds32.h: Likewise.
3670 * config/nds32/nds32.md: Likewise.
3671 * config/nds32/nds32-predicates.c: Likewise.
3672 * config/nds32/constraints.md: Likewise.
3673 * common/config/nds32/nds32-common.c: Likewise.
3675 2017-11-16 Julia Koval <julia.koval@intel.com>
3678 * config/i386/gfniintrin.h: Add sse check.
3679 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
3681 2017-11-16 Julia Koval <julia.koval@intel.com>
3683 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
3684 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
3685 (ix86_handle_option): Handle -mavx512vbmi2.
3686 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
3687 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3688 * config/i386/i386-c.c (__AVX512VBMI2__): New.
3689 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
3690 (ix86_valid_target_attribute_inner_p): Ditto.
3691 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
3692 * config/i386/i386.opt (mavx512vbmi2): New option.
3693 * doc/invoke.texi: Add new option.
3695 2017-11-16 Julia Koval <julia.koval@intel.com>
3697 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
3698 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
3699 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
3700 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
3701 _mm512_gf2p8mul_epi8): New intrinsics.
3702 * config/i386/i386-builtin-types.def
3703 (V64QI_FTYPE_V64QI_V64QI): New type.
3704 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
3705 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
3706 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
3707 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
3708 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
3709 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3711 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
3713 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
3714 explicitly as a stream of bytes.
3716 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3718 * config/rs6000/altivec.h (vec_xst_be): New #define.
3719 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
3720 and externalize from *altivec_vperm_<mode>_internal.
3721 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
3723 (XL_BE_V8HI): Likewise.
3724 (XL_BE_V4SI): Likewise.
3725 (XL_BE_V4SI): Likewise.
3726 (XL_BE_V2DI): Likewise.
3727 (XL_BE_V4SF): Likewise.
3728 (XL_BE_V2DF): Likewise.
3729 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
3730 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
3731 all array entries with these keys: VSX_BUILTIN_VEC_XL,
3732 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
3733 VSX_BUILTIN_VEC_XST_BE.
3734 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
3735 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
3737 (altivec_init_builtins): Replace conditional calls to def_builtin
3738 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
3739 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
3740 with unconditional calls. Remove calls to def_builtin for
3741 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
3742 __builtin_vec_xst_be.
3743 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
3744 to define_expand, and add alternate RTL generation for P8.
3745 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
3746 vsx_ld_elemrev_v8hi.
3747 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
3748 add alternate RTL generation for P8.
3749 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
3750 vsx_ld_elemrev_v16qi.
3751 (vsx_st_elemrev_v8hi): Convert define_insn
3752 to define_expand, and add alternate RTL generation for P8.
3753 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
3754 vsx_st_elemrev_v8hi.
3755 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
3756 add alternate RTL generation for P8.
3757 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
3758 vsx_st_elemrev_v16qi.
3760 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
3763 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
3764 TARGET_AVX512ER check.
3765 (ix86_option_override_internal): Set MASK_VZEROUPPER if
3766 neither -mzeroupper nor -mno-zeroupper is used and
3767 TARGET_EMIT_VZEROUPPER is set.
3768 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
3769 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
3771 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
3773 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3774 folding of vector compares.
3775 (fold_build_vec_cmp): New helper function.
3776 (fold_compare_helper): New helper function.
3777 (builtin_function_type): Add compare builtins to the list of functions
3778 having unsigned arguments. Cosmetic updates to comment indentation.
3779 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
3780 the not+eq combination.
3782 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3784 PR tree-optimization/82726
3785 PR tree-optimization/70754
3786 * tree-predcom.c (order_drefs_by_pos): New function.
3787 (combine_chains): Move code setting has_max_use_after to...
3788 (try_combine_chains): ...here. New parameter. Sort combined chains
3789 according to position information.
3790 (tree_predictive_commoning_loop): Update call to above function.
3791 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
3793 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3795 PR tree-optimization/82726
3797 2017-01-23 Bin Cheng <bin.cheng@arm.com>
3799 PR tree-optimization/70754
3800 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
3801 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
3802 combined stmt before it if not NULL.
3803 (combine_chains): Process refs reversely and compute dominance point
3807 2017-02-23 Bin Cheng <bin.cheng@arm.com>
3809 PR tree-optimization/79663
3810 * tree-predcom.c (combine_chains): Process refs in reverse order
3811 only for ZERO length chains, and add explaining comment.
3813 2017-11-15 Tamar Christina <tamar.christina@arm.com>
3815 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
3816 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
3817 * doc/invoke.texi (ARM Options): Add armv8.3-a.
3819 2017-11-15 Tamar Christina <tamar.christina@arm.com>
3821 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
3823 2017-11-15 Martin Liska <mliska@suse.cz>
3825 * tree-cfg.c (pass_warn_function_return::execute):
3826 Compare warn_return_type for greater than zero.
3828 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
3832 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
3833 to return true on Xeon and not on Xeon Phi.
3834 (ix86_check_avx256_register): Changed to ...
3835 (ix86_check_avx_upper_register): ... this. Add extra check for
3836 VALID_AVX512F_REG_OR_XI_MODE.
3837 (ix86_avx_u128_mode_needed): Changed
3838 ix86_check_avx256_register to ix86_check_avx_upper_register.
3839 (ix86_check_avx256_stores): Changed to ...
3840 (ix86_check_avx_upper_stores): ... this. Changed
3841 ix86_check_avx256_register to ix86_check_avx_upper_register.
3842 (ix86_avx_u128_mode_after): Changed
3843 avx_reg256_found to avx_upper_reg_found. Changed
3844 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
3845 (ix86_avx_u128_mode_entry): Changed
3846 ix86_check_avx256_register to ix86_check_avx_upper_register.
3847 (ix86_avx_u128_mode_exit): Ditto.
3848 * config/i386/i386.h: (host_detect_local_cpu): New define.
3850 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3852 * config/arm/xgene1.md (xgene1): Split into automatons
3853 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
3854 (xgene1_f_load): Adjust reservations and/or types.
3855 (xgene1_f_store): Likewise.
3856 (xgene1_load_pair): Likewise.
3857 (xgene1_store_pair): Likewise.
3858 (xgene1_fp_load1): Likewise.
3859 (xgene1_load1): Likewise.
3860 (xgene1_store1): Likewise.
3861 (xgene1_move): Likewise.
3862 (xgene1_alu): Likewise.
3863 (xgene1_simd): Likewise.
3864 (xgene1_bfm): Likewise.
3865 (xgene1_neon_load1): Likewise.
3866 (xgene1_neon_store1): Likewise.
3867 (xgene1_neon_logic): Likewise.
3868 (xgene1_neon_st1): Likewise.
3869 (xgene1_neon_ld1r): Likewise.
3870 (xgene1_alu_cond): Added.
3871 (xgene1_shift_reg): Likwise.
3872 (xgene1_bfx): Likewise.
3873 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
3875 2017-11-15 Jakub Jelinek <jakub@redhat.com>
3878 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
3880 (can_widen_mult_without_libcall): New function.
3881 (expand_mul_overflow): If only checking unsigned mul overflow,
3882 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
3883 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
3884 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
3885 (expand_DIVMOD): Formatting fix.
3886 * expmed.h (expand_mult): Add NO_LIBCALL argument.
3887 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
3888 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
3890 PR tree-optimization/82977
3891 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
3892 constructed temporary to strlen_to_stridx.put.
3894 2017-11-15 Martin Liska <mliska@suse.cz>
3896 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
3897 * configure: Regenerate.
3899 2017-11-15 Martin Liska <mliska@suse.cz>
3902 * config/sh/sh-mem.cc: Use proper probability for
3905 2017-11-14 Jeff Law <law@redhat.com>
3907 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
3908 the red zone for stack_clash_protection_final_dynamic_probe targets
3909 when the total dynamic stack size is zero bytes.
3911 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
3912 blocks is post order.
3914 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
3916 * dumpfile.h (TDF_COMPARE_DEBUG): New.
3917 * final.c (rest_of_clean_state): Set it for the
3918 -fcompare-debug dump.
3919 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
3920 class when TDF_COMPARE_DEBUG is set.
3922 * dwarf2out.c (gen_producer_string): Discard
3925 2017-11-15 Joseph Myers <joseph@codesourcery.com>
3928 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
3929 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
3930 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
3931 (__TGMATH_REAL_2_3): Remove macros.
3932 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
3933 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
3935 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
3937 (remquo): Define using __TGMATH_REAL_3.
3939 2017-11-14 Jeff Law <law@redhat.com>
3941 * vr-values.c: New file with contents extracted from tree-vrp.c.
3942 * Makefile.in (OBJS): Add vr-values.o
3943 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
3944 (set_value_range, set_and_canonicalize_value_range): Likewise.
3945 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
3946 (value_range_constant_singleton, symbolic_range_p): Likewise.
3947 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
3948 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
3949 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
3950 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
3951 (find_case_label_range, find_case_label_index): Likewise.
3952 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
3953 (range_int_cst_singleton_p, value_inside_range): Likewise.
3954 (get_single_symbol): Likewise.
3955 (switch_update): Move structure definition here.
3956 (to_remove_edges, to_update_switch_stmts): Provide externs.
3957 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
3958 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
3959 (vrp_val_is_min, set_value_range): Likewise.
3960 (set_and_canonicalize_value_range, copy_value_range): Likewise.
3961 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
3962 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
3963 (range_is_nonnull, range_int_cst_p): Likewwise.
3964 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
3965 (get_single_symbol, operand_less_p): Likewise
3966 (compare_values_warnv, compare_values): Likewise.
3967 (value_inside_range, value_range_constant_singleton): Likewise.
3968 (zero_nonzero_bitgs_from_vr): Likewise.
3969 (extract_range_from_binary_expr_1): Likewise.
3970 (overflow_comparison_p): Likewise.
3971 (to_remove_edges, to_update_switch_stmts): Likewise.
3972 (find_case_label-index, find_case_label_range): Likewise.
3973 (switch_update, set_value_range_to_nonnegative): Remove.
3974 (set_value_range_to_truthvalue): Likewise.
3975 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
3976 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
3977 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
3978 (find_case_label_ranges, test_for_singularity): Likewise.
3979 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
3980 (x_vr_values): Move to its remaining use site.
3982 2017-11-10 Jeff Law <law@redhat.com>
3984 * vr-values.h (VR_INITIALIZER): Move #define here.
3985 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
3986 * Makefile.in (OBJS): Add tree-evrp.o
3987 * tree-vrp.h (assert_info): Move structure definition here.
3988 (set_value_range_to_varying): Prototype.
3989 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3990 (infer_value_range, register_edge_assert_for): Likewise.
3991 (stmt_interesting_for_vrp): Likewise.
3992 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
3993 (set_value_range_to_varying): No longer static.
3994 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3995 (infer_value_range, register_edge_assert_for): Likewise.
3997 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
3999 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
4000 generate the XXBRD instruction.
4002 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
4003 (rs6000_builtin_type_compatible): Treat _Float128 and long double
4004 as being compatible if -mabi=ieeelongdouble.
4005 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
4006 to setup float128 built-ins with hardware support.
4007 (BU_FLOAT128_HW_2): Likewise.
4008 (BU_FLOAT128_HW_3): Likewise.
4009 (BU_FLOAT128_HW_VSX_1): Likewise.
4010 (BU_FLOAT128_HW_VSX_2): Likewise.
4011 (scalar_extract_expq): Change float128 built-in functions to
4012 accommodate having both KFmode and TFmode functions. Use the
4013 KFmode variant as the default.
4014 (scalar_extract_sigq): Likewise.
4015 (scalar_test_neg_qp): Likewise.
4016 (scalar_insert_exp_q): Likewise.
4017 (scalar_insert_exp_qp): Likewise.
4018 (scalar_test_data_class_qp): Likewise.
4019 (sqrtf128_round_to_odd): Delete processing the round to odd
4020 built-in functions as special built-in functions, and define them
4021 as float128 built-ins. Use the KFmode variant as the default.
4022 (truncf128_round_to_odd): Likewise.
4023 (addf128_round_to_odd): Likewise.
4024 (subf128_round_to_odd): Likewise.
4025 (mulf128_round_to_odd): Likewise.
4026 (divf128_round_to_odd): Likewise.
4027 (fmaf128_round_to_odd): Likewise.
4028 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
4029 support for KFmode and TFmode xststdcqp calls.
4030 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
4031 point, switch the built-in handlers for the get/set float128
4032 exponent, get float128 mantissa, float128 test built-ins, and the
4033 float128 round to odd built-in functions. Eliminate creating the
4034 float128 round to odd built-in functions as special built-ins.
4035 (rs6000_init_builtins): Eliminate special creation of the float128
4036 round to odd built-in functions.
4037 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
4038 function insns to support both TFmode and KFmode variants.
4039 (xsxsigqp_<mode>): Likewise.
4040 (xsiexpqpf_<mode>): Likewise.
4041 (xsiexpqp_<mode>): Likewise.
4042 (xststdcqp_<mode>): Likewise.
4043 (xststdcnegqp_<mode>): Likewise.
4044 (xststdcqp_<mode>): Likewise.
4046 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4048 * tree-ssa-threadupdate.c (compute_path_counts): Remove
4049 unused path_in_freq_ptr parameter.
4050 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
4052 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4054 * ipa-inline.c (edge_badness): Dump sreal frequency.
4055 (compute_inlined_call_time): Match natural implementaiton ...
4056 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
4057 forgotten division by CGRAPH_FREQ_BASE.
4059 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4061 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
4062 Solaris 11. Update comment.
4063 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
4065 * config/sol2.h (STARTFILE_SPEC): Likewise.
4066 * configure: Regenerate.
4068 2017-11-14 Carl Love <cel@us.ibm.com>
4070 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
4071 le_ and be_ prefixes to swap* variables. Remove
4072 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
4075 2017-11-14 Jason Merrill <jason@redhat.com>
4077 Support GTY((cache)) on hash_map.
4078 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
4079 (ggc_cache_remove): Override it instead of ggc_mx.
4080 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
4081 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
4082 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
4083 (simple_cache_map_traits): Override maybe_mx.
4084 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
4085 (hash_map): Friend gt_cleare_cache.
4086 (gt_cleare_cache): New.
4087 * tree.h (tree_cache_traits): New hash_map traits class.
4088 (tree_cache_map): New typedef.
4090 2017-11-14 Richard Biener <rguenther@suse.de>
4092 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
4093 paramter and handling.
4094 (cleanup_control_flow_bb): Likewise.
4095 (cleanup_control_flow_pre): New helper performing a DFS walk
4096 to call cleanup_control_flow_bb in PRE order.
4097 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
4098 via cleanup_control_flow_pre.
4100 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
4102 * config/aarch64/aarch64-simd.md
4103 (aarch64_simd_bsl<mode>_internal): Remove DImode.
4104 (*aarch64_simd_bsl<mode>_alt): Likewise.
4105 (aarch64_simd_bsldi_internal): New.
4106 (aarch64_simd_bsldi_alt): Likewise.
4108 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4110 * tracer.c (better_p): Do not compare frequencies.
4111 * reg-stack.c (better_edge): Likewise.
4112 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
4115 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4117 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
4118 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
4119 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
4120 * lto-streamer-in.c (input_function): Use update_max_bb_count.
4121 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
4122 * predict.c (maybe_hot_frequency_p): Inline to ...
4123 (maybe_hot_count_p): ... here; rewrite to counts.
4124 (counts_to_freqs): Rename to ...
4125 (update_max_bb_count): ... this one.
4126 (expensive_function_p): Use counts.
4127 (estimate_bb_frequencies): Update.
4128 (rebuild_frequencies): Update.
4129 * predict.h (counts_to_freqs): Rename to ...
4130 (update_max_bb_count): ... this one.
4131 * profile.c (compute_branch_probabilities): Add debug info
4132 * tree-inline.c (expand_call_inline): Update debug info.
4133 (optimize_inline_calls): Use update_max_bb_count..
4134 (tree_function_versioning): Use update_max_bb_count..
4135 * value-prof.c (gimple_value_profile_transformations):
4136 Do not use update_max_bb_count.
4138 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4140 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
4141 always use frequencies.
4143 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4145 * bb-reorder.c: Remove frequencies from comments.
4146 (better_edge_p): Use profile counts.
4147 (find_traces): Dump profile counts.
4148 (rotate_loop): Use profile counts.
4149 (find_traces_1_round): Likewise.
4150 (connect_better_edge_p): Use counts instead of probabilities for
4152 (copy_bb_p): Drop early check for non-0 frequency.
4153 (sanitize_hot_paths): Update comments.
4155 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4157 * ipa-split.c (struct split_point): Add count.
4158 (consider_split): Do not compute incoming frequency; compute incoming
4159 count and store it to split_point.
4160 (split_function): Set count of the call to split part correctly.
4162 2017-11-13 Carl Love <cel@us.ibm.com>
4164 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
4166 2017-11-13 Tom Tromey <tom@tromey.com>
4168 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
4170 2017-11-13 Carl Love <cel@us.ibm.com>
4172 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
4173 Add support for builtins:
4174 unsigned int vec_first_{,miss}_match_{,or_eos}index,
4175 vector {un,}signed {char,int,short},
4176 vector {un,}signed {char,int,short}) arguments.
4177 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
4178 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
4179 Add BU_P9V_AV_2 expansions for the builtins.
4180 * config/rs6000/altivec.h (vec_first_match_index,
4181 vec_first_mismatch_index, vec_first_match_or_eos_index,
4182 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
4183 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
4184 new extern declaration.
4185 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
4186 * config/rs6000/vsx.md (first_match_index_<mode>,
4187 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
4188 first_mismatch_or_eos_index_<mode>): Add define expand.
4189 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
4190 * doc/extend.texi: Update the built-in documenation file for the new
4193 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
4195 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
4196 into the min/max operations for _Float<N> and _Float<N>X types.
4198 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4201 * dwarf2out.c (do_eh_frame): New static variable.
4202 (dwarf2out_begin_prologue): Set it.
4203 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
4205 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4207 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
4209 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
4211 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4214 * debug.h (dwarf2out_do_eh_frame): Declare.
4215 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
4216 (dwarf2out_do_frame): Use it.
4217 (dwarf2out_do_cfi_asm): Likewise.
4218 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
4219 (dwarf2out_assembly_start): Likewise.
4220 (dwarf2out_begin_prologue): Fix comment.
4221 * toplev.c (compile_file): Always call dwarf2out_frame_finish
4222 if the target needs either debug or unwind DWARF2 info.
4223 * lto-opts.c (lto_write_options): Do not save -fexceptions,
4224 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
4225 -ftrapping-math, -ftrapv and -fwrapv.
4227 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4229 * cgraph.c (cgraph_edge::sreal_frequency): New function.
4230 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
4231 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
4232 (estimate_edge_size_and_time): Likewise.
4233 (ipa_merge_fn_summary_after_inlining): Likewise.
4234 * ipa-inline.c (cgraph_freq_base_rec): Remove.
4235 (compute_uninlined_call_time): Use sreal_frequency.
4236 (compute_inlined_call_time): Likewise.
4237 (ipa_inline): Do not initialize cgraph_freq_base_rec.
4238 * profile-count.c: Include sreal.h.
4239 (profile_count::to_sreal_scale): New.
4240 * profile-count.h: Forward declare sreal.
4241 (profile_count::to_sreal_scale): Declare.
4243 2017-11-13 Nathan Sidwell <nathan@acm.org>
4245 * diagnostic.c (maybe_line_and_column): New.
4246 (diagnostic_get_location_text): Use it.
4247 (diagnostic_report_current_module): Likewise.
4248 (test_diagnostic_get_location_text): Add tests.
4250 2017-11-13 Luis Machado <luis.machado@linaro.org>
4252 * doc/md.texi (Specifying processor pipeline description): Fix
4253 incorrect latency for the div instruction example.
4255 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4257 PR tree-optimization/78821
4258 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
4259 that bit_not_p is the same.
4260 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
4261 (split_group): Count precisely bit_not_p bits in each statement.
4262 (invert_op): New function.
4263 (imm_store_chain_info::output_merged_store): Use invert_op to
4264 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
4265 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
4267 2017-11-13 Martin Liska <mliska@suse.cz>
4269 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
4270 (struct source_info): Likewise.
4271 (add_branch_counts): Likewise.
4272 (add_line_counts): Likewise.
4273 (function_summary): Likewise.
4274 (output_intermediate_line): Likewise.
4275 (generate_results): Likewise.
4277 2017-11-13 Martin Liska <mliska@suse.cz>
4279 * gcov.c (struct block_info): Remove typedef for block_t.
4280 (struct line_info): Likewise.
4281 (line_info::has_block): Likewise.
4282 (EXIT_BLOCK): Likewise.
4283 (unblock): Likewise.
4284 (circuit): Likewise.
4285 (get_cycles_count): Likewise.
4286 (process_file): Likewise.
4287 (read_graph_file): Likewise.
4288 (solve_flow_graph): Likewise.
4289 (find_exception_blocks): Likewise.
4290 (add_line_counts): Likewise.
4291 (accumulate_line_info): Likewise.
4292 (output_line_details): Likewise.
4294 2017-11-13 Martin Liska <mliska@suse.cz>
4296 * gcov.c (struct arc_info): Remove typedef for arc_t.
4297 (struct line_info): Likewise.
4298 (add_branch_counts): Likewise.
4299 (output_branch_count): Likewise.
4300 (function_info::~function_info): Likewise.
4301 (circuit): Likewise.
4302 (output_intermediate_line): Likewise.
4303 (read_graph_file): Likewise.
4304 (solve_flow_graph): Likewise.
4305 (find_exception_blocks): Likewise.
4306 (add_line_counts): Likewise.
4307 (accumulate_line_info): Likewise.
4308 (output_line_details): Likewise.
4309 (output_function_details): Likewise.
4311 2017-11-13 Martin Liska <mliska@suse.cz>
4313 * gcov.c (struct function_info): Remove typedef for function_t.
4314 (struct source_info): Likewise.
4315 (source_info::get_functions_at_location): Likewise.
4316 (solve_flow_graph): Likewise.
4317 (find_exception_blocks): Likewise.
4318 (add_line_counts): Likewise.
4319 (output_intermediate_file): Likewise.
4320 (process_file): Likewise.
4321 (generate_results): Likewise.
4322 (release_structures): Likewise.
4323 (read_graph_file): Likewise.
4324 (read_count_file): Likewise.
4325 (accumulate_line_counts): Likewise.
4326 (output_lines): Likewise.
4328 2017-11-13 Martin Liska <mliska@suse.cz>
4330 * gcov.c (function_info::function_info): Remove num_counts
4331 and add vector<gcov_type>.
4332 (function_info::~function_info): Use the vector.
4333 (process_file): Likewise.
4334 (read_graph_file): Likewise.
4335 (read_count_file): Likewise.
4336 (solve_flow_graph): Likewise.
4338 2017-11-13 Martin Liska <mliska@suse.cz>
4340 * gcov.c (function_info::is_artificial): New function.
4341 (process_file): Erase all artificial early.
4342 (generate_results): Skip as all artificial are already
4345 2017-11-13 Martin Liska <mliska@suse.cz>
4347 * gcov.c (read_graph_file): Store to global vector of functions.
4348 (read_count_file): Iterate the vector.
4349 (process_file): Likewise.
4350 (generate_results): Likewise.
4351 (release_structures): Likewise.
4353 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4355 PR tree-optimization/82954
4356 * gimple-ssa-store-merging.c
4357 (imm_store_chain_info::coalesce_immediate_stores): If
4358 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
4360 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
4362 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
4363 Upddate call to ENDIAN_LANE_N.
4364 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
4365 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
4366 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
4367 and use aarch64_endian_lane_rtx.
4368 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
4370 2017-11-12 Tom de Vries <tom@codesourcery.com>
4372 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
4374 2017-11-12 Tom de Vries <tom@codesourcery.com>
4376 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
4378 2017-11-12 Tom de Vries <tom@codesourcery.com>
4380 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
4382 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4383 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4384 * defaults.h (ASM_OUTPUT_LABELREF): Same.
4386 2017-11-11 Martin Sebor <msebor@redhat.com>
4389 * doc/extend.texi (attribute nonstring): Remove spurious argument.
4392 * prefic.c (translate_name): Replace strncpy with memcpy to
4393 avoid -Wstringop-truncation.
4395 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4397 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
4399 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4401 * predict.c (maybe_hot_frequency_p): Do not use cfun.
4403 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4405 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
4408 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
4410 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
4411 enable generating XXBRH if the value is in a vector register.
4412 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
4413 value is in a vector register.
4414 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
4415 register to register bswap64's instead of doing the GPR sequence
4416 used on previous machines.
4417 (bswapdi2_xxbrd): New insn.
4418 (bswapdi2_reg): Disallow on ISA 3.0.
4419 (register to register bswap64 splitter): Do not split the insn on
4420 ISA 3.0 systems that use XXBRD.
4422 2017-11-10 Martin Sebor <msebor@redhat.com>
4425 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
4427 (find_subframework_file): Same.
4429 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4431 * auto-profile.c (afdo_indirect_call): Drop frequency.
4432 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
4433 (cgraph_node::create_edge): Drop frequency argument.
4434 (cgraph_node::create_indirect_edge): Drop frequency argument.
4435 (cgraph_edge::make_speculative): Drop frequency arguments.
4436 (cgraph_edge::resolve_speculation): Do not update frequencies
4437 (cgraph_edge::dump_edge_flags): Do not dump frequency.
4438 (cgraph_node::dump): Check consistency in IPA mode.
4439 (cgraph_edge::maybe_hot_p): Use IPA counter.
4440 (cgraph_edge::verify_count_and_frequency): Rename to ...
4441 (cgraph_edge::verify_count): ... this one; drop frequency checking.
4442 (cgraph_node::verify_node): Update.
4443 * cgraph.h (struct cgraph_edge): Drop frequency.
4444 (cgraph_edge::frequency): New function.
4445 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
4447 (cgraph_edge::rebuild_edges): Likewise.
4448 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
4449 (duplicate_thunk_for_node): Do not pass frequency.
4450 (cgraph_node::create_clone): Scale only counts.
4451 (cgraph_node::create_virtual_clone): Do not pass frequency.
4452 (cgraph_node::create_edge_including_clones): Do not pass frequency.
4453 (cgraph_node::create_version_clone): Do not pass frequency.
4454 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
4455 (cgraph_node::expand_thunk): Do not pass frequency.
4456 (cgraph_node::create_wrapper): Do not pass frequency.
4457 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
4459 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
4460 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
4461 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
4462 (gather_caller_stats): Use frequency function.
4463 (ipcp_cloning_candidate_p): Use frequency function.
4464 (ipcp_propagate_stage): Use frequency function.
4465 (get_info_about_necessary_edges): Use frequency function.
4466 (update_profiling_info): Update only IPA profile.
4467 (update_specialized_profile): Use frequency functoin.
4468 (perhaps_add_new_callers): Update only IPA profile.
4469 * ipa-devirt.c (ipa_devirt): Use IPA profile.
4470 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
4471 (dump_ipa_call_summary): Use frequency function.
4472 (estimate_edge_size_and_time): Use frequency function.
4473 (ipa_merge_fn_summary_after_inlining): Use frequency function.
4474 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
4475 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
4476 (update_noncloned_counts): ... ths one; scale counts only.
4477 (clone_inlined_nodes): Do not scale frequency.
4478 (inline_call): Do not pass frequency.
4479 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
4480 (compute_inlined_call_time): Use IPA profile.
4481 (want_inline_small_function_p): Use IPA profile.
4482 (want_inline_self_recursive_call_p): Use IPA profile.
4483 (edge_badness): Use IPA profile.
4484 (lookup_recursive_calls): Use IPA profile.
4485 (recursive_inlining): Do not pass frequency.
4486 (resolve_noninline_speculation): Do not update frequency.
4487 (inline_small_functions): Collect max of IPA profile.
4488 (dump_overall_stats): Dump IPA porfile.
4489 (dump_inline_stats): Dump IPA porfile.
4490 (ipa_inline): Collect IPA stats.
4491 * ipa-inline.h (clone_inlined_nodes): Update prototype.
4492 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
4493 (ipa_propagate_frequency): Use frequency function.
4494 (ipa_profile): Cleanup.
4495 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
4496 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
4497 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
4498 (input_node): Do not stream frequency.
4499 (input_edge): Do not stream frequency.
4500 (merge_profile_summaries): Scale only IPA profiles.
4501 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
4502 * predict.c (drop_profile): Do not recompute frequency.
4503 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
4504 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
4505 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
4506 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
4507 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
4508 * tree-inline.c (copy_bb): Do not scale frequency.
4509 (expand_call_inline): Do not scale frequency.
4510 (tree_function_versioning): Do not scale frequency.
4511 * ubsan.c (ubsan_create_edge): Do not pass frequency.
4513 2017-11-10 Julia Koval <julia.koval@intel.com>
4515 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
4516 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
4517 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
4518 (_mm256_maskz_gf2p8affine_epi64_epi8)
4519 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
4520 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
4521 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
4522 (__builtin_ia32_vgf2p8affineqb_v32qi)
4523 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
4524 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
4526 2017-11-10 Tamar Christina <tamar.christina@arm.com>
4530 (arm_option_override): Refactor.
4531 (arm_option_reconfigure_globals): New.
4532 (arm_options_perform_arch_sanity_checks): New.
4533 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
4535 (arm_options_perform_arch_sanity_checks): Likewise
4537 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
4539 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
4540 (power9-qpmul): New.
4541 * rs6000/rs6000.md ("type" attr): Add qmul.
4542 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
4543 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
4544 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
4546 2017-11-10 Martin Sebor <msebor@redhat.com>
4549 * builtins.c (compute_objsize): Handle arrays that
4550 compute_builtin_object_size likes to fail for. Make extern.
4551 * builtins.h (compute_objsize): Declare.
4552 (check_strncpy_sizes): New function.
4553 (expand_builtin_strncpy): Call check_strncpy_sizes.
4554 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
4555 -Wstringop-truncation.
4556 (gimple_fold_builtin_strncat): Same.
4557 * gimple.c (gimple_build_call_from_tree): Set call location.
4558 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
4559 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
4560 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
4561 (handle_builtin_strlen): Use strlen_to_stridx.
4562 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
4564 Use strlen_to_stridx.
4565 (pass_strlen::execute): Release strlen_to_stridx.
4566 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
4567 (-Wstringop-truncation): Document new option.
4569 2017-11-10 Martin Liska <mliska@suse.cz>
4571 PR gcov-profile/82702
4572 * gcov.c (main): Handle intermediate files in a different
4574 (get_gcov_intermediate_filename): New function.
4575 (output_gcov_file): Remove support of intermediate files.
4576 (generate_results): Allocate intermediate file.
4577 (release_structures): Clean-up properly fn_end.
4578 (output_intermediate_file): Start iterating with line 1.
4580 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4582 PR tree-optimization/82929
4583 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
4584 ops_swapped_p non-static data member.
4585 (store_immediate_info::store_immediate_info): Clear it.
4586 (imm_store_chain_info::coalesce_immediate_stores): If swapping
4587 ops set ops_swapped_p.
4588 (count_multiple_uses): Handle ops_swapped_p.
4590 2017-11-10 Martin Liska <mliska@suse.cz>
4592 * coverage.c (coverage_init): Stream information about
4593 support of has_unexecuted_blocks.
4594 * doc/gcov.texi: Document that.
4595 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
4596 * gcov.c (read_graph_file): Likewise.
4597 (output_line_beginning): Fix a small issue with
4600 2017-11-10 Bin Cheng <bin.cheng@arm.com>
4602 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
4603 reference of trivial component.
4605 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4608 * gimple-ssa-store-merging.c
4609 (pass_store_merging::terminate_all_aliasing_chains): For
4610 gimple_store_p stmts also call refs_output_dependent_p.
4612 PR rtl-optimization/82913
4613 * compare-elim.c (try_merge_compare): Punt if def_insn is not
4616 2017-11-09 Jeff Law <law@redhat.com>
4618 * vr-values.h: New file with vr_values class.
4619 * tree-vrp.c: Include vr-values.h
4620 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
4621 data objects into the vr_values class.
4622 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
4623 (get_value_range): Make it a member function within vr_values class.
4624 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
4625 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
4626 (op_with_constant_singleton_value_range): Likewise.
4627 (extract_range_for_var_from_comparison_expr): Likewise.
4628 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
4629 (extract_range_from_binary_expr): Likewise.
4630 (extract_range_from_unary_expr): Likewise.
4631 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
4632 (check_for_binary_op_overflow, extract_range_basic): Likewise.
4633 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
4634 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
4635 (compare_name_with_value, compare_names): Likewise.
4636 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
4637 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
4638 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
4639 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
4640 (extract_range_from_phi_node): Likewise.
4641 (simplify_truth_ops_using_ranges): Likewise.
4642 (simplify_div_or_mod_using_ranges): Likewise.
4643 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
4644 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
4645 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
4646 (simplify_float_conversion_using_ranges): Likewise.
4647 (simplify_internal_call_using_ranges): Likewise.
4648 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
4649 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
4650 poitner via x_vr_values for calls into gimple folder.
4651 (vrp_initialize_lattice): Make this the vr_values ctor.
4652 (vrp_free_lattice): Make this the vr_values dtor.
4653 (set_vr_value): New function.
4654 (class vrp_prop): Add vr_values data member. Add various member
4655 functions as well as member functions that delegate to vr_values.
4656 (check_array_ref): Make a member function within vrp_prop class.
4657 (search_for_addr_array, vrp_initialize): Likewise.
4658 (vrp_finalize): Likewise. Revamp to avoid direct access to
4659 vr_value, values_propagated, etc.
4660 (check_array_bounds): Extract vrp_prop class instance pointer from
4661 walk info structure. Use it to call member functions.
4662 (check_all_array_refs): Make a member function within vrp_prop class.
4663 Smuggle class instance pointer via walk info structure.
4664 (x_vr_values): New local static.
4665 (vrp_valueize): Use x_vr_values to get class instance.
4666 (vr_valueize_1): Likewise.
4667 (class vrp_folder): Add vr_values data member. Add various member
4668 functions as well as member functions that delegate to vr_values.
4669 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
4670 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
4671 class instance from vr_values. Use it to call member functions.
4672 (vrp_dom_walker): Add vr_values data member.
4673 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
4674 instance via x_vr_values.
4675 (identify_jump_threads): Accept vr_values as argument. Store
4676 it into the walker structure.
4677 (evrp_dom_walker): Add vr_values class data member. Add various
4679 (evrp_dom_walker::try_find_new_range): Use vr_values data
4680 member to access the memory allocator.
4681 (evrp_dom_walker::before_dom_children): Store vr_values class
4682 instance into the vrp_folder class.
4683 (evrp_dom_walker::push_value_range): Rework to avoid direct
4684 access to num_vr_values and vr_value.
4685 (evrp_dom_walker::pop_value_range): Likewise.
4686 (execute_early_vrp): Remove call to vrp_initialize_lattice.
4687 Use vr_values to get to dump_all_value_ranges member function.
4688 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
4689 and simplify_cond_using_ranges_2 via vrp_prop class instance.
4690 Pass vr_values class instance down to identify_jump_threads.
4691 Remove call to vrp_free_lattice.
4692 (debug_all_value_ranges): Remove.
4694 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
4695 (vrp_folder): Likewise.
4697 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
4698 Get it from the existing bitmap instead.
4699 (vrp_intersect_ranges_1): Likewise.
4701 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4703 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
4705 (store_immediate_info::store_immediate_info): Add bitnotp argument,
4706 set bit_not_p to it.
4707 (imm_store_chain_info::coalesce_immediate_stores): Break group
4708 if bit_not_p is different.
4709 (count_multiple_uses, split_group,
4710 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
4711 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
4712 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
4713 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
4714 to store_immediate_info ctor.
4716 2017-11-09 Jim Wilson <jimw@sifive.com>
4718 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
4719 (scan_prog_file): Likewise.
4721 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4723 * bb-reorder.c (max_entry_frequency): Remove.
4724 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
4725 connect_traces, push_to_next_round_p): Remove prototypes.
4726 (find_traces_1_round): Use counts only.
4727 (push_to_next_round_p): Likewise.
4728 (find_traces): Likewise.
4729 (rotate_loop): Likewise.
4730 (find_traces_1_round): Likewise.
4731 (connect_traces): Likewise.
4732 (edge_order): Likewise.
4734 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
4736 * config/arm/arm.c (output_return_instruction): Add comments to
4737 indicate requirement for cmse_nonsecure_entry return to account
4738 for the size of clearing instruction output here.
4739 (thumb_exit): Likewise.
4740 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
4741 return in hardfloat mode.
4743 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
4745 * config/rs6000/rs6000.c (machine_function): Add a bool,
4746 "toc_is_wrapped_separately".
4747 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
4748 if it wasn't explicitly set or unset, we are optimizing for speed, and
4749 doing separate shrink-wrapping.
4750 (rs6000_get_separate_components): Enable the TOC component if
4751 saving the TOC register in the prologue.
4752 (rs6000_components_for_bb): Handle the TOC component.
4753 (rs6000_emit_prologue_components): Store the TOC register where needed.
4754 (rs6000_set_handled_components): Mark TOC as handled, if handled.
4755 (rs6000_emit_prologue): Don't save the TOC if that is already done.
4757 2017-11-09 Martin Jambor <mjambor@suse.cz>
4759 * ipa-param-manipulation.c: New file.
4760 * ipa-param-manipulation.h: Likewise.
4761 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
4762 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
4763 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
4764 (ipa_parm_adjustment): Likewise.
4765 (ipa_parm_adjustment_vec): Likewise.
4766 (ipa_get_vector_of_formal_parms): Moved declaration to
4767 ipa-param-manipulation.h.
4768 (ipa_get_vector_of_formal_parm_types): Likewise.
4769 (ipa_modify_formal_parameters): Likewise.
4770 (ipa_modify_call_arguments): Likewise.
4771 (ipa_combine_adjustments): Likewise.
4772 (ipa_dump_param_adjustments): Likewise.
4773 (ipa_modify_expr): Likewise.
4774 (ipa_get_adjustment_candidate): Likewise.
4775 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
4776 ipa-param-manipulation.c.
4777 (ipa_get_vector_of_formal_parm_types): Likewise.
4778 (ipa_modify_formal_parameters): Likewise.
4779 (ipa_modify_call_arguments): Likewise.
4780 (ipa_modify_expr): Likewise.
4781 (get_ssa_base_param): Likewise.
4782 (ipa_get_adjustment_candidate): Likewise.
4783 (index_in_adjustments_multiple_times_p): Likewise.
4784 (ipa_combine_adjustments): Likewise.
4785 (ipa_dump_param_adjustments): Likewise.
4786 * tree-sra.c: Also include ipa-param-manipulation.h
4787 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
4790 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4791 Alan Hayward <alan.hayward@arm.com>
4792 David Sherwood <david.sherwood@arm.com>
4794 * doc/sourcebuild.texi (vect_masked_store): Document.
4796 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4797 Alan Hayward <alan.hayward@arm.com>
4798 David Sherwood <david.sherwood@arm.com>
4800 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
4802 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4803 Alan Hayward <alan.hayward@arm.com>
4804 David Sherwood <david.sherwood@arm.com>
4806 * doc/sourcebuild.texi (vect_variable_length): Document.
4808 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4809 Alan Hayward <alan.hayward@arm.com>
4810 David Sherwood <david.sherwood@arm.com>
4812 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
4814 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4815 Alan Hayward <alan.hayward@arm.com>
4816 David Sherwood <david.sherwood@arm.com>
4818 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
4820 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4821 Alan Hayward <alan.hayward@arm.com>
4822 David Sherwood <david.sherwood@arm.com>
4824 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
4825 previously undocumented selectors.
4826 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
4828 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4830 * doc/rtl.texi (const_vector): Say that elements can be
4831 const_wide_ints too.
4832 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
4833 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
4834 (gen_vec_duplicate): Use it instead of CONSTANT_P.
4835 * optabs.c (expand_vector_broadcast): Likewise.
4837 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4838 Alan Hayward <alan.hayward@arm.com>
4839 David Sherwood <david.sherwood@arm.com>
4841 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
4842 scaled index even if the unscaled address was invalid.
4843 Don't increase the complexity of using a scale in that case.
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/rtl.texi: Rewrite the subreg rules so that they partition
4850 the inner register into REGMODE_NATURAL_SIZE bytes rather than
4851 UNITS_PER_WORD bytes.
4852 * emit-rtl.c (validate_subreg): Divide subregs into blocks
4853 based on REGMODE_NATURAL_SIZE of the inner mode.
4854 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
4855 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
4856 * expmed.c (lowpart_bit_field_p): Divide the value up into
4857 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
4858 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
4859 whether something is likely to occupy more than one register.
4861 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4864 * ipa-inline-transform.c (update_noncloned_frequencies): Use
4865 profile_count::adjust_for_ipa_scaling.
4866 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
4867 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
4869 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
4871 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4873 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
4874 (split_group): Add total_orig and total_new arguments, estimate the
4875 number of statements related to the store group without store merging
4876 and with store merging.
4877 (imm_store_chain_info::output_merged_store): Adjust split_group
4878 callers, punt if estimated number of statements with store merging
4879 is not smaller than estimated number of statements without it.
4881 (handled_load): Remove has_single_use checks.
4882 (pass_store_merging::process_store): Likewise.
4884 2017-11-09 Richard Biener <rguenther@suse.de>
4886 PR tree-optimization/82902
4887 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
4889 2017-11-09 Martin Liska <mliska@suse.cz>
4892 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
4895 2017-11-09 Martin Liska <mliska@suse.cz>
4897 PR tree-optimization/82669
4898 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
4900 2017-11-09 Martin Liska <mliska@suse.cz>
4902 PR gcov-profile/48463
4903 * coverage.c (coverage_begin_function): Output also end locus
4904 of a function and information whether the function is
4906 * gcov-dump.c (tag_function): Parse and print the information.
4907 * gcov.c (INCLUDE_MAP): Add include.
4908 (INCLUDE_SET): Likewise.
4909 (struct line_info): Move earlier in the source file because
4910 of vector<line_info> in function_info structure.
4911 (line_info::line_info): Likewise.
4912 (line_info::has_block): Likewise.
4913 (struct source_info): Add new member index.
4914 (source_info::get_functions_at_location): New function.
4915 (function_info::group_line_p): New function.
4916 (output_intermediate_line): New function.
4917 (output_intermediate_file): Use the mentioned function.
4918 (struct function_start): New.
4919 (struct function_start_pair_hash): Likewise.
4920 (process_file): Add code that identifies group functions.
4921 Assign lines either to global or function scope.
4922 (generate_results): Skip artificial functions.
4923 (find_source): Assign index for each source file.
4924 (read_graph_file): Read new flag artificial and end_line.
4925 (add_line_counts): Assign it either to global of function scope.
4926 (accumulate_line_counts): Isolate core of the function to
4927 accumulate_line_info and call it for both function and global
4929 (accumulate_line_info): New function.
4930 (output_line_beginning): Fix GNU coding style.
4931 (print_source_line): New function.
4932 (output_line_details): Likewise.
4933 (output_function_details): Likewise.
4934 (output_lines): Iterate both source (global) scope and function
4936 (struct function_line_start_cmp): New class.
4937 * doc/gcov.texi: Reflect changes in documentation.
4939 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4942 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
4943 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
4944 and similarly for not instead of neg.
4946 2017-11-08 Andi Kleen <ak@linux.intel.com>
4948 * config/i386/i386.opt: Add -mforce-indirect-call.
4949 * config/i386/predicates.md: Check for flag_force_indirect_call.
4950 * doc/invoke.texi: Document -mforce-indirect-call
4952 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
4954 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
4956 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
4957 config/riscv/riscv.c (predict.h): New include.
4958 (riscv_slow_unaligned_access_p): No longer static.
4959 (riscv_block_move_straight): Add require.
4960 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
4962 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4965 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
4966 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
4967 nonimmediate_operand predicate for operand 1 instead of
4970 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4972 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
4974 * config/aarch64/constraints.md (Uml): New constraint.
4975 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
4978 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4980 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
4981 of two vec_duplicates into a vec_concat.
4983 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4985 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4986 Simplify vec_merge of vec_duplicate and vec_concat.
4987 * config/aarch64/constraints.md (Utq): New constraint.
4988 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
4991 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4993 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4994 Simplify vec_merge of vec_duplicate and const_vector.
4995 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
4997 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
4998 mode iterator. Update predicate on operand 1 to
4999 handle non-const_vec constants. Delete constraints.
5000 (*aarch64_combinez_be<mode>): Likewise for operand 2.
5002 2017-11-08 Jakub Jelinek <jakub@redhat.com>
5004 PR tree-optimization/78821
5005 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
5007 (store_operand_info::store_operand_info): Initialize it to false.
5008 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
5009 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
5010 store in the group, and if chain_info is non-NULL, to ignore altogether
5012 (compatible_load_p): Fail if bit_not_p does not match.
5013 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
5014 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
5015 (pass_store_merging::process_store): Adjust
5016 terminate_all_aliasing_chains calls to pass NULL in all current spots,
5017 call terminate_all_aliasing_chains newly when adding a store into
5018 a chain with non-NULL chain_info.
5020 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
5022 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
5024 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
5026 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
5028 (aarch64_layout_frame): Initialise emit_frame_chain.
5029 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
5030 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
5032 2017-11-08 Martin Liska <mliska@suse.cz>
5034 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
5035 of gimple_call_internal_p.
5037 2017-11-07 Tom de Vries <tom@codesourcery.com>
5039 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
5041 2017-11-07 Tom de Vries <tom@codesourcery.com>
5043 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5046 2017-11-08 Martin Liska <mliska@suse.cz>
5049 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
5051 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
5053 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
5055 2017-11-07 Jakub Jelinek <jakub@redhat.com>
5058 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
5059 (*cmp<mode>_ccz_1): New insn with $k alternative.
5062 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
5063 RTX_COMM_COMPARE as commutative as well.
5064 (ix86_binary_operator_ok): Formatting fix.
5065 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
5066 *<code><mode>3<mask_name><round_saeonly_name>,
5067 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
5068 *<s>mul<mode>3_highpart<mask_name>,
5069 *vec_widen_umult_even_v16si<mask_name>,
5070 *vec_widen_umult_even_v8si<mask_name>,
5071 *vec_widen_umult_even_v4si<mask_name>,
5072 *vec_widen_smult_even_v16si<mask_name>,
5073 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
5074 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
5075 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
5076 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
5077 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
5078 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
5079 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
5080 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
5081 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
5082 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
5083 ix86_binary_operator_ok. Formatting fixes.
5084 (*<plusminus_insn><mode>3<mask_name><round_name>,
5085 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
5088 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
5090 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
5091 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
5092 eq and ne if TARGET_ISEL.
5093 (cmp): New code_iterator.
5094 (UNS, UNSU_, UNSIK): New code_attrs.
5095 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
5096 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
5098 ("eq<mode>3"): ... this.
5099 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
5101 ("ne<mode>3"): ... this.
5103 2017-11-07 Julia Koval <julia.koval@intel.com>
5106 * common/config/i386/i386-common.c
5107 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
5108 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
5109 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
5110 * config/i386/i386.opt: Ditto.
5111 * config/i386/i386.c (ix86_target_string): Ditto.
5112 (ix86_option_override_internal): Ditto.
5113 (ix86_init_mpx_builtins): Move MPX to args2.
5114 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
5115 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
5116 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
5117 __builtin_ia32_vgf2p8affineinvqb_v32qi,
5118 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
5119 __builtin_ia32_vgf2p8affineinvqb_v16qi,
5120 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
5122 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
5125 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
5126 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
5127 (zero-extendsidi peephole2): Remove peephole.
5129 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
5132 * stor-layout.c: Include attribs.h.
5133 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
5134 explicit lookup of "aligned" attribute.
5136 2017-11-07 Andrew Waterman <andrew@sifive.com>
5138 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
5139 (riscv_expand_block_move): Likewise.
5140 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
5142 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
5143 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
5144 * config/riscv/riscv.c (riscv_block_move_straight): New function.
5145 (riscv_adjust_block_mem): Likewise.
5146 (riscv_block_move_loop): Likewise.
5147 (riscv_expand_block_move): Likewise.
5148 * config/riscv/riscv.md (movmemsi): New pattern.
5150 2017-11-07 Michael Clark <michaeljclark@mac.com>
5152 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
5153 (MUSL_DYNAMIC_LINKER): Likewise.
5155 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
5157 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
5158 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
5161 2017-11-07 Richard Biener <rguenther@suse.de>
5163 * match.pd: Fix build.
5165 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
5166 Jackson Woodruff <jackson.woodruff@arm.com>
5168 PR tree-optimization/71026
5169 * match.pd: Canonicalize negate in division.
5171 2017-11-07 Sudakshina Das <sudi.das@arm.com>
5174 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
5176 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5178 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
5179 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
5181 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5183 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
5184 non-scalar integral types.
5185 * match.pd (negate_expr_p): Handle MINUS_EXPR.
5186 (-(A-B), -(~A)): New transformations.
5188 2017-11-07 Tom de Vries <tom@codesourcery.com>
5190 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
5191 semicolon after "do {} while (0)".
5192 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5193 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5194 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5195 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5196 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5198 2017-11-07 Tom de Vries <tom@codesourcery.com>
5200 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
5201 after "do {} while (0)".
5202 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5203 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5204 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5205 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5206 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5208 2017-11-07 Tom de Vries <tom@codesourcery.com>
5211 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
5213 (arm_rtx_costs_internal): Add missing semicolon after
5214 HANDLE_NARROW_SHIFT_ARITH call.
5216 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5218 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
5219 disable isel if it was not set explicitly.
5221 2017-11-06 James Bowman <james.bowman@ftdichip.com>
5223 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
5224 (add_type_attribute) likewise.
5226 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
5228 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
5229 of struct ix86_frame.
5230 (ix86_initial_elimination_offset): Likewise.
5231 (ix86_expand_split_stack_prologue): Likewise.
5233 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
5235 * tree-vrp.h (enum value_range_type): Update stale comment.
5237 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5238 Alan Hayward <alan.hayward@arm.com>
5239 David Sherwood <david.sherwood@arm.com>
5241 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
5242 (aarch64_expand_vec_perm_const): Take the number of units too.
5243 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
5244 (aarch64_expand_vec_perm_const): Likewise.
5245 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
5246 (vec_perm<mode>): Update accordingly.
5248 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5249 Alan Hayward <alan.hayward@arm.com>
5250 David Sherwood <david.sherwood@arm.com>
5252 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
5253 Take the number of units too.
5254 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
5255 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
5256 but check for a vector mode before rather than after the call.
5257 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
5258 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
5259 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
5260 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
5261 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
5262 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
5263 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
5264 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
5265 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
5266 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
5267 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
5268 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
5269 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
5270 (aarch64_sqdmull2_n<mode>): Update accordingly.
5272 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5273 Alan Hayward <alan.hayward@arm.com>
5274 David Sherwood <david.sherwood@arm.com>
5276 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
5277 the number of units too.
5278 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
5279 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
5280 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
5281 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
5282 (vec_store_lanesxi<mode>): Update accordingly.
5284 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5285 Alan Hayward <alan.hayward@arm.com>
5286 David Sherwood <david.sherwood@arm.com>
5288 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
5289 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
5290 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
5291 of units rather than the mode.
5292 * config/aarch64/iterators.md (nunits): New mode attribute.
5293 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
5294 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
5295 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
5296 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
5297 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
5298 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
5299 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
5300 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
5301 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
5302 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
5303 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5304 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
5305 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
5306 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
5307 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
5308 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
5309 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
5310 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
5311 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
5312 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
5313 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
5314 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
5315 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
5316 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
5317 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
5318 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
5319 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
5320 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
5321 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
5322 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
5323 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
5324 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
5325 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
5326 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
5327 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
5328 (aarch64_simd_vec_setv2di): Likewise.
5330 2017-11-06 Carl Love <cel@us.ibm.com>
5332 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
5334 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
5335 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
5337 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
5339 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
5340 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
5341 Add power 8 macro expansions.
5342 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
5343 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
5344 power 8 instructions. (VSX_XXBR): Add iterator.
5346 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
5348 * config/arm/arm.md (predicable_short_it): Change default to "no",
5349 improve documentation, remove uses that are identical to the default.
5350 (enabled_for_depr_it): Rename to enabled_for_short_it.
5351 * gcc/config/arm/arm-fixed.md (predicable_short_it):
5352 Remove default uses.
5353 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
5354 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
5355 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
5356 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
5358 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
5361 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
5362 float128 helper macros, which are no longer used after deleting
5363 the old 'q' built-in functions, and moving the round to odd
5364 built-in functions to being special built-in functions.
5365 (BU_FLOAT128_2): Likewise.
5366 (BU_FLOAT128_1_HW): Likewise.
5367 (BU_FLOAT128_2_HW): Likewise.
5368 (BU_FLOAT128_3_HW): Likewise.
5369 (FABSQ): Delete old 'q' built-in functions.
5370 (COPYSIGNQ): Likewise.
5371 (SQRTF128_ODD): Move round to odd built-in functions to be
5372 special built-in functions, so that we can handle
5373 -mabi=ieeelongdouble.
5374 (TRUNCF128_ODD): Likewise.
5375 (ADDF128_ODD): Likewise.
5376 (SUBF128_ODD): Likewise.
5377 (MULF128_ODD): Likewise.
5378 (DIVF128_ODD): Likewise.
5379 (FMAF128_ODD): Likewise.
5380 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
5381 built-in names to 'f128'.
5382 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
5383 old 'q' built-in functions, as the machine independent code for
5384 'f128' built-in functions handles this.
5385 (rs6000_expand_builtin): Add expansion for float128 round to odd
5386 functions, keying off on -mabi=ieeelongdouble of whether to use
5387 the KFmode or TFmode variant.
5388 (rs6000_init_builtins): Initialize the _Float128 round to odd
5390 * doc/extend.texi (PowerPC Built-in Functions): Document the old
5391 _Float128 'q' built-in functions are now mapped into the new
5392 'f128' built-in functions.
5394 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
5396 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
5397 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
5399 2017-11-06 David Malcolm <dmalcolm@redhat.com>
5402 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
5403 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
5404 * toplev.c: Include "ipa-fnsummary.h".
5405 (toplev::finalize): Call ipa_fnsummary_c_finalize.
5407 2017-11-06 Jakub Jelinek <jakub@redhat.com>
5409 PR tree-optimization/82838
5410 * gimple-ssa-store-merging.c
5411 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
5412 on a separate gimple_seq which is then appended to seq.
5414 2017-11-06 Jeff Law <law@redhat.com>
5417 * config/i386/i386.c (PROBE_INTERVAL): Remove.
5418 (get_probe_interval): New functions.
5419 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
5420 (ix86_adjust_stack_and_probe): Likewise.
5421 (output_adjust_stack_and_probe): Likewise.
5422 (ix86_emit_probe_stack_range): Likewise.
5423 (ix86_expand_prologue): Likewise.
5425 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5427 PR tree-optimization/82816
5428 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
5429 if the modes of the two types are the same.
5430 (convert_plusminus_to_widen): Likewise.
5432 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5434 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
5436 (usadv16qi): New define_expand.
5437 (usadv8hi): New define_expand.
5439 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
5442 * ipa-inline-transform.c (update_noncloned_frequencies): Always
5444 (inline_transform): Likewise.
5445 * predict.c (counts_to_freqs): Remove useless conditional.
5446 * profile-count.h (profile_count::apply_scale): Move sanity check.
5447 * tree-inline.c (copy_bb): Always scale.
5448 (copy_cfg_body): Likewise.
5450 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
5453 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
5455 (*cmp_ite0): Add enabled_for_depr_it attribute.
5456 (*cmp_ite1): Likewise.
5458 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5460 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
5463 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5465 * tree-vrp.c (vrp_int_const_binop): Return true on success and
5466 return the value by pointer.
5467 (extract_range_from_multiplicative_op_1): Update accordingly.
5468 Return as soon as an operation fails.
5470 2017-11-05 Tom de Vries <tom@codesourcery.com>
5473 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
5474 (DEF_SANITIZER_BUILTIN): ... here.
5475 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
5476 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
5478 2017-11-05 Tom de Vries <tom@codesourcery.com>
5481 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
5483 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
5484 ASM_OUTPUT_BEFORE_CASE_LABEL call.
5485 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
5487 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5488 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5489 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5491 2017-11-05 Tom de Vries <tom@codesourcery.com>
5494 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
5497 2017-11-04 Michael Clark <michaeljclark@mac.com>
5499 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
5500 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
5502 (*addsi3_extended): Likewise.
5503 (*addsi3_extended2): Likewise.
5504 (<optab>si3): Likewise.
5505 (<optab>di3): Likewise.
5506 (<optab><mode>3): Likewise.
5507 (<*optabe>si3_internal): Likewise.
5508 (zero_extendqi<SUPERQI:mode>2): Likewise.
5509 (*add<mode>hi3): Likewise.
5510 (*xor<mode>hi3): Likewise.
5511 (<optab>di3): Likewise.
5512 (*<optab>si3_extend): Likewise.
5513 (*sge<u>_<X:mode><GPR:mode>): Likewise.
5514 (*slt<u>_<X:mode><GPR:mode>): Likewise.
5515 (*sle<u>_<X:mode><GPR:mode>): Likewise.
5517 2017-11-04 Andrew Waterman <andrew@sifive.com>
5519 * config/riscv/riscv.c (riscv_option_override): Conditionally set
5520 TARGET_STRICT_ALIGN based upon -mtune argument.
5522 2017-11-04 Andrew Waterman <andrew@sifive.com>
5524 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
5526 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
5528 * config/i386/i386.c (choose_basereg): Use optional scratch
5529 register and add assertion.
5530 (x86_emit_outlined_ms2sysv_save): Use scratch register when
5531 needed, and don't allocate stack.
5532 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
5533 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
5534 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
5536 2017-11-03 Jeff Law <law@redhat.com>
5538 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
5539 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
5540 to probe at the start of a noreturn function.
5542 2017-11-03 Jakub Jelinek <jakub@redhat.com>
5544 PR tree-optimization/78821
5545 * gimple-ssa-store-merging.c: Update the file comment.
5546 (MAX_STORE_ALIAS_CHECKS): Define.
5547 (struct store_operand_info): New type.
5548 (store_operand_info::store_operand_info): New constructor.
5549 (struct store_immediate_info): Add rhs_code and ops data members.
5550 (store_immediate_info::store_immediate_info): Add rhscode, op0r
5551 and op1r arguments to the ctor, initialize corresponding data members.
5552 (struct merged_store_group): Add load_align_base and load_align
5554 (merged_store_group::merged_store_group): Initialize them.
5555 (merged_store_group::do_merge): Update them.
5556 (merged_store_group::apply_stores): Pick the constant for
5557 encode_tree_to_bitpos from one of the two operands, or skip
5558 encode_tree_to_bitpos if neither operand is a constant.
5559 (class pass_store_merging): Add process_store method decl. Remove
5560 bool argument from terminate_all_aliasing_chains method decl.
5561 (pass_store_merging::terminate_all_aliasing_chains): Remove
5562 var_offset_p argument and corresponding handling.
5563 (stmts_may_clobber_ref_p): New function.
5564 (compatible_load_p): New function.
5565 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
5566 if there is overlap and rhs_code is not INTEGER_CST. For
5567 non-overlapping stores terminate group if rhs is not mergeable.
5568 (get_alias_type_for_stmts): Change first argument from
5569 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
5570 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
5571 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
5573 (get_location_for_stmts): Change first argument from
5574 auto_vec<gimple *> & to vec<gimple *> &.
5575 (struct split_store): Remove orig_stmts data member, add orig_stores.
5576 (split_store::split_store): Create orig_stores rather than orig_stmts.
5577 (find_constituent_stmts): Renamed to ...
5578 (find_constituent_stores): ... this. Change second argument from
5579 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
5580 to info structures rather than the statements.
5581 (split_group): Rename ALLOW_UNALIGNED argument to
5582 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
5583 it. Adjust find_constituent_stores caller.
5584 (imm_store_chain_info::output_merged_store): Handle rhs_code other
5585 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
5586 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
5587 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
5588 (mem_valid_for_store_merging): New function.
5589 (handled_load): New function.
5590 (pass_store_merging::process_store): New method.
5591 (pass_store_merging::execute): Use process_store method. Adjust
5592 terminate_all_aliasing_chains caller.
5594 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5596 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
5597 Return true for more constants, symbols and label references.
5598 (aarch64_valid_floating_const): Remove unused function.
5600 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5603 * config/aarch64/aarch64.c (aarch64_layout_frame):
5604 Undo forcing of LR at bottom of frame.
5606 2017-11-03 Jeff Law <law@redhat.com>
5609 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
5610 for int_registers_saved.
5612 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
5613 extracted from tree-ssa-dom.c.
5614 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
5615 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
5616 (record_equivalences_from_incoming_edge): Add additional argument
5617 to single_pred_edge_ignoring_loop_edges call.
5618 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
5619 (uncprop_dom_walker::before_dom_children): Add additional argument
5620 to single_pred_edge_ignoring_loop_edges call.
5621 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
5622 single_pred_edge_ignoring_loop_edges rather than open coding.
5623 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
5625 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
5627 * match.pd (-(-A)): Rewrite.
5629 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5631 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
5632 (rs6000_emit_int_cmove): New declaration.
5633 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
5634 (rs6000_emit_sISEL): Delete.
5635 (rs6000_emit_int_cmove): Make non-static.
5636 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
5637 instead of rs6000_emit_sISEL.
5639 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
5641 * asan.c (create_cond_insert_point): Maintain profile.
5642 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
5644 * basic-block.h (struct basic_block_def): Remove frequency.
5645 (EDGE_FREQUENCY): Use to_frequency
5646 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
5648 (find_traces): Update to use to_frequency.
5649 (find_traces_1_round): Likewise; use only IPA counts.
5650 (bb_to_key): Likewise.
5651 (connect_traces): Use IPA counts only.
5652 (copy_bb_p): Update to use to_frequency.
5653 (fix_up_crossing_landing_pad): Likewise.
5654 (sanitize_hot_paths): Likewise.
5655 * bt-load.c (basic_block_freq): Likewise.
5656 * cfg.c (init_flow): Set count_max to uninitialized.
5657 (check_bb_profile): Remove frequencies; check counts.
5658 (dump_bb_info): Do not dump frequencies.
5659 (update_bb_profile_for_threading): Update counts only.
5660 (scale_bbs_frequencies_int): Likewise.
5661 (MAX_SAFE_MULTIPLIER): Remove.
5662 (scale_bbs_frequencies_gcov_type): Update counts only.
5663 (scale_bbs_frequencies_profile_count): Update counts only.
5664 (scale_bbs_frequencies): Update counts only.
5665 * cfg.h (struct control_flow_graph): Add count-max.
5666 (update_bb_profile_for_threading): Update prototype.
5667 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
5668 (find_many_sub_basic_blocks): Likewise.
5669 * cfgcleanup.c (try_forward_edges): Likewise.
5670 (try_crossjump_to_edge): Likewise.
5671 * cfgexpand.c (expand_gimple_cond): Likewise.
5672 (expand_gimple_tailcall): Likewise.
5673 (construct_init_block): Likewise.
5674 (construct_exit_block): Likewise.
5675 * cfghooks.c (verify_flow_info): Check consistency of counts.
5676 (dump_bb_for_graph): Do not dump frequencies.
5677 (split_block_1): Do not update frequencies.
5678 (split_edge): Do not update frequencies.
5679 (make_forwarder_block): Do not update frequencies.
5680 (duplicate_block): Do not update frequencies.
5681 (account_profile_record): Do not update frequencies.
5682 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
5683 for global heuristics.
5684 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
5685 (expected_loop_iterations_unbounded): Use counts only.
5686 * cfgloopmanip.c (scale_loop_profile): Simplify.
5687 (create_empty_loop_on_edge): Simplify
5689 (duplicate_loop_to_header_edge): Simplify
5690 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
5691 (update_br_prob_note): Take care of removing note when profile
5693 (relink_block_chain): Do not dump frequency.
5694 (rtl_account_profile_record): Use to_frequency.
5695 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
5696 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
5697 (cgraph_update_edges_for_call_stmt_node): Likewise.
5698 (cgraph_edge::verify_count_and_frequency): Update.
5699 (cgraph_node::verify_node): Temporarily disable frequency verification.
5700 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
5701 to_cgraph_frequency.
5702 (cgraph_edge::rebuild_edges): Convert to ipa counts.
5703 * cgraphunit.c (init_lowered_empty_function): Do not initialize
5705 (cgraph_node::expand_thunk): Update profile.
5706 * except.c (dw2_build_landing_pads): Do not update frequency.
5707 * final.c (compute_alignments): Use to_frequency.
5708 (dump_basic_block_info): Do not dump frequency.
5709 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
5710 (dump_gimple_bb_header): Do not dump frequency.
5711 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
5713 * gimple-streamer-in.c (input_bb): Do not stream frequency.
5714 * gimple-streamer-out.c (output_bb): Do not stream frequency.
5715 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
5716 (init_before_recovery): Do not update frequency.
5717 (sched_create_recovery_edges): Do not update frequency.
5718 * hsa-gen.c (convert_switch_statements): Do not update frequency.
5719 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
5720 (ipa_cp_c_finalize): Set max_count to uninitialized.
5721 * ipa-fnsummary.c (get_minimal_bb): Use counts.
5722 (param_change_prob): Use counts.
5723 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
5725 * ipa-split.c (consider_split): Use to_frequency.
5726 (split_function): Use to_frequency.
5727 * ira-build.c (loop_compare_func): Likewise.
5728 (mark_loops_for_removal): Likewise.
5729 (mark_all_loops_for_removal): Likewise.
5730 * loop-doloop.c (doloop_modify): Do not update frequency.
5731 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
5733 * lto-streamer-in.c (input_function): Update count_max.
5734 * omp-expand.c (expand_omp_taskreg): Update count_max.
5735 * omp-simd-clone.c (simd_clone_adjust): Update profile.
5736 * predict.c (maybe_hot_frequency_p): Use to_frequency.
5737 (maybe_hot_count_p): Use ipa counts only.
5738 (maybe_hot_bb_p): Simplify.
5739 (maybe_hot_edge_p): Simplify.
5740 (probably_never_executed): Do not take frequency argument.
5741 (probably_never_executed_bb_p): Do not pass frequency.
5742 (probably_never_executed_edge_p): Likewise.
5743 (combine_predictions_for_bb): Check that profile is nonzero.
5744 (propagate_freq): Do not set frequency.
5745 (drop_profile): Simplify.
5746 (counts_to_freqs): Simplify.
5747 (expensive_function_p): Use to_frequency.
5748 (propagate_unlikely_bbs_forward): Simplify.
5749 (determine_unlikely_bbs): Simplify.
5750 (estimate_bb_frequencies): Add hack to silence graphite issues.
5751 (compute_function_frequency): Use ipa counts.
5752 (pass_profile::execute): Update.
5753 (rebuild_frequencies): Use counts only.
5754 (force_edge_cold): Use counts only.
5755 * profile-count.c (profile_count::dump): Dump new count types.
5756 (profile_count::differs_from_p): Check compatiblity.
5757 (profile_count::to_frequency): New function.
5758 (profile_count::to_cgraph_frequency): New function.
5759 * profile-count.h (struct function): Declare.
5760 (enum profile_quality): Add profile_guessed_local and
5761 profile_guessed_global0.
5762 (class profile_proability): Decrease number of bits to 29;
5763 update from_reg_br_prob_note and to_reg_br_prob_note.
5764 (class profile_count: Update comment; decrease number of bits
5765 to 61. Check compatibility.
5766 (profile_count::compatible_p): New private member function.
5767 (profile_count::ipa_p): New member function.
5768 (profile_count::operator<): Handle global zero correctly.
5769 (profile_count::operator>): Handle global zero correctly.
5770 (profile_count::operator<=): Handle global zero correctly.
5771 (profile_count::operator>=): Handle global zero correctly.
5772 (profile_count::nonzero_p): New member function.
5773 (profile_count::force_nonzero): New member function.
5774 (profile_count::max): New member function.
5775 (profile_count::apply_scale): Handle IPA scalling.
5776 (profile_count::guessed_local): New member function.
5777 (profile_count::global0): New member function.
5778 (profile_count::ipa): New member function.
5779 (profile_count::to_frequency): Declare.
5780 (profile_count::to_cgraph_frequency): Declare.
5781 * profile.c (OVERLAP_BASE): Delete.
5782 (compute_frequency_overlap): Delete.
5783 (compute_branch_probabilities): Do not use compute_frequency_overlap.
5784 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
5785 * sched-ebb.c (rank): Use counts only.
5786 * shrink-wrap.c (handle_simple_exit): Use counts only.
5787 (try_shrink_wrapping): Use counts only.
5788 (place_prologue_for_one_component): Use counts only.
5789 * tracer.c (find_best_predecessor): Use to_frequency.
5790 (find_trace): Use to_frequency.
5791 (tail_duplicate): Use to_frequency.
5792 * trans-mem.c (expand_transaction): Do not update frequency.
5793 * tree-call-cdce.c: Do not update frequency.
5794 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
5795 (gimple_merge_blocks): Likewise.
5796 (gimple_split_edge): Likewise.
5797 (gimple_duplicate_sese_region): Likewise.
5798 (gimple_duplicate_sese_tail): Likewise.
5799 (move_sese_region_to_fn): Likewise.
5800 (gimple_account_profile_record): Likewise.
5801 (insert_cond_bb): Likewise.
5802 * tree-complex.c (expand_complex_div_wide): Likewise.
5803 * tree-eh.c (lower_resx): Update profile.
5804 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
5806 (initialize_cfun): Do not initialize frequencies
5807 (freqs_to_counts): Delete.
5808 (copy_cfg_body): Ignore count parameter.
5809 (copy_body): Update.
5810 (expand_call_inline): Update count_max.
5811 (optimize_inline_calls): Update count_max.
5812 (tree_function_versioning): Update count_max.
5813 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
5814 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
5816 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
5817 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
5818 (try_peel_loop): Likewise.
5819 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
5821 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
5822 (tree_transform_and_unroll_loop): Do not use frequencies
5823 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
5824 Use reliable prediction only.
5825 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
5826 * tree-ssa-sink.c (select_best_block): Use to_frequency.
5827 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
5828 probability scaling.
5829 * tree-ssa-threadupdate.c (create_block_for_threading): Do
5830 not update frequency
5831 (any_remaining_duplicated_blocks): Likewise.
5832 (update_profile): Likewise.
5833 (estimated_freqs_path): Delete.
5834 (freqs_to_counts_path): Delete.
5835 (clear_counts_path): Delete.
5836 (ssa_fix_duplicate_block_edges): Likewise.
5837 (duplicate_thread_path): Likewise.
5838 * tree-switch-conversion.c (gen_inbound_check): Use counts.
5839 * tree-tailcall.c (decrease_profile): Do not update frequency.
5840 (eliminate_tail_call): Likewise.
5841 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
5842 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
5843 (optimize_mask_stores): Likewise.
5844 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
5845 * ubsan.c (ubsan_expand_null_ifn): Update profile.
5846 (ubsan_expand_ptr_ifn): Update profile.
5847 * value-prof.c (gimple_ic): Simplify.
5848 * value-prof.h (gimple_ic): Update prototype.
5849 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
5850 * ipa-inline.c (compute_uninlined_call_time): Be sure that
5852 (want_inline_self_recursive_call_p): Likewise.
5853 (resolve_noninline_speculation): Only cummulate defined counts.
5854 (inline_small_functions): Use nonzero_p.
5855 (ipa_inline): Do not access freed node.
5857 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5859 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5860 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
5862 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
5864 * config/riscv/riscv.c (riscv_legitimize_move): Handle
5865 non-legitimate address.
5867 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5869 * config/rs6000/rs6000.md (*lt0_disi): Delete.
5870 (*lt0_<mode>di, *lt0_<mode>si): New.
5872 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5874 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
5875 TARGET_PAIRED_FLOAT.
5877 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
5878 Jim Wilson <jim.wilson@linaro.org>
5880 * config/aarch64/aarch64-cores.def (saphira): New CPU.
5881 * config/aarch64/aarch64-tune.md: Regenerated.
5882 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
5883 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
5885 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
5887 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
5888 (arc_expand_prologue): Restore blink for millicode.
5889 * config/arc/linux.h (LINK_EH_SPEC): Defined.
5891 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
5894 * config/i386/i386.c (ix86_vector_duplicate_value): Use
5895 gen_vec_duplicate after forcing the scalar into a register.
5897 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
5899 * combine (try_combine): Print the insns input to try_combine to the
5902 2017-11-02 Steve Ellcey <sellcey@cavium.com>
5905 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
5906 Remove second argument from aarch64_process_target_attr call.
5907 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
5909 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
5911 (aarch64_handle_attr_arch): Remove second argument.
5912 (aarch64_handle_attr_cpu): Ditto.
5913 (aarch64_handle_attr_tune): Ditto.
5914 (aarch64_handle_attr_isa_flags): Ditto.
5915 (aarch64_process_one_target_attr): Ditto.
5916 (aarch64_process_target_attr): Ditto.
5917 (aarch64_option_valid_attribute_p): Remove second argument.
5918 on aarch64_process_target_attr call.
5920 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5922 * diagnostic.c: Include "selftest-diagnostic.h".
5923 (selftest::assert_location_text): New function.
5924 (selftest::test_diagnostic_get_location_text): New function.
5925 (selftest::diagnostic_c_tests): Call it.
5927 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5929 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
5930 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
5931 (class selftest::test_diagnostic_context): Move to...
5932 * selftest-diagnostic.c: New file.
5933 * selftest-diagnostic.h: New file.
5935 2017-11-02 James Bowman <james.bowman@ftdichip.com>
5937 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
5938 offset range for FT32B.
5939 * config/ft32/ft32.h: option "mcompress" enables relaxation.
5940 * config/ft32/ft32.md: Add TARGET_NOPM.
5941 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
5942 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
5944 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
5946 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
5948 2017-11-02 Jeff Law <law@redhat.com>
5950 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
5951 virtual keyword on FINAL OVERRIDE members.
5953 * tree-ssa-propagate.h (ssa_propagation_engine): Group
5954 virtuals together. Add virtual destructor.
5955 (substitute_and_fold_engine): Similarly.
5957 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
5959 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
5961 2017-11-02 Richard Biener <rguenther@suse.de>
5963 PR tree-optimization/82795
5964 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
5966 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5968 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
5970 (gcc_GAS_CHECK_FEATURE): Remove.
5971 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
5972 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
5973 * configure: Regenerate.
5975 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
5977 * config/arc/arc.c (hwloop_optimize): Account for empty
5980 2017-11-02 Richard Biener <rguenther@suse.de>
5983 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
5984 Truncate ARRAY_REF index and element size.
5986 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
5988 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
5990 2017-11-01 Jeff Law <law@redhat.com>
5992 * tree-ssa-ccp.c (ccp_folder): New class derived from
5993 substitute_and_fold_engine.
5994 (ccp_folder::get_value): New member function.
5995 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
5996 (ccp_fold_stmt): Remove prototype.
5997 (ccp_finalize): Call substitute_and_fold from the ccp_class.
5998 * tree-ssa-copy.c (copy_folder): New class derived from
5999 substitute_and_fold_engine.
6000 (copy_folder::get_value): Renamed from get_value.
6001 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
6002 * tree-vrp.c (vrp_folder): New class derived from
6003 substitute_and_fold_engine.
6004 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
6005 (vrp_folder::get_value): New member function.
6006 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
6007 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
6008 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
6009 provide a class interface to folder/substitute routines.
6010 (ssa_prop_fold_stmt_fn): Remove typedef.
6011 (ssa_prop_get_value_fn): Likewise.
6012 (subsitute_and_fold): Remove prototype.
6013 (replace_uses_in): Likewise.
6014 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
6015 Renamed from replace_uses_in. Call the virtual member function
6016 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
6017 (substitute_and_fold_dom_walker): Remove initialization of
6018 data member entries for calbacks. Add substitute_and_fold_engine
6019 member and initialize it.
6020 (substitute_and_fold_dom_walker::before_dom_children0: Use the
6021 member functions for get_value, replace_phi_args_in c
6022 replace_uses_in, and fold_stmt calls.
6023 (substitute_and_fold_engine::substitute_and_fold): Renamed from
6024 substitute_and_fold. Remove assert. Update ctor call.
6026 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
6027 (ssa_prop_visit_phi_fn): Likewise.
6028 (class ssa_propagation_engine): New class to provide an interface
6030 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
6032 (ssa_prop_visit_phi): Likewise.
6033 (ssa_propagation_engine::simulate_stmt): Moved into class.
6034 Call visit_phi/visit_stmt from the class rather than via
6035 file scoped static variables.
6036 (ssa_propagation_engine::simulate_block): Moved into class.
6037 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
6038 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
6039 set file scoped statics for the visit_stmt/visit_phi callbacks.
6040 * tree-complex.c (complex_propagate): New class derived from
6041 ssa_propagation_engine.
6042 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
6043 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
6044 (tree_lower_complex): Call ssa_propagate via the complex_propagate
6046 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
6047 ssa_propagation_engine.
6048 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
6049 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
6050 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
6051 * tree-ssa-copy.c (copy_prop): New class derived from
6052 ssa_propagation_engine.
6053 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
6054 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
6055 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
6056 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
6057 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
6058 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
6059 (execute_vrp): Call ssa_propagate from the vrp_prop class.
6061 2017-11-01 Jakub Jelinek <jakub@redhat.com>
6063 PR rtl-optimization/82778
6064 PR rtl-optimization/82597
6065 * compare-elim.c (struct comparison): Add in_a_setter field.
6066 (find_comparison_dom_walker::before_dom_children): Remove killed
6067 bitmap and df_simulate_find_defs call, instead walk the defs.
6068 Compute last_setter and initialize in_a_setter. Merge definitions
6069 with first initialization for a few variables.
6070 (try_validate_parallel): Use insn_invalid_p instead of
6071 recog_memoized. Return insn rather than just the pattern.
6072 (try_merge_compare): Fix up comment. Don't uselessly test if
6073 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
6075 (execute_compare_elim_after_reload): Remove df_chain_add_problem
6078 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6079 Alan Hayward <alan.hayward@arm.com>
6080 David Sherwood <david.sherwood@arm.com>
6082 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
6083 aarch64_hard_regno_nregs to get the number of registers
6086 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6087 Alan Hayward <alan.hayward@arm.com>
6088 David Sherwood <david.sherwood@arm.com>
6090 * config/aarch64/constraints.md (Upl): Rename to...
6092 * config/aarch64/aarch64.md
6093 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
6096 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6097 Alan Hayward <alan.hayward@arm.com>
6098 David Sherwood <david.sherwood@arm.com>
6100 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
6101 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
6104 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6105 Alan Hayward <alan.hayward@arm.com>
6106 David Sherwood <david.sherwood@arm.com>
6108 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
6109 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
6110 (aarch64_evpc_dup): Generate rtl direcly, rather than using
6112 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
6113 of a single element.
6114 * config/aarch64/iterators.md: Add a comment above the permute
6115 unspecs to say that they are generated directly by
6116 aarch64_expand_vec_perm_const.
6117 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
6119 2017-11-01 Nathan Sidwell <nathan@acm.org>
6121 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
6123 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
6125 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
6126 and medany code models, and describe what they do.
6128 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6130 Revert accidental duplicate:
6132 * combine.c (can_change_dest_mode): Reject changes in
6133 REGMODE_NATURAL_SIZE.
6135 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
6137 PR rtl-optimization/64682
6138 PR rtl-optimization/69567
6139 PR rtl-optimization/69737
6140 PR rtl-optimization/82683
6141 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
6142 register mentioned in the note, drop the note, unless it came from I3,
6143 in which case it should go to I3 again.
6145 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6147 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
6148 and return false if not.
6149 (clear_bytes_written_by, live_bytes_read): Update accordingly.
6151 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6153 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
6154 range is known to be empty.
6156 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6157 Alan Hayward <alan.hayward@arm.com>
6158 David Sherwood <david.sherwood@arm.com>
6160 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
6161 and CONST_VECTOR_NUNITS instead of computing the number of units from
6162 the byte sizes of the vector and element.
6163 (simplify_binary_operation_1): Likewise.
6164 (simplify_const_binary_operation): Likewise.
6165 (simplify_ternary_operation): Likewise.
6167 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6168 Alan Hayward <alan.hayward@arm.com>
6169 David Sherwood <david.sherwood@arm.com>
6171 * var-tracking.c (INT_MEM_OFFSET): Replace with...
6172 (int_mem_offset): ...this new function.
6173 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
6174 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
6175 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
6178 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6179 Alan Hayward <alan.hayward@arm.com>
6180 David Sherwood <david.sherwood@arm.com>
6182 * lower-subreg.c (interesting_mode_p): New function.
6183 (compute_costs, find_decomposable_subregs, decompose_register)
6184 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
6185 (resolve_clobber, dump_choices): Use it.
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 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
6193 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6194 Alan Hayward <alan.hayward@arm.com>
6195 David Sherwood <david.sherwood@arm.com>
6197 * alias.c (find_base_value, find_base_term): Only process integer
6198 truncations. Check the precision rather than the size.
6200 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6201 Alan Hayward <alan.hayward@arm.com>
6202 David Sherwood <david.sherwood@arm.com>
6204 * machmode.h (is_narrower_int_mode): New function
6205 * optabs.c (expand_float, expand_fix): Use it.
6206 * dwarf2out.c (rotate_loc_descriptor): Likewise.
6208 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6209 Alan Hayward <alan.hayward@arm.com>
6210 David Sherwood <david.sherwood@arm.com>
6212 * rtl.h (narrower_subreg_mode): New function.
6213 * ira-color.c (update_costs_from_allocno): Use it.
6215 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6216 Alan Hayward <alan.hayward@arm.com>
6217 David Sherwood <david.sherwood@arm.com>
6219 * optabs-query.h (convert_optab_p): New function, split out from...
6220 (convert_optab_handler): ...here.
6221 (widening_optab_handler): Delete.
6222 (find_widening_optab_handler): Remove permit_non_widening parameter.
6223 (find_widening_optab_handler_and_mode): Likewise. Provide an
6224 override that operates on mode class wrappers.
6225 * optabs-query.c (widening_optab_handler): Delete.
6226 (find_widening_optab_handler_and_mode): Remove permit_non_widening
6227 parameter. Assert that the two modes are the same class and that
6228 the "from" mode is narrower than the "to" mode. Use
6229 convert_optab_handler instead of widening_optab_handler.
6230 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
6231 instead of widening_optab_handler.
6232 * expr.c (expand_expr_real_2): Update calls to
6233 find_widening_optab_handler.
6234 * optabs.c (expand_widen_pattern_expr): Likewise.
6235 (expand_binop_directly): Take the insn_code as a parameter.
6236 (expand_binop): Only call find_widening_optab_handler for
6237 conversion optabs; use optab_handler otherwise. Update calls
6238 to find_widening_optab_handler and expand_binop_directly.
6239 Use convert_optab_handler instead of widening_optab_handler.
6240 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
6241 find_widening_optab_handler and use scalar_mode rather than
6243 (convert_plusminus_to_widen): Likewise.
6245 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6246 Alan Hayward <alan.hayward@arm.com>
6247 David Sherwood <david.sherwood@arm.com>
6249 * machmode.h (fixed_size_mode): New class.
6250 * rtl.h (get_pool_mode): Return fixed_size_mode.
6251 * gengtype.c (main): Add fixed_size_mode.
6252 * target.def (get_raw_result_mode): Return a fixed_size_mode.
6253 (get_raw_arg_mode): Likewise.
6254 * doc/tm.texi: Regenerate.
6255 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
6256 * targhooks.c (default_get_reg_raw_mode): Likewise.
6257 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
6258 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
6259 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
6260 (msp430_get_raw_result_mode): Likewise.
6261 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
6262 * dbxout.c (dbxout_parms): Require fixed-size modes.
6263 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
6264 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
6265 * omp-low.c (lower_oacc_reductions): Likewise.
6266 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
6267 (simplify_subreg): Update accordingly.
6268 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
6269 (force_const_mem): Update accordingly. Return NULL_RTX for modes
6270 that aren't fixed-size.
6271 (get_pool_mode): Return a fixed_size_mode.
6272 (output_constant_pool_2): Take a fixed_size_mode.
6274 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6275 Alan Hayward <alan.hayward@arm.com>
6276 David Sherwood <david.sherwood@arm.com>
6278 * doc/rtl.texi (vec_series): Document.
6279 (const): Say that the operand can be a vec_series.
6280 * rtl.def (VEC_SERIES): New rtx code.
6281 * rtl.h (const_vec_series_p_1): Declare.
6282 (const_vec_series_p): New function.
6283 * emit-rtl.h (gen_const_vec_series): Declare.
6284 (gen_vec_series): Likewise.
6285 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
6286 (gen_vec_series): Likewise.
6287 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
6288 * simplify-rtx.c (simplify_unary_operation): Handle negations
6290 (simplify_binary_operation_series): New function.
6291 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
6292 (test_vector_ops_series): New function.
6293 (test_vector_ops): Call it.
6294 * config/powerpcspe/altivec.md (altivec_lvsl): Use
6295 gen_const_vec_series.
6296 (altivec_lvsr): Likewise.
6297 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
6299 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6300 Alan Hayward <alan.hayward@arm.com>
6301 David Sherwood <david.sherwood@arm.com>
6303 * doc/rtl.texi (const): Update description of address constants.
6304 Say that vector constants are allowed too.
6305 * common.md (E, F): Use CONSTANT_P instead of checking for
6307 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
6308 checking for CONST_VECTOR.
6309 * expmed.c (make_tree): Use build_vector_from_val for a CONST
6311 * expr.c (expand_expr_real_2): Check for vector modes instead
6312 of checking for CONST_VECTOR.
6313 * rtl.h (const_vec_p): New function.
6314 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
6315 (unwrap_const_vec_duplicate): Handle them here too.
6317 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6318 David Malcolm <dmalcolm@redhat.com>
6319 Alan Hayward <alan.hayward@arm.com>
6320 David Sherwood <david.sherwood@arm.com>
6322 * rtl.h (vec_duplicate_p): New function.
6323 * selftest-rtl.c (assert_rtx_eq_at): New function.
6324 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
6325 (assert_rtx_eq_at): Declare.
6326 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
6327 * selftest-run-tests.c (selftest::run_tests): Call it.
6328 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
6329 (simplify_unary_operation_1): Recursively handle vector duplicates.
6330 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
6332 (simplify_subreg): Handle subregs of vector duplicates.
6333 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
6334 (selftest::simplify_rtx_c_tests): New functions.
6336 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6337 Alan Hayward <alan.hayward@arm.com>
6338 David Sherwood <david.sherwood@arm.com>
6340 * emit-rtl.h (gen_const_vec_duplicate): Declare.
6341 (gen_vec_duplicate): Likewise.
6342 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
6344 (gen_const_vector): ...here.
6345 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
6346 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
6347 whose elements are all equal.
6348 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
6349 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
6350 (simplify_relational_operation): Likewise.
6351 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
6353 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
6354 (aarch64_expand_vector_init): Likewise.
6355 * config/arm/arm.c (neon_vdup_constant): Likewise.
6356 (neon_expand_vector_init): Likewise.
6357 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
6358 (arm_block_set_unaligned_vect): Likewise.
6359 (arm_block_set_aligned_vect): Likewise.
6360 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
6361 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
6362 (expand_vec_perm_even_odd_pack): Likewise.
6363 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
6364 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
6365 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
6366 gen_const_vec_duplicate.
6367 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
6368 * config/mips/mips.c (mips_gen_const_int_vector): Use
6369 gen_const_vec_duplicate.
6370 (mips_expand_vector_init): Use CONST0_RTX.
6371 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
6372 (define_split): Use gen_const_vec_duplicate.
6373 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
6374 (define_split): Use gen_const_vec_duplicate.
6375 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
6376 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
6377 * config/spu/spu.c (spu_const): Likewise.
6379 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6380 Alan Hayward <alan.hayward@arm.com>
6381 David Sherwood <david.sherwood@arm.com>
6383 * combine.c (can_change_dest_mode): Reject changes in
6384 REGMODE_NATURAL_SIZE.
6386 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
6388 * configure.ac (--enable-libssp): New.
6389 (gcc_cv_libc_provides_ssp): Check for explicit setting before
6390 trying to determine target-specific default. Adjust indentation.
6391 * configure: Regenerated.
6392 * doc/install.texi (Configuration): Expand --disable-libssp
6395 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
6397 config/i386/i386.c (ix86_expand_epilogue): Correct stack
6400 2017-10-31 Martin Jambor <mjambor@suse.cz>
6403 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
6405 2017-10-31 David Malcolm <dmalcolm@redhat.com>
6407 * auto-profile.c (autofdo_source_profile::read): Use
6408 UNKNOWN_LOCATION rather than 0.
6409 * diagnostic-core.h (warning_at_rich_loc): Rename to...
6410 (warning_at): ...this overload.
6411 (warning_at_rich_loc_n): Rename to...
6412 (warning_n): ...this overload.
6413 (error_at_rich_loc): Rename to...
6414 (error_at): ...this overload.
6415 (pedwarn_at_rich_loc): Rename to...
6416 (pedwarn): ...this overload.
6417 (permerror_at_rich_loc): Rename to...
6418 (permerror): ...this overload.
6419 (inform_at_rich_loc): Rename to...
6420 (inform): ...this overload.
6421 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
6422 (diagnostic_n_impl_richloc): Rename to...
6423 (diagnostic_n_impl): ...this rich_location *-based decl.
6424 (inform_at_rich_loc): Rename to...
6425 (inform): ...this, and add an assertion.
6426 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
6427 (warning_at_rich_loc): Rename to...
6428 (warning_at): ...this, and add an assertion.
6429 (warning_at_rich_loc_n): Rename to...
6430 (warning_n): ...this, and add an assertion.
6431 (warning_n): Update location_t-based implementation for removal of
6432 location_t-based diagnostic_n_impl.
6433 (pedwarn_at_rich_loc): Rename to...
6434 (pedwarn): ...this, and add an assertion.
6435 (permerror_at_rich_loc): Rename to...
6436 (permerror): ...this, and add an assertion.
6437 (error_n): Update for removal of location_t-based diagnostic_n_impl.
6438 (error_at_rich_loc): Rename to...
6439 (error_at): ...this, and add an assertion.
6440 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
6441 (driver::do_spec_on_infiles): Likewise.
6442 * substring-locations.c (format_warning_va): Update for renaming
6443 of inform_at_rich_loc.
6445 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
6447 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
6448 _Float<N>X built-in functions so that the variant without the
6449 "__builtin_" prefix is only enabled for the GNU C and Objective C
6450 languages when they are in non-strict ANSI/ISO mode.
6451 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
6452 * target.def (floatn_builtin_p): Add a target hook to control
6453 whether _Float<N> and _Float<N>X built-in functions without the
6454 "__builtin_" prefix are enabled, and return true for C and
6455 Objective C in the default hook. Include langhooks.h in
6457 * targhooks.h (default_floatn_builtin_p): Likewise.
6458 * targhooks.c (default_floatn_builtin_p): Likewise.
6459 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
6460 floatn_builtin_p target hook.
6461 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
6463 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
6464 Eric Botcazou <ebotcazou@adacore.com>
6466 PR rtl-optimization/81803
6467 * lra-constraints.c (curr_insn_transform): Also reload the whole
6468 register for a strict subreg no wider than a word if this is for
6469 a WORD_REGISTER_OPERATIONS target.
6471 2017-10-31 Jason Merrill <jason@redhat.com>
6473 * gdbinit.in: Skip over inlines from timevar.h.
6475 2017-10-31 Martin Liska <mliska@suse.cz>
6477 * doc/gcov.texi: Document new option.
6478 * gcov.c (print_usage): Likewise print it.
6479 (process_args): Support the argument.
6480 (format_count): New function.
6481 (format_gcov): Use the function.
6483 2017-10-31 Martin Liska <mliska@suse.cz>
6485 * gcov.c (struct name_map): do not use typedef.
6486 Define operator== and operator<.
6487 (name_search): Remove.
6488 (name_sort): Remove.
6489 (main): Do not allocate names.
6490 (process_file): Add vertical space.
6491 (generate_results): Use std::find.
6492 (release_structures): Do not release memory.
6493 (find_source): Use std::find.
6495 2017-10-31 Martin Liska <mliska@suse.cz>
6497 * gcov.c (struct line_info): Remove it's typedef.
6498 (line_info::line_info): Add proper ctor.
6499 (line_info::has_block): Do not use a typedef.
6500 (struct source_info): Do not use typedef.
6501 (circuit): Likewise.
6502 (get_cycles_count): Likewise.
6503 (output_intermediate_file): Iterate via vector iterator.
6504 (add_line_counts): Use std::vector methods.
6505 (accumulate_line_counts): Likewise.
6506 (output_lines): Likewise.
6508 2017-10-31 Martin Liska <mliska@suse.cz>
6510 * gcov.c (struct source_info): Remove typedef.
6511 (source_info::source_info): Add proper ctor.
6512 (accumulate_line_counts): Use struct, not it's typedef.
6513 (output_gcov_file): Likewise.
6514 (output_lines): Likewise.
6515 (main): Do not allocate an array.
6516 (output_intermediate_file): Use size of vector container.
6517 (process_file): Resize the vector.
6518 (generate_results): Do not preallocate, use newly added vector
6520 (release_structures): Do not release sources.
6521 (find_source): Use vector methods.
6522 (add_line_counts): Do not use typedef.
6524 2017-10-31 Martin Liska <mliska@suse.cz>
6526 * doc/gcov.texi: Document that.
6527 * gcov.c (add_line_counts): Mark lines with a non-executed
6529 (output_line_beginning): Handle such lines.
6530 (output_lines): Pass new argument.
6531 (output_intermediate_file): Print it in intermediate format.
6533 2017-10-31 Martin Liska <mliska@suse.cz>
6535 * color-macros.h: New file.
6536 * diagnostic-color.c: Factor out color related to macros to
6538 * doc/gcov.texi: Document -k option.
6539 * gcov.c (INCLUDE_STRING): Include string.h.
6540 (print_usage): Add -k option.
6541 (process_args): Parse it.
6542 (pad_count_string): New function.
6543 (output_line_beginning): Likewise.
6544 (DEFAULT_LINE_START): New macro.
6545 (output_lines): Support color output.
6547 2017-10-31 Martin Liska <mliska@suse.cz>
6549 PR gcov-profile/82633
6550 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
6551 their interaction with GCOV infrastructure.
6552 * configure.ac: Add -fkeep-{inline,static}-functions to
6554 * configure: Regenerate.
6556 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
6559 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
6561 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
6564 * config/rs6000/rs6000.md (allocate_stack): Force update interval
6565 into a register if it does not fit into an immediate offset field.
6567 2017-10-31 Olivier Hainque <hainque@adacore.com>
6569 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
6571 2017-10-31 Julia Koval <julia.koval@intel.com>
6573 * config.gcc: Add gfniintrin.h.
6574 * config/i386/gfniintrin.h: New.
6575 * config/i386/i386-builtin-types.def
6576 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
6577 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
6578 __builtin_ia32_vgf2p8affineinvqb_v32qi,
6579 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
6580 __builtin_ia32_vgf2p8affineinvqb_v16qi,
6581 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
6582 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
6583 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
6584 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
6585 V64QI_FTYPE_V64QI_V64QI_INT): New types.
6586 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
6587 * config/i386/immintrin.h: Include gfniintrin.h.
6588 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
6590 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
6592 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
6594 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
6596 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
6597 (arm_ashldi3_1bit): Remove pattern.
6598 (ashrdi3): Remove shift by 1 expansion.
6599 (arm_ashrdi3_1bit): Remove pattern.
6600 (lshrdi3): Remove shift by 1 expansion.
6601 (arm_lshrdi3_1bit): Remove pattern.
6602 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
6603 cost of ashldi3 by 1.
6604 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
6605 (<shift>di3_neon): Likewise.
6607 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6609 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
6610 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
6611 and (*aarch64_simd_mov<VQ:mode>).
6612 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
6613 pattern alternative.
6614 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
6615 attributes to match pattern alternative.
6617 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
6619 * config.gcc (powerpc*-*-*): Add emmintrin.h.
6620 * config/rs6000/emmintrin.h: New file.
6621 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
6623 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
6625 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
6626 * (movdi_vfp_cortexa8): Remove pattern.
6628 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6630 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
6632 (Specific, alpha*-dec-osf5.1): Remove.
6633 (Specific, mips-sgi-irix5): Remove.
6634 (Specific, mips-sgi-irix6): Remove.
6636 2017-10-30 Jakub Jelinek <jakub@redhat.com>
6639 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
6640 arguments to clear_bit_region_be.
6642 2017-10-30 Jim Wilson <wilson@tuliptree.org>
6644 * gimplify.c: Include memmodel.h.
6646 2017-10-30 Martin Jambor <mjambor@suse.cz>
6648 * omp-grid.c (grid_attempt_target_gridification): Also insert a
6649 condition whether loop should be executed at all.
6651 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
6653 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6654 gimple folding of vec_madd() intrinsics.
6655 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
6656 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
6657 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
6659 2017-10-30 Richard Biener <rguenther@suse.de>
6661 PR tree-optimization/82762
6663 2017-10-23 Richard Biener <rguenther@suse.de>
6665 PR tree-optimization/82129
6667 2017-08-01 Richard Biener <rguenther@suse.de>
6669 PR tree-optimization/81181
6670 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6671 (compute_antic): ... end of iteration here.
6673 2017-10-30 Joseph Myers <joseph@codesourcery.com>
6675 * doc/invoke.texi (C Dialect Options): Document -std=c17,
6676 -std=iso9899:2017 and -std=gnu17.
6677 * doc/standards.texi (C Language): Document C17 support.
6678 * doc/cpp.texi (Overview): Mention -std=c17.
6679 (Standard Predefined Macros): Document C11 and C17 values of
6680 __STDC_VERSION__. Do not refer to C99 support as incomplete.
6681 * doc/extend.texi (Inline): Do not list individual options for
6682 standards newer than C99.
6683 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
6685 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
6688 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
6690 * asan.c (asan_finish_file): Align asan globals array by shadow
6693 2017-10-30 Jakub Jelinek <jakub@redhat.com>
6696 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
6697 (struct store_immediate_info): Add bitregion_start and bitregion_end
6699 (store_immediate_info::store_immediate_info): Add brs and bre
6700 arguments and initialize bitregion_{start,end} from those.
6701 (struct merged_store_group): Add bitregion_start, bitregion_end,
6702 align_base and mask fields. Drop unnecessary struct keyword from
6703 struct store_immediate_info. Add do_merge method.
6704 (clear_bit_region_be): Use memset instead of loop storing zeros.
6705 (merged_store_group::do_merge): New method.
6706 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
6707 stores as long as the surrounding bitregions have no gaps.
6708 (merged_store_group::merge_overlapping): Use do_merge.
6709 (merged_store_group::apply_stores): Test that bitregion_{start,end}
6710 is byte aligned, rather than requiring that start and width are
6711 byte aligned. Drop unnecessary struct keyword from
6712 struct store_immediate_info. Allocate and populate also mask array.
6713 Make start of the arrays relative to bitregion_start rather than
6714 start and size them according to bitregion_{end,start} difference.
6715 (struct imm_store_chain_info): Drop unnecessary struct keyword from
6716 struct store_immediate_info.
6717 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
6718 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
6719 struct keyword from struct store_immediate_info.
6720 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
6721 between stores as long as the surrounding bitregions have no gaps.
6723 (struct split_store): Add orig non-static data member.
6724 (split_store::split_store): Initialize orig to false.
6725 (find_constituent_stmts): Return store_immediate_info *, non-NULL
6726 if there is exactly a single original stmt. Change stmts argument
6727 to pointer from reference, if NULL, don't push anything to it. Add
6728 first argument, use it to optimize skipping over orig stmts that
6729 are known to be before bitpos already. Simplify.
6730 (split_group): Return unsigned int count how many stores are or
6731 would be needed rather than a bool. Add allow_unaligned argument.
6732 Change split_stores argument from reference to pointer, if NULL,
6733 only do a dry run computing how many stores would be produced.
6734 Rewritten algorithm to use both alignment and misalign if
6735 !allow_unaligned and handle bitfield stores with gaps.
6736 (imm_store_chain_info::output_merged_store): Set start_byte_pos
6737 from bitregion_start instead of start. Compute allow_unaligned
6738 here, if true, do 2 split_group dry runs to compute which one
6739 produces fewer stores and prefer aligned if equal. Punt if
6740 new count is bigger or equal than original before emitting any
6741 statements, rather than during that. Remove no longer needed
6742 new_ssa_names tracking. Replace num_stmts with
6743 split_stores.length (). Use 32-bit stack allocated entries
6744 in split_stores auto_vec. Try to reuse original store lhs/rhs1
6745 if possible. Handle bitfields with gaps.
6746 (pass_store_merging::execute): Ignore bitsize == 0 stores.
6747 Compute bitregion_{start,end} for the stores and construct
6748 store_immediate_info with that. Formatting fixes.
6750 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
6753 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
6754 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
6756 2017-10-29 Jim Wilson <wilson@tuliptree.org>
6758 * gimplify.c: Include tm_p.h.
6760 * common.opt (gcoff): Re-add as ignored option.
6761 (gcoff1, gcoff2, gcoff3): Likewise.
6763 * Makefile.in (OBJS): Delete sdbout.o.
6764 (GTFILES): Delete $(srcdir)/sdbout.c.
6765 * debug.h: Delete sdb_debug_hooks.
6766 * final.c: Delete sdbout.h include.
6767 (final_scan_insn): Delete SDB_DEBUG check.
6768 (rest_of_clean_state): Likewise.
6769 * output.h: Delete sdb_begin_function_line.
6772 * toplev.c: Delete sdbout.h include.
6773 (process_options): Delete SDB_DEBUG check.
6774 * tree-core.h (tree_type_common): Delete pointer field of
6776 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
6777 TYPE_SYMTAB_POINTER.
6778 * tree.h (TYPE_SYMTAB_POINTER): Delete.
6779 (TYPE_SYMTAB_IS_POINTER): Delete.
6780 (TYPE_SYMTAB_IS_DIE): Renumber.
6781 * xcoffout.c: Refer to former sdbout.c file.
6782 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
6784 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
6785 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
6786 (Debugging Options): Delete -gcoff.
6787 (-gstabs): Delete SDB reference.
6789 (-gcoff@var{level}): Delete.
6790 * doc/passes.texi (Debugging information output): Delete SDB and
6791 sdbout.c references.
6792 * doc/tm.texi: Regenerate.
6793 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
6794 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
6796 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
6797 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
6798 (SDB_DEBUGGING_INFO): Delete.
6799 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
6800 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
6801 * target.def (output_source_filename): Delete COFF reference.
6803 * common.opt (gcoff): Delete.
6804 (gxcoff+): Update Negative chain.
6805 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
6807 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
6808 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
6809 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
6811 (expand_function_start): Change sdb reference to past tense.
6812 (expand_function_end): Change sdb reference to past tense.
6813 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
6814 * opts.c (debug_type_names): Delete coff entry.
6815 (common_handle_option): Delete OPT_gcoff case.
6816 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
6818 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
6819 * config/cris/cris.h: Delete SDB reference in comment.
6820 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
6821 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
6822 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
6823 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
6825 (ix86_expand_prologue): Likewise.
6826 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
6827 * config/ia64/ia64.h: Likewise.
6828 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
6829 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
6831 * config/mmix/mmix.h: Likewise.
6832 * config/nds32/nds32.c: Likewise.
6833 * config/stormy/storym16.h: Likewise.
6834 * config/visium/visium.h: Likewise.
6835 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
6837 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
6839 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
6840 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
6841 FRAME_POINTER_REGNUM point at high end of local var area.
6843 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
6845 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
6846 Move comment around. Do not reset best_edge for a copiable
6847 destination if the copy would cause a partition change.
6848 (better_edge_p): Remove redundant check.
6850 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6852 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
6854 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
6856 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
6857 for math functions that have _Float<N> and _Float<N>X variants.
6858 (mathfn_built_in_2): Add support for math functions that have
6859 _Float<N> and _Float<N>X variants.
6860 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
6861 (expand_builtin_mathfn_ternary): Add support for fma with
6862 _Float<N> and _Float<N>X variants.
6863 (expand_builtin): Likewise.
6864 (fold_builtin_3): Likewise.
6865 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
6866 create math function _Float<N> and _Float<N>X variants as external
6868 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
6869 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
6870 the __builtin_ prefix and if not strict ansi, without the prefix.
6871 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
6872 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
6873 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
6874 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
6875 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
6876 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
6877 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
6878 function signatures for fma _Float<N> and _Float<N>X variants.
6879 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
6880 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
6881 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
6882 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
6883 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
6884 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
6885 * gencfn-macros.c (print_case_cfn): Add support for math functions
6886 that have _Float<N> and _Float<N>X variants.
6887 (print_define_operator_list): Likewise.
6888 (fltfn_suffixes): Likewise.
6890 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
6891 for math functions that have _Float<N> and _Float<N>X variants.
6892 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
6893 and _Float<N>X variants.
6894 (COPYSIGN): Likewise.
6897 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
6898 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
6900 (integer_valued_read_call_p): Likewise.
6901 * fold-const-call.c (fold_const_call_ss): Likewise.
6902 (fold_const_call_sss): Add support for copysign, fmin, and fmax
6903 _Float<N> and _Float<N>X variants.
6904 (fold_const_call_ssss): Add support for fma _Float<N> and
6905 _Float<N>X variants.
6906 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
6907 support for copysign and fma _Float<N> and _Float<N>X variants.
6908 (backprop::process_builtin_call_use): Likewise.
6909 * tree-call-cdce.c (can_test_argument_range); Add support for
6910 sqrt _Float<N> and _Float<N>X variants.
6911 (edom_only_function): Likewise.
6912 (get_no_error_domain): Likewise.
6913 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
6914 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
6915 copysign _Float<N> and _Float<N>X variants.
6916 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
6917 handled by machine independent code.
6918 (FMAF128): Likewise.
6919 * doc/cpp.texi (Common Predefined Macros): Document defining
6920 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
6921 fma _Float<N> and _Float<N>X variants.
6923 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6926 * config/i386/i386-modes.def (CCFPU): Remove definition.
6927 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
6928 (ix86_cc_modes_compatible): Ditto.
6929 (ix86_expand_carry_flag_compare): Ditto.
6930 (ix86_expand_int_movcc): Ditto.
6931 (ix86_expand_int_addcc): Ditto.
6932 (ix86_reverse_condition): Ditto.
6933 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
6934 Return true/false for unordered/ordered fp comparisons.
6935 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
6936 (ix86_prepare_fp_compare_args): Update for rename.
6937 (ix86_expand_fp_compare): Update for rename. Generate unordered
6938 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
6939 (ix86_expand_sse_compare_and_jump): Ditto.
6940 * config/i386/predicates.md (fcmov_comparison_operator):
6941 Remove CCFPU mode handling.
6942 (ix86_comparison_operator): Ditto.
6943 (ix86_carry_flag_operator): Ditto.
6944 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
6945 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
6946 (*cmpu<mode>_cc_i387): Ditto.
6947 (FPCMP): Remove mode iterator.
6948 (unord): Remove mode attribute.
6949 (unord_subst): New define_subst transformation
6950 (unord): New define_subst attribute.
6952 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
6953 (*cmpi<unord>xf_i387): Ditto.
6954 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
6955 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
6956 using unord_subst transformation.
6957 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
6958 (round_saeonly): Also handle CCFP mode.
6959 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
6960 Remove UNSPEC_SAHF unspec handling.
6962 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6964 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
6966 2017-10-27 Jeff Law <law@redhat.com>
6968 * gimple-ssa-sprintf.c: Include domwalk.h.
6969 (class sprintf_dom_walker): New class, derived from dom_walker.
6970 (sprintf_dom_walker::before_dom_children): New function.
6971 (struct call_info): Moved into sprintf_dom_walker class
6972 (compute_formath_length, handle_gimple_call): Likewise.
6973 (sprintf_length::execute): Call the dominator walker rather
6974 than walking the statements.
6976 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
6977 gimple statement locations.
6978 (check_array_bounds): Corresponding changes. Get the statement's
6979 location directly from wi->stmt.
6981 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
6984 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
6986 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6988 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
6989 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
6991 2017-10-27 Jakub Jelinek <jakub@redhat.com>
6994 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
6995 * config/i386/i386.c (maybe_get_pool_constant): Removed.
6996 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
6997 maybe_get_pool_constant.
6998 * config/i386/predicates.md (zero_extended_scalar_load_operand):
7001 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7003 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
7004 2.26 caveat. Update gas and gld versions.
7005 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
7008 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7010 * cgraph.h (set_malloc_flag): Declare.
7011 * cgraph.c (set_malloc_flag_1): New function.
7012 (set_malloc_flag): Likewise.
7013 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
7014 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
7016 (read_ipa_call_summary): Add support for reading is_return_callee.
7017 (write_ipa_call_summary): Stream is_return_callee.
7018 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
7019 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
7020 ipa-prop.h, ipa-fnsummary.h.
7021 (pure_const_names): Change to static.
7022 (malloc_state_e): Define.
7023 (malloc_state_names): Define.
7024 (funct_state_d): Add field malloc_state.
7025 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
7026 (check_retval_uses): New function.
7027 (malloc_candidate_p): Likewise.
7028 (analyze_function): Add support for malloc attribute.
7029 (pure_const_write_summary): Stream malloc_state.
7030 (pure_const_read_summary): Add support for reading malloc_state.
7031 (dump_malloc_lattice): New function.
7032 (propagate_malloc): New function.
7033 (warn_function_malloc): New function.
7034 (ipa_pure_const::execute): Call propagate_malloc and
7035 ipa_free_fn_summary.
7036 (pass_local_pure_const::execute): Add support for malloc attribute.
7037 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
7038 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
7040 2017-10-27 Martin Liska <mliska@suse.cz>
7042 PR gcov-profile/82457
7043 * doc/invoke.texi: Document that one needs a non-strict ISO mode
7044 for fork-like functions to be properly instrumented.
7046 2017-10-27 Richard Biener <rguenther@suse.de>
7049 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
7050 info when we redirected EH.
7052 2017-10-26 Michael Collison <michael.collison@arm.com>
7054 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
7056 (<optab>_trunchf<GPI:mode>2: New pattern.
7057 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
7058 * config/aarch64/iterators.md (wv): New mode attribute.
7059 (vf, VF): New mode attributes.
7060 (vgp, VGP): New mode attributes.
7061 (s): Update attribute with SImode and DImode prefixes.
7063 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
7065 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
7066 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
7067 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
7068 (nios2_option_overide): Initialize it. Don't allow R0-relative
7069 addressing with PIC.
7070 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
7071 (nios2_symbolic_constant_p): Likewise.
7072 (nios2_legitimate_address_p): Likewise.
7073 (nios2_r0rel_section_name_p): New.
7074 (nios2_symbol_ref_in_r0rel_data_p): New.
7075 (nios2_emit_move_sequence): Handle r0rel_constant_p.
7076 (r0rel_constant_p): New.
7077 (nios2_print_operand_address): Handle r0rel_constant_p.
7078 (nios2_cdx_narrow_form_p): Likewise.
7079 * config/nios2/nios2.opt (mr0rel-sec=): New option.
7080 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
7081 (Nios II Options): Document -mr0rel-sec.
7083 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
7085 * config/nios2/nios2.c: Include xregex.h.
7086 (nios2_gprel_sec_regex): New.
7087 (nios2_option_overide): Initialize it. Don't allow GP-relative
7088 addressing with PIC.
7089 (nios2_small_section_name_p): Check for regex match.
7090 * config/nios2/nios2.opt (mgprel-sec=): New option.
7091 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
7092 (Nios II Options): Document -mgprel-sec.
7094 2017-10-26 Jim Wilson <wilson@tuliptree.org>
7096 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
7098 2017-10-26 Tom de Vries <tom@codesourcery.com>
7100 PR tree-optimization/82707
7101 * gimple.c (gimple_copy): Fix unsharing of
7102 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
7104 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
7106 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
7107 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
7108 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
7109 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
7110 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
7111 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
7112 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
7113 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
7114 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
7115 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
7116 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
7117 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
7118 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
7119 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
7120 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
7121 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
7122 _mm512_mask_cmpunord_ps_mask): New intrinsics.
7124 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7126 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
7128 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
7129 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
7130 warning to rs6000.c. Remove the Undocumented flag, since it has
7132 (-mabi=ibmlongdouble): Likewise.
7133 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
7134 already set, set the default format for long double.
7135 (rs6000_debug_reg_global): Print whether long double is IBM or
7137 (rs6000_option_override_internal): Rework setting long double
7138 format. Only warn if the user is changing the long double default
7139 and they did not use -Wno-psabi.
7140 * doc/invoke.texi (PowerPC options): Update the documentation for
7141 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
7143 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7144 Alan Hayward <alan.hayward@arm.com>
7145 David Sherwood <david.sherwood@arm.com>
7147 * rtl.h (wider_subreg_mode): New function.
7148 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
7149 rather than an unsigned int *.
7150 * ira-color.c (regno_max_ref_width): Replace with...
7151 (regno_max_ref_mode): ...this new variable.
7152 (coalesced_pseudo_reg_slot_compare): Update accordingly.
7153 Use wider_subreg_mode.
7154 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
7155 rather than an unsigned int *.
7156 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
7157 (process_alt_operands): Likewise.
7158 (invariant_p): Likewise.
7159 * lra-spills.c (assign_mem_slot): Likewise.
7160 (add_pseudo_to_slot): Likewise.
7161 * lra.c (collect_non_operand_hard_regs): Likewise.
7162 (add_regs_to_insn_regno_info): Likewise.
7163 * reload1.c (regno_max_ref_width): Replace with...
7164 (regno_max_ref_mode): ...this new variable.
7165 (reload): Update accordingly. Update call to
7166 ira_sort_regnos_for_alter_reg.
7167 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
7168 (init_eliminable_invariants): Update to use regno_max_ref_mode.
7169 (scan_paradoxical_subregs): Likewise.
7171 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7173 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
7174 (aarch64_frame): Add emit_frame_chain boolean.
7175 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7176 Move eh_return case to aarch64_layout_frame.
7177 (aarch64_layout_frame): Initialize emit_frame_chain.
7178 (aarch64_expand_prologue): Use emit_frame_chain.
7180 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7182 * config/aarch64/aarch64.c (aarch64_layout_frame):
7183 Ensure LR is always stored at the bottom of the callee-saves.
7184 Remove rarely used frame layout which saves callee-saves at top of
7185 frame, so the store of LR can be used as a valid probe in all cases.
7187 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7189 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
7190 Improve unaligned TImode/TFmode base/offset split.
7192 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7193 Alan Hayward <alan.hayward@arm.com>
7194 David Sherwood <david.sherwood@arm.com>
7196 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
7197 * combine.c (find_single_use_1): Likewise.
7198 (expand_field_assignment): Likewise.
7199 (move_deaths): Likewise.
7200 * lra-constraints.c (simplify_operand_subreg): Likewise.
7201 (curr_insn_transform): Likewise.
7202 * lra.c (collect_non_operand_hard_regs): Likewise.
7203 (add_regs_to_insn_regno_info): Likewise.
7204 * rtlanal.c (reg_referenced_p): Likewise.
7205 (covers_regno_no_parallel_p): Likewise.
7207 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7209 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
7210 Don't print any bits outside the precision of the value.
7211 * wide-int.cc (test_printing): Add some new tests.
7213 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7215 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
7216 supports -xbrace_comment option.
7217 * configure: Regenerate.
7218 * config.in: Regenerate.
7219 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
7220 (ASM_CPU_SPEC): Use it.
7222 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7224 * target.def (static_rtx_alignment): New hook.
7225 * targhooks.h (default_static_rtx_alignment): Declare.
7226 * targhooks.c (default_static_rtx_alignment): New function.
7227 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
7228 * doc/tm.texi: Regenerate.
7229 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
7230 instead of targetm.constant_alignment. Remove call to
7232 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7233 (cris_preferred_mininum_alignment): New function, split out from...
7234 (cris_constant_alignment): ...here.
7235 (cris_static_rtx_alignment): New function.
7236 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
7238 (ix86_constant_alignment): ...here.
7239 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7240 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7241 (mmix_static_rtx_alignment): New function.
7242 * config/spu/spu.c (spu_static_rtx_alignment): New function.
7243 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7245 2017-10-26 Tamar Christina <tamar.christina@arm.com>
7248 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
7249 Add flag_trapping_math and flag_fp_int_builtin_inexact.
7251 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
7253 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
7254 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
7255 mark as a sign-extending load.
7256 (local_pic_load_u): Define.
7258 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
7261 * fold-const.c (operand_equal_for_comparison_p): Also return true
7262 if ARG0 is a simple variant of ARG1 with narrower precision.
7263 (fold_ternary_loc): Always pass unstripped operands to the predicate.
7265 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
7267 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
7269 * i386.h (processor_costs): Add gather_static, gather_per_elt,
7270 scatter_static, scatter_per_elt.
7271 * x86-tune-costs.h: Add new cost entries.
7273 2017-10-25 Richard Biener <rguenther@suse.de>
7275 * tree-ssa-sccvn.h (vn_eliminate): Declare.
7276 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
7277 class pass_fre): Move to ...
7278 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
7279 class pass_fre): ... here and adjust for statistics.
7281 2017-10-25 Jakub Jelinek <jakub@redhat.com>
7284 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
7285 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
7286 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
7287 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
7290 2017-10-25 Richard Biener <rguenther@suse.de>
7292 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
7293 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
7294 eliminate_push_avail, eliminate_insert): Move inside...
7295 (class eliminate_dom_walker): ... this class in preparation
7297 (fini_eliminate): Remove by merging with ...
7298 (eliminate): ... this function. Adjust for class changes.
7299 (pass_pre::execute): Remove fini_eliminate call.
7300 (pass_fre::execute): Likewise.
7302 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7305 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
7306 (VPERMI2, VPERMI2I): New mode iterators.
7307 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
7308 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
7310 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
7311 mode iterator. Remove 3 old define_insn patterns.
7312 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
7313 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
7314 VPERMI2 mode iterator, remove the other two expanders.
7315 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
7316 to use VPERMI2 mode iterator, add another alternative for vpermi2*
7317 instructions, remove the other two patterns.
7318 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
7319 mode iterator, remove the other two patterns.
7320 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
7321 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
7322 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
7323 and adjust argument order accordingly.
7324 (ix86_expand_vec_perm): Adjust caller.
7325 (expand_vec_perm_1): Likewise.
7326 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
7327 (expand_vec_perm_vpermt2_vpshub2): ... this.
7328 (ix86_expand_vec_perm_const_1): Adjust caller.
7329 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
7332 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
7333 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
7334 (vec_shl_<mode>): Remove unused expander.
7335 (avx512bw_<shift_insn><mode>3): New define_insn.
7336 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
7337 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
7339 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
7342 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
7345 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7347 PR rtl-optimization/82396
7348 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
7349 (autopref_multipass_init): Simplify initialization.
7350 (autopref_rank_data): Simplify sort order.
7351 * gcc/sched-int.h (autopref_multipass_data_): Remove
7352 multi_mem_insn_p, min_offset and max_offset.
7354 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7357 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7358 Check special value of flag_omit_frame_pointer.
7359 (aarch64_can_eliminate): Likewise.
7360 (aarch64_override_options_after_change_1): Simplify handling of
7361 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
7363 2017-10-24 Richard Biener <rguenther@suse.de>
7365 PR tree-optimization/82697
7366 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
7367 zero for conditional load and unconditional store.
7369 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7371 * doc/install.texi: Document bootstrap-cet.
7373 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7376 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
7377 ENDBR instruction at function entrance if function is only
7380 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7383 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
7384 patterns to better describe from which operation the CF is computed.
7385 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
7386 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
7387 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
7388 is 0, use _0 suffixed expanders instead of emitting a comparison
7391 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
7393 * config/i386/i386.md(*movsf_internal, *movdf_internal):
7394 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
7396 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
7399 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
7400 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
7401 * loop-iv.c (iv_get_reaching_def): Likewise.
7402 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
7403 variable is promoted and the partition contains undefined values.
7405 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7407 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
7409 (nios2_address_cost): Define.
7410 (nios2_legitimize_address): Recognize (exp + constant) directly.
7411 (TARGET_ADDRESS_COST): Define.
7413 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7415 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
7416 (nios2_symbolic_memory_operand_p): Declare.
7417 (nios2_split_large_constant): Declare.
7418 (nios2_split_symbolic_memory_operand): Declare.
7419 * config/nios2/nios2.c: Adjust includes.
7420 (nios2_symbolic_constant_allowed): New.
7421 (nios2_symbolic_constant_p): New.
7422 (nios2_plus_symbolic_constant_p): New.
7423 (nios2_valid_addr_expr_p): Recognize addresses involving
7425 (nios2_legitimate_address_p): Likewise, also LO_SUM.
7426 (nios2_symbolic_memory_operand_p): New.
7427 (nios2_large_constant_p): New.
7428 (nios2_split_large_constant): New.
7429 (nios2_split_plus_large_constant): New.
7430 (nios2_split_symbolic_memory_operand): New.
7431 (nios2_legitimize_address): Code refactoring. Handle addresses
7432 involving symbolic constants.
7433 (nios2_emit_move_sequence): Likewise.
7434 (nios2_print_operand): Improve error output.
7435 (nios2_print_operand_address): Handle LO_SUM.
7436 (nios2_cdx_narrow_form_p): Likewise.
7437 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
7438 operands involving symbolic constants.
7439 (movhi_internal, movsi_internal): Likewise.
7440 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
7441 (extendhisi2, extendqi<mode>2): Likewise.
7443 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7445 * tree-pass.h (PROP_rtl_split_insns): Define.
7446 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
7448 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7450 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
7452 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7455 * target.def (const_not_ok_for_debug_p): Default to
7456 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
7457 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
7458 * targhooks.c (default_const_not_ok_for_debug_p): New function.
7459 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
7460 which targetm.const_not_ok_for_debug_p returned true.
7461 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
7463 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
7465 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
7466 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
7467 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
7469 (ix86_const_not_ok_for_debug_p): New function.
7470 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
7471 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
7473 2017-10-23 David Malcolm <dmalcolm@redhat.com>
7476 * system.h: Conditionally include "unique-ptr.h" if
7477 INCLUDE_UNIQUE_PTR is defined.
7478 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
7479 of defining INCLUDE_UNIQUE_PTR before including "system.h".
7481 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
7483 * config/rl78/rl78.md: New define_expand "subdi3".
7485 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
7488 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
7489 DF_REF_INSN if DF_REF_INSN_INFO is false.
7491 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
7493 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
7494 xmm_move instead of sse_move.
7495 (sse_store_index): New function.
7496 (ix86_register_move_cost): Be more sensible about mismatch stall;
7497 model AVX moves correctly; make difference between sse->integer and
7499 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
7500 moves; make difference between SSE and AVX.
7501 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
7502 and zmm_move. Increase size of sse load and store tables;
7503 add unaligned load and store tables; add ssemmx_to_integer.
7504 * x86-tune-costs.h: Update all entries according to real
7505 move latencies from Agner Fog's manual and chip documentation.
7507 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7510 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
7511 * config/i386/constraints.md (Wf): New constraint.
7512 * config/i386/i386.md (UNSPEC_SBB): New unspec.
7513 (cmp<dwi>_doubleword): Removed.
7514 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
7515 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
7516 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
7517 expand with cmp<dwi>_doubleword. For LTU and GEU use
7518 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
7520 * common.opt (gcolumn-info): Enable by default.
7521 * doc/invoke.texi (gcolumn-info): Document new default.
7523 2017-10-23 Richard Biener <rguenther@suse.de>
7525 PR tree-optimization/82672
7526 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
7527 Fold the stmt if we propagated into it.
7529 2017-10-23 Richard Biener <rguenther@suse.de>
7531 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
7532 (bitmap_remove_expr_from_set): ... this. All callers call this
7533 for non-constant values.
7534 (bitmap_set_subtract): Rename to...
7535 (bitmap_set_subtract_expressions): ... this. Adjust and
7537 (bitmap_set_contains_value): Remove superfluous check.
7538 (bitmap_set_replace_value): Inline into single caller ...
7539 (bitmap_value_replace_in_set): ... here and simplify.
7540 (dependent_clean): Merge into ...
7541 (clean): ... this using an overload. Adjust.
7542 (prune_clobbered_mems): Adjust.
7543 (compute_antic_aux): Likewise.
7544 (compute_partial_antic_aux): Likewise.
7546 2017-10-23 Richard Biener <rguenther@suse.de>
7548 PR tree-optimization/82129
7550 2017-08-01 Richard Biener <rguenther@suse.de>
7552 PR tree-optimization/81181
7553 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7554 (compute_antic): ... end of iteration here.
7556 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7558 * target.def (starting_frame_offset): New hook.
7559 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
7560 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
7561 * doc/tm.texi.in: Regenerate.
7562 * hooks.h (hook_hwi_void_0): Declare.
7563 * hooks.c (hook_hwi_void_0): New function.
7564 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
7565 STARTING_FRAME_OFFSET.
7566 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
7567 * reload1.c (reload): Likewise.
7568 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
7569 instead of STARTING_FRAME_OFFSET.
7570 * function.c (try_fit_stack_local): Likewise.
7571 (assign_stack_local_1): Likewise
7572 (instantiate_virtual_regs): Likewise.
7573 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
7574 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
7575 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
7576 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
7577 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
7578 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
7579 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
7580 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
7581 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
7582 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
7583 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
7584 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
7585 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
7586 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
7587 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
7588 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
7589 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
7590 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
7591 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
7592 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
7593 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
7594 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
7595 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
7596 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
7597 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
7598 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
7599 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
7600 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
7601 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
7602 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
7603 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
7604 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
7605 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
7606 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
7607 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
7608 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
7609 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
7610 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
7611 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
7612 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
7613 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
7614 * config/avr/avr.c (avr_starting_frame_offset): Make static and
7615 return a HOST_WIDE_INT.
7616 (avr_builtin_setjmp_frame_value): Use it instead of
7617 STARTING_FRAME_OFFSET.
7618 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7619 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
7620 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
7622 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7623 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
7624 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
7625 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7626 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
7627 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
7628 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7629 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
7630 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
7631 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7632 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
7633 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
7635 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7636 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
7637 * config/mips/mips.c (mips_compute_frame_info): Refer to
7638 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
7639 (mips_starting_frame_offset): New function.
7640 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7641 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
7642 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
7643 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
7644 and return a HOST_WIDE_INT.
7645 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7646 (mmix_initial_elimination_offset): Refer to
7647 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
7648 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
7649 * config/pa/pa.c (pa_starting_frame_offset): New function.
7650 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
7651 (pa_expand_prologue): Likewise.
7652 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7653 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
7654 !FRAME_GROWS_DOWNWARD handling to...
7655 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7656 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
7657 !FRAME_GROWS_DOWNWARD handling to...
7658 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7659 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
7660 !FRAME_GROWS_DOWNWARD handling to...
7661 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7662 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
7664 (rs6000_starting_frame_offset): New function.
7665 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
7666 !FRAME_GROWS_DOWNWARD handling to...
7667 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7668 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
7669 !FRAME_GROWS_DOWNWARD handling to...
7670 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7671 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
7672 !FRAME_GROWS_DOWNWARD handling to...
7673 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7674 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
7675 (rs6000_starting_frame_offset): New function.
7676 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
7677 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
7678 * config/vax/vax.c (vax_starting_frame_offset): New function.
7679 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
7680 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7681 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
7682 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
7683 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7684 * system.h (STARTING_FRAME_OFFSET): Poison.
7686 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7688 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
7689 SCALAR_TYPE_MODE instead of TYPE_MODE.
7691 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7692 Alan Hayward <alan.hayward@arm.com>
7693 David Sherwood <david.sherwood@arm.com>
7695 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
7697 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7698 Alan Hayward <alan.hayward@arm.com>
7699 David Sherwood <david.sherwood@arm.com>
7701 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
7703 2017-10-23 Richard Biener <rguenther@suse.de>
7705 PR tree-optimization/82129
7706 * tree-ssa-pre.c (bitmap_set_and): Remove.
7707 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
7708 canonicalizing expressions in the set to those with lowest
7709 ID rather than taking that from the first edge.
7711 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7713 * combine.c (rtx_equal_for_field_assignment_p): Use
7714 byte_lowpart_offset.
7716 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7717 Alan Hayward <alan.hayward@arm.com>
7718 David Sherwood <david.sherwood@arm.com>
7720 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
7721 to a SUBREG_PROMOTED_VAR.
7723 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7724 Alan Hayward <alan.hayward@arm.com>
7725 David Sherwood <david.sherwood@arm.com>
7727 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
7728 (expand_debug_source_expr): Likewise.
7729 * combine.c (combine_simplify_rtx): Likewise.
7730 * cse.c (fold_rtx): Likewise.
7731 * optabs.c (expand_float): Likewise.
7732 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7733 (simplify_binary_operation_1): Likewise.
7735 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7736 Alan Hayward <alan.hayward@arm.com>
7737 David Sherwood <david.sherwood@arm.com>
7739 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
7740 (record_promoted_value): Likewise.
7741 * expr.c (expand_expr_real_2): Likewise.
7742 * ree.c (update_reg_equal_equiv_notes): Likewise.
7743 (combine_set_extension): Likewise.
7744 * rtlanal.c (low_bitmask_len): Likewise.
7745 * simplify-rtx.c (neg_const_int): Likewise.
7746 (simplify_binary_operation_1): Likewise.
7748 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7749 Alan Hayward <alan.hayward@arm.com>
7750 David Sherwood <david.sherwood@arm.com>
7752 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
7753 * regcprop.c (maybe_mode_change): Likewise.
7754 * reload1.c (alter_reg): Likewise.
7756 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7758 * inchash.h (inchash::hash::add_wide_int): New function.
7759 * lto-streamer-out.c (hash_tree): Use it.
7761 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7763 * inchash.h (inchash::hash::add_wide_int): Rename to...
7764 (inchash::hash::add_hwi): ...this.
7765 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
7766 (polymorphic_call_target_hasher::hash): Likewise.
7767 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
7768 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
7769 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
7770 * lto-streamer-out.c (hash_tree): Likewise.
7771 * optc-save-gen.awk: Likewise.
7772 * tree.c (add_expr): Likewise.
7774 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
7777 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
7778 for ordered inequality comparisons even with TARGET_IEEE_FP.
7780 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
7783 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
7784 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
7785 Expand with cmp<dwi>_doubleword.
7787 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7789 * extend.texi: Add x86 specific to 'nocf_check' attribute.
7790 List CET intrinsics.
7791 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
7792 specific to -fcf-protection option.
7794 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7796 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
7797 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
7798 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
7799 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
7800 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
7801 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
7802 (extra_objs): Add cet.o for Linux/x86 targets.
7803 (tmake_file): Add i386/t-cet for Linux/x86 targets.
7804 * config/i386/cet.c: New file.
7805 * config/i386/cetintrin.h: Likewise.
7806 * config/i386/t-cet: Likewise.
7807 * config/i386/cpuid.h (bit_SHSTK): New.
7808 (bit_IBT): Likewise.
7809 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
7810 pass IBT and SHSTK bits.
7811 * config/i386/i386-builtin-types.def
7812 (VOID_FTYPE_UNSIGNED_PVOID): New.
7813 (VOID_FTYPE_UINT64_PVOID): Likewise.
7814 * config/i386/i386-builtin.def: Add CET intrinsics.
7815 * config/i386/i386-c.c (ix86_target_macros_internal): Add
7816 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
7817 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
7818 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
7820 * config/i386/i386.c (rest_of_insert_endbranch): New.
7821 (pass_data_insert_endbranch): Likewise.
7822 (pass_insert_endbranch): Likewise.
7823 (make_pass_insert_endbranch): Likewise.
7824 (ix86_notrack_prefixed_insn_p): Likewise.
7825 (ix86_target_string): Add -mibt, -mshstk flags.
7826 (ix86_option_override_internal): Add flag_cf_protection
7828 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
7829 (ix86_print_operand): Add 'notrack' prefix output.
7830 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
7831 (ix86_expand_builtin): Expand CET intrinsics.
7832 (x86_output_mi_thunk): Add 'endbranch' instruction.
7833 * config/i386/i386.h (TARGET_IBT): New.
7834 (TARGET_IBT_P): Likewise.
7835 (TARGET_SHSTK): Likewise.
7836 (TARGET_SHSTK_P): Likewise.
7837 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
7838 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
7839 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
7840 (builtin_setjmp_setup): New pattern.
7841 (builtin_longjmp): Likewise.
7842 (rdssp<mode>): Likewise.
7843 (incssp<mode>): Likewise.
7844 (saveprevssp): Likewise.
7845 (rstorssp): Likewise.
7846 (wrss<mode>): Likewise.
7847 (wruss<mode>): Likewise.
7848 (setssbsy): Likewise.
7849 (clrssbsy): Likewise.
7850 (nop_endbr): Likewise.
7851 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
7853 * config/i386/immintrin.h: Include <cetintrin.h>.
7854 * config/i386/linux-common.h
7855 (file_end_indicate_exec_stack_and_cet): New prototype.
7856 (TARGET_ASM_FILE_END): New.
7858 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7860 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
7861 latencies instead of having separate table; make difference between
7862 integer and float costs.
7863 * i386.h (processor_costs): Remove scalar_stmt_cost,
7864 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
7865 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
7867 * x86-tune-costs.h: Remove entries which has been removed in
7868 procesor_costs from all tables; make cond_taken_branch_cost
7869 and cond_not_taken_branch_cost COST_N_INSNS based.
7871 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7873 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
7875 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7877 * config/i386/i386.md (isa): Remove fma_avx512f.
7878 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
7879 <avx512>_fmadd_<mode>_mask3<round_name>,
7880 <avx512>_fmsub_<mode>_mask<round_name>,
7881 <avx512>_fmsub_<mode>_mask3<round_name>,
7882 <avx512>_fnmadd_<mode>_mask<round_name>,
7883 <avx512>_fnmadd_<mode>_mask3<round_name>,
7884 <avx512>_fnmsub_<mode>_mask<round_name>,
7885 <avx512>_fnmsub_<mode>_mask3<round_name>,
7886 <avx512>_fmaddsub_<mode>_mask<round_name>,
7887 <avx512>_fmaddsub_<mode>_mask3<round_name>,
7888 <avx512>_fmsubadd_<mode>_mask<round_name>,
7889 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
7890 (*vec_widen_umult_even_v16si<mask_name>,
7891 *vec_widen_smult_even_v16si<mask_name>): Likewise.
7892 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
7894 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7896 * extend.texi: Add 'nocf_check' documentation.
7897 * gimple.texi: Add second parameter to
7898 gimple_build_call_from_tree.
7899 * invoke.texi: Add -fcf-protection documentation.
7900 * rtl.texi: Add REG_CALL_NOTRACK documenation.
7902 2017-10-20 Richard Biener <rguenther@suse.de>
7904 PR tree-optimization/82473
7905 * tree-vect-loop.c (vectorizable_reduction): Properly get at
7906 the largest input type.
7908 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7910 * c-attribs.c (handle_nocf_check_attribute): New function.
7911 (c_common_attribute_table): Add 'nocf_check' handling.
7912 * gimple-parser.c: Add second argument NULL to
7913 gimple_build_call_from_tree.
7914 * attrib.c (comp_type_attributes): Check nocf_check attribute.
7915 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
7917 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
7918 * common.opt: Add fcf-protection flag.
7919 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
7920 * flag-types.h: Add enum cf_protection_level.
7921 * gimple.c (gimple_build_call_from_tree): Add second parameter.
7922 Add 'nocf_check' attribute propagation to gimple call.
7923 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
7924 (gimple_build_call_from_tree): Update prototype.
7925 (gimple_call_nocf_check_p): New function.
7926 (gimple_call_set_nocf_check): Likewise.
7927 * gimplify.c: Add second argument to gimple_build_call_from_tree.
7928 * ipa-icf.c: Add nocf_check attribute in statement hash.
7929 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
7930 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
7931 * toplev.c (process_options): Add flag_cf_protection handling.
7933 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7935 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
7937 2017-10-20 Richard Biener <rguenther@suse.de>
7939 PR tree-optimization/82603
7940 * tree-if-conv.c (predicate_mem_writes): Make sure to only
7941 remove false predicated stores.
7943 2017-10-20 Richard Biener <rguenther@suse.de>
7945 * graphite-isl-ast-to-gimple.c
7946 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
7947 Remove return value and simplify, dump copied stmt after lhs
7949 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
7950 Reduce dump verbosity.
7951 (gsi_insert_earliest): Likewise.
7952 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
7953 * graphite.c (print_global_statistics): Adjust dumping.
7954 (print_graphite_scop_statistics): Likewise.
7955 (print_graphite_statistics): Do not dump loops here.
7956 (graphite_transform_loops): But here.
7958 2017-10-20 Nicolas Roche <roche@adacore.com>
7960 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
7961 * configure: Regenerate.
7963 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7966 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
7967 functions when optimizing replace GIMPLE_RETURN stmts with
7968 calls to __builtin_unreachable ().
7971 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
7972 for -fsanitize=thread link of executables.
7973 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
7974 link of executables.
7977 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
7979 (<shift_insn><mode>3<mask_name>): Change the last of the 3
7980 define_insns for logical vector shifts to use VI248_AVX512BW
7981 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
7982 condition, useless isa and prefix attributes. Change the first
7983 2 of these define_insns to ...
7984 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
7985 define_insn for avx512vl.
7986 (<shift_insn><mode>3): ... and this, new define_insn without
7987 masking for non-avx512vl.
7990 * config/i386/sse.md (*andnot<mode>3,
7991 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
7992 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
7993 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
7994 not applied use empty suffix even for TARGET_AVX512VL.
7995 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
7996 is applied, supply evex,evex or evex,evex,evex instead of just
7999 2017-10-20 Julia Koval <julia.koval@intel.com>
8001 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
8002 (OPTION_MASK_ISA_GFNI_UNSET): New.
8003 (ix86_handle_option): Handle OPT_mgfni.
8004 * config/i386/cpuid.h (bit_GFNI): New.
8005 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
8006 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
8007 * config/i386/i386.c (ix86_target_string): Add -mgfni.
8008 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
8009 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
8010 * config/i386/i386.opt: Add mgfni.
8012 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
8014 * config/msp430/msp430.c (msp430_option_override): Disable
8015 -fdelete-null-pointer-checks.
8016 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
8018 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8020 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
8021 of x87 and SSE instructions.
8023 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8025 * asan.c (create_cond_insert_point): Do not update edge count.
8026 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
8027 (afdo_propagate_circuit): Likewise.
8028 (afdo_calculate_branch_prob): Likewise.
8029 (afdo_annotate_cfg): Likewise.
8030 * basic-block.h (struct edge_def): Remove count.
8031 (edge_def::count): New accessor.
8032 * bb-reorder.c (rotate_loop): Update.
8033 (find_traces_1_round): Update.
8034 (connect_traces): Update.
8035 (sanitize_hot_paths): Update.
8036 * cfg.c (unchecked_make_edge): Update.
8037 (make_single_succ_edge): Update.
8038 (check_bb_profile): Update.
8039 (dump_edge_info): Update.
8040 (update_bb_profile_for_threading): Update.
8041 (scale_bbs_frequencies_int): Update.
8042 (scale_bbs_frequencies_gcov_type): Update.
8043 (scale_bbs_frequencies_profile_count): Update.
8044 (scale_bbs_frequencies): Update.
8045 * cfganal.c (connect_infinite_loops_to_exit): Update.
8046 * cfgbuild.c (compute_outgoing_frequencies): Update.
8047 (find_many_sub_basic_blocks): Update.
8048 * cfgcleanup.c (try_forward_edges): Update.
8049 (try_crossjump_to_edge): Update
8050 * cfgexpand.c (expand_gimple_cond): Update
8051 (expand_gimple_tailcall): Update
8052 (construct_exit_block): Update
8053 * cfghooks.c (verify_flow_info): Update
8054 (redirect_edge_succ_nodup): Update
8055 (split_edge): Update
8056 (make_forwarder_block): Update
8057 (duplicate_block): Update
8058 (account_profile_record): Update
8059 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
8060 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
8061 * cfgloopmanip.c (scale_loop_profile): Update.
8063 (lv_adjust_loop_entry_edge): Update.
8064 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
8065 (force_nonfallthru_and_redirect): Update.
8066 (purge_dead_edges): Update.
8067 (rtl_flow_call_edges_add): Update.
8068 * cgraphunit.c (init_lowered_empty_function): Update.
8069 (cgraph_node::expand_thunk): Update.
8070 * gimple-pretty-print.c (dump_probability): Update.
8071 (dump_edge_probability): Update.
8072 * gimple-ssa-isolate-paths.c (isolate_path): Update.
8073 * haifa-sched.c (sched_create_recovery_edges): Update.
8074 * hsa-gen.c (convert_switch_statements): Update.
8075 * ifcvt.c (dead_or_predicable): Update.
8076 * ipa-inline-transform.c (inline_transform): Update.
8077 * ipa-split.c (split_function): Update.
8078 * ipa-utils.c (ipa_merge_profiles): Update.
8079 * loop-doloop.c (add_test): Update.
8080 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
8081 * lto-streamer-in.c (input_cfg): Update.
8082 (input_function): Update.
8083 * lto-streamer-out.c (output_cfg): Update.
8084 * modulo-sched.c (sms_schedule): Update.
8085 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
8086 * predict.c (maybe_hot_edge_p): Update.
8087 (unlikely_executed_edge_p): Update.
8088 (probably_never_executed_edge_p): Update.
8089 (dump_prediction): Update.
8090 (drop_profile): Update.
8091 (propagate_unlikely_bbs_forward): Update.
8092 (determine_unlikely_bbs): Update.
8093 (force_edge_cold): Update.
8094 * profile.c (compute_branch_probabilities): Update.
8095 * reg-stack.c (better_edge): Update.
8096 * shrink-wrap.c (handle_simple_exit): Update.
8097 * tracer.c (better_p): Update.
8098 * trans-mem.c (expand_transaction): Update.
8099 (split_bb_make_tm_edge): Update.
8100 * tree-call-cdce.c: Update.
8101 * tree-cfg.c (gimple_find_sub_bbs): Update.
8102 (gimple_split_edge): Update.
8103 (gimple_duplicate_sese_region): Update.
8104 (gimple_duplicate_sese_tail): Update.
8105 (gimple_flow_call_edges_add): Update.
8106 (insert_cond_bb): Update.
8107 (execute_fixup_cfg): Update.
8108 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
8109 * tree-complex.c (expand_complex_div_wide): Update.
8110 * tree-eh.c (lower_resx): Update.
8111 (unsplit_eh): Update.
8112 (cleanup_empty_eh_move_lp): Update.
8113 * tree-inline.c (copy_edges_for_bb): Update.
8114 (freqs_to_counts): Update.
8115 (copy_cfg_body): Update.
8116 * tree-ssa-dce.c (remove_dead_stmt): Update.
8117 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
8118 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
8119 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
8120 (unloop_loops): Update.
8121 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8122 * tree-ssa-loop-split.c (connect_loops): Update.
8123 (split_loop): Update.
8124 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
8125 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
8126 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
8127 * tree-ssa-reassoc.c (branch_fixup): Update.
8128 * tree-ssa-tail-merge.c (replace_block_by): Update.
8129 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
8130 (compute_path_counts): Update.
8131 (update_profile): Update.
8132 (recompute_probabilities): Update.
8133 (update_joiner_offpath_counts): Update.
8134 (estimated_freqs_path): Update.
8135 (freqs_to_counts_path): Update.
8136 (clear_counts_path): Update.
8137 (ssa_fix_duplicate_block_edges): Update.
8138 (duplicate_thread_path): Update.
8139 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
8140 (case_bit_test_cmp): Update.
8141 (collect_switch_conv_info): Update.
8142 (gen_inbound_check): Update.
8143 (do_jump_if_equal): Update.
8144 (emit_cmp_and_jump_insns): Update.
8145 * tree-tailcall.c (decrease_profile): Update.
8146 (eliminate_tail_call): Update.
8147 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
8148 (vect_do_peeling): Update.
8149 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8150 * ubsan.c (ubsan_expand_null_ifn): Update.
8151 (ubsan_expand_ptr_ifn): Update.
8152 * value-prof.c (gimple_divmod_fixed_value): Update.
8153 (gimple_mod_pow2): Update.
8154 (gimple_mod_subtract): Update.
8155 (gimple_ic): Update.
8156 (gimple_stringop_fixed_value): Update.
8158 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8161 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
8163 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
8165 PR rtl-optimization/82395
8166 * ira-color.c (allocno_priority_compare_func): Fix comparison step
8167 based on non_spilled_static_chain_regno_p.
8169 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8171 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
8172 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
8173 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
8175 2017-10-19 Martin Sebor <msebor@redhat.com>
8177 PR tree-optimization/82596
8178 * tree.c (array_at_struct_end_p): Handle STRING_CST.
8180 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8182 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
8183 (get_mem_refs_of_builtin_call): Likewise.
8184 * builtins.c (expand_builtin_apply): Adjust call to
8185 allocate_dynamic_stack_space.
8186 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
8187 the third argument to allocate_dynamic_stack_space, otherwise -1.
8188 (expand_builtin): Deal with all alloca variants.
8189 (is_inexpensive_builtin): Likewise.
8190 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
8191 * calls.c (special_function_p): Deal with all alloca variants.
8192 (initialize_argument_information): Adjust call to
8193 allocate_dynamic_stack_space.
8194 (expand_call): Likewise.
8195 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
8196 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
8197 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
8198 use it for the stack usage computation.
8199 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
8200 * function.c (gimplify_parameters): Call build_alloca_call_expr.
8201 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
8202 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
8203 (in_loop_p): Remove first argument and useless check.
8204 (pass_walloca::execute): Remove useless test and adjust call to above.
8205 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
8206 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
8207 (gimplify_call_expr): Deal with all alloca variants.
8208 * hsa-gen.c (gen_hsa_alloca): Likewise.
8209 (gen_hsa_insns_for_call): Likewise.
8210 * ipa-pure-const.c (special_builtin_state): Likewise.
8211 * tree-chkp.c (chkp_build_returned_bound): Likewise.
8212 * tree-object-size.c (alloc_object_size): Likewise.
8213 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
8214 (call_may_clobber_ref_p_1): Likewise.
8215 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
8216 (ccp_fold_stmt): Likewise.
8217 (optimize_stack_restore): Likewise.
8218 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
8219 (mark_all_reaching_defs_necessary_1): Likewise.
8220 (propagate_necessity): Likewise.
8221 (eliminate_unnecessary_stmts): Likewise.
8222 * tree.c (build_common_builtin_nodes): Build
8223 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
8224 (build_alloca_call_expr): New function.
8225 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
8226 (CASE_BUILT_IN_ALLOCA): Likewise.
8227 (build_alloca_call_expr): Declare.
8228 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
8230 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8233 * dwarf2out.c (new_die_raw): New static inline function.
8234 (new_die): Use it to create the DIE.
8235 (add_AT_external_die_ref): Likewise.
8236 (clone_die): Likewise.
8237 (clone_as_declaration): Likewise.
8238 (dwarf2out_vms_debug_main_pointer): Likewise.
8239 (base_type_die): Likewise. Remove early return for corner cases.
8240 Do not call add_pubtype on the DIE here.
8241 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
8242 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
8243 typedefs for base types with DW_AT_endianity. Make sure a DIE with
8244 native order exists for base types, attach the DIE manually and call
8245 add_pubtype on it. Do not equate a reverse order DIE to the type.
8247 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
8249 * config/arm/arm.c (align_ok_ldrd_strd): New function.
8250 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
8252 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
8254 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8256 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
8257 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
8258 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
8259 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
8260 * opts.c (sanitizer_opts): Add builtin.
8261 * ubsan.c (instrument_builtin): New function.
8262 (pass_ubsan::execute): Call it.
8263 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
8264 * doc/invoke.texi: Document -fsanitize=builtin.
8266 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
8267 builtins, store max (log2 (align), 0) into uchar field instead of
8268 align into uptr field.
8269 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
8270 store uchar 0 field instead of uptr 0 field.
8271 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
8272 instead of passing one address of struct with 2 locations pass
8273 two addresses of structs with 1 location each.
8274 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
8275 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
8276 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
8277 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
8278 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
8279 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
8280 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
8281 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
8283 2017-10-19 Martin Liska <mliska@suse.cz>
8286 * file-find.c (remove_prefix): Remove.
8287 * file-find.h (remove_prefix): Likewise.
8288 * gcc-ar.c: Remove smartness of lookup.
8290 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
8292 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
8293 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
8294 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
8296 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8299 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
8300 (setcc + and to xor + setcc): New peephole2.
8302 2017-10-19 Tom de Vries <tom@codesourcery.com>
8304 * doc/sourcebuild.texi (Test Directives, Variants of
8305 dg-require-support): Add dg-require-stack-size.
8307 2017-10-19 Martin Liska <mliska@suse.cz>
8310 * gimplify.c (gimplify_decl_expr): Do not instrument variables
8311 that have a large alignment.
8312 (gimplify_target_expr): Likewise.
8314 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
8316 PR rtl-optimization/82602
8317 * ira.c (rtx_moveable_p): Return false for volatile asm.
8319 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
8322 * config/i386/i386-modes.def (CCGZ): New CC mode.
8323 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
8324 * config/i386/predicates.md (ix86_comparison_operator):
8326 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
8327 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
8328 with double-word subtraction.
8329 (put_condition_code): Handle CCGZmode.
8331 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
8333 * wide-int.cc (debug (const wide_int &)): New.
8334 (debug (const wide_int *)): New.
8335 (debug (const widest_int &)): New.
8336 (debug (const widest_int *)): New.
8338 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
8341 * lra-constraints.c (curr_insn_transform): Use non-input operand
8342 instead of output one for matched reload.
8344 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8346 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
8347 (tree-ssa-loop-ivopts.h): New header file.
8348 (struct builtin_info): New fields.
8349 (classify_builtin_1): Compute and record base and offset parts for
8350 memset builtin partition by calling strip_offset.
8351 (offset_cmp, fuse_memset_builtins): New functions.
8352 (finalize_partitions): Fuse adjacent memset partitions by calling
8354 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
8355 Expose the interface.
8356 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
8358 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8360 PR tree-optimization/82574
8361 * tree-loop-distribution.c (find_single_drs): New parameter. Check
8362 that data reference must be executed exactly once per iteration
8363 against the outermost loop in nest.
8364 (classify_partition): Update call to above function.
8366 2017-10-18 Richard Biener <rguenther@suse.de>
8368 PR tree-optimization/82591
8369 * graphite.c (graphite_transform_loops): Move code gen message
8371 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8372 Here. Handle scop_to_isl_ast failing.
8373 (scop_to_isl_ast): Limit the number of ISL operations.
8375 2017-10-18 Richard Biener <rguenther@suse.de>
8377 * graphite-isl-ast-to-gimple.c
8378 (translate_isl_ast_to_gimple::set_rename): Simplify.
8379 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
8380 (graphite_copy_stmts_from_block): ... here.
8381 (copy_bb_and_scalar_dependences): Simplify.
8382 (add_parameters_to_ivs_params): Canonicalize.
8383 (generate_entry_out_of_ssa_copies): Simplify.
8384 * graphite-sese-to-poly.c (extract_affine_name): Simplify
8385 by passing in ISL dimension.
8386 (parameter_index_in_region_1): Rename to ...
8387 (parameter_index_in_region): ... this.
8388 (extract_affine): Adjust assert, pass down parameter index.
8389 (add_param_constraints): Use range-info when available.
8390 (build_scop_context): Adjust.
8391 * sese.c (new_sese_info): Adjust.
8392 (free_sese_info): Likewise.
8393 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
8394 Remove unused typedefs.
8395 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
8397 2017-10-18 Martin Liska <mliska@suse.cz>
8399 * combine.c (simplify_compare_const): Add gcc_fallthrough.
8401 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8403 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
8404 (s390_sched_init): Do not reset s390_sched_state if we entered the
8405 current basic block via a fallthru edge and all others are unlikely.
8407 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8409 * config/s390/s390.c (NUM_SIDES): New variable.
8410 (LONGRUNNING_THRESHOLD): New variable.
8411 (LATENCY_FACTOR): New variable.
8412 (s390_sched_score): Decrease score for long-running instructions on
8414 (s390_sched_variable_issue): Perform bookkeeping for long-running
8417 2017-10-18 Richard Biener <rguenther@suse.de>
8419 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
8420 Simplify with removal of the parameter rename map.
8421 (set_rename): Likewise.
8422 (should_copy_to_new_region): Likewise.
8423 (graphite_copy_stmts_from_block): Likewise.
8424 (copy_bb_and_scalar_dependences): Remove initialization of
8425 unused copied_bb_map.
8427 (copy_internal_parameters): Likewise.
8428 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
8429 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
8430 Use INTEGRAL_TYPE_P.
8431 (parameter_index_in_region_1): Rename to ...
8432 (assign_parameter_index_in_region): ... this. Assert we have
8433 a parameter we handle.
8434 (scan_tree_for_params): Adjust.
8435 * sese.h (parameter_rename_map_t): Remove.
8436 (struct sese_info_t): Remove unused parameter_rename_map and
8437 copied_bb_map members.
8438 * sese.c (new_sese_info): Adjust.
8439 (free_sese_info): Likewise.
8441 2017-10-18 Martin Liska <mliska@suse.cz>
8444 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
8445 on an abnormal edge.
8447 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
8449 * doc/invoke.texi (ffunction-sections and fdata-sections):
8452 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8454 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
8455 the use statement can throw internally.
8457 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8459 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
8460 any RTX present on the RHS of a SET.
8461 * compare-elim.c (try_eliminate_compare): Restore comment.
8463 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8465 * langhooks.h (struct lang_hooks): Document that tree_size langhook
8466 may be also called on tcc_type nodes.
8467 * langhooks.c (lhd_tree_size): Likewise.
8469 2017-10-17 David Malcolm <dmalcolm@redhat.com>
8471 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
8472 format_warning_at_substring.
8473 (maybe_warn): Convert source_range * param to a location_t. Pass
8474 UNKNOWN_LOCATION rather than NULL to fmtwarn.
8475 (format_directive): Remove code to extract source_ranges and
8476 source_range * in favor of just a location_t.
8477 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
8479 * substring-locations.c (format_warning_va): Convert
8480 source_range * param to a location_t.
8481 (format_warning_at_substring): Likewise.
8482 * substring-locations.h (format_warning_va): Likewise.
8483 (format_warning_at_substring): Likewise.
8485 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
8487 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
8489 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
8490 and scatter/gather ops.
8492 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
8493 vec_gather_load and vec_scatter_store.
8494 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
8495 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
8496 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
8497 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
8498 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
8499 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
8501 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
8503 * reg-stack.c (compare_for_stack_reg): Add bool argument.
8504 Detect FTST instruction and handle its register pops. Only pop
8505 second operand if can_pop_second_op is true.
8506 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
8507 set can_pop_second_op to false in the compare_for_stack_reg call.
8509 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
8510 output_fp_compare for stack register operands.
8511 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
8512 instructions here. Do not emit stack register pops here. Assert
8513 that FCOMPP pops next to top stack register. Rewrite function.
8515 2017-10-17 Nathan Sidwell <nathan@acm.org>
8518 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
8519 use DECL_ASSEMBLER_NAME_RAW.
8522 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
8525 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
8526 Wilco Dijkstra <wilco.dijkstra@arm.com>
8528 * builtins.c (expand_builtin_update_setjmp_buf): Add a
8529 converstion to Pmode from the buf_addr.
8531 2017-10-17 Richard Biener <rguenther@suse.de>
8533 * graphite-dependences.c (scop_get_reads_and_writes): Change
8534 output parameters to references.
8536 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
8538 PR 71026/tree-optimization
8539 * fold-const.c (distribute_real_division): Removed.
8540 (fold_binary_loc): Remove calls to distribute_real_divison.
8542 2017-10-17 Richard Biener <rguenther@suse.de>
8544 * graphite-scop-detection.c
8545 (scop_detection::stmt_has_simple_data_refs_p): Always use
8546 the full nest as region.
8547 (try_generate_gimple_bb): Likewise.
8548 * sese.c (scalar_evolution_in_region): Simplify now that
8549 SCEV can handle instantiation in regions.
8550 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
8551 in the non-loop part of a function if requested.
8553 2017-10-17 Richard Biener <rguenther@suse.de>
8555 PR tree-optimization/82563
8556 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
8558 (graphite_regenerate_ast_isl): Call it.
8559 * graphite-scop-detection.c (build_scops): Remove entry edge split.
8561 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8563 PR tree-optimization/82549
8564 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
8565 Formatting fixes. Instead of calling make_bit_field_ref with negative
8568 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
8570 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
8571 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
8572 _mm_maskz_reduce_ss): New.
8573 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
8574 __builtin_ia32_reducess_mask): Ditto..
8575 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
8576 * config/i386/sse.md (reduces<mode>): Renamed to ...
8577 (reduces<mode><mask_scalar_name>): ... this.
8578 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
8580 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
8581 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
8583 2017-10-16 David Malcolm <dmalcolm@redhat.com>
8585 * Makefile.in (OBJS): Add unique-ptr-tests.o.
8586 * selftest-run-tests.c (selftest::run_tests): Call
8587 selftest::unique_ptr_tests_cc_tests.
8588 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
8589 * unique-ptr-tests.cc: New file.
8591 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
8594 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8596 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8597 (make_hard_regno_born, make_hard_regno_dead): Update
8598 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
8600 2017-10-16 Jeff Law <law@redhat.com>
8602 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
8604 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
8606 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
8608 2017-10-16 Olivier Hainque <hainque@adacore.com>
8610 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
8611 with_cpu if we were configured for an e500v2 target cpu name.
8613 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
8615 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
8616 * doc/invoke.texi: Document +nodsp as a valid extension for
8619 2017-10-16 Martin Liska <mliska@suse.cz>
8621 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
8622 (test_set_range): Likewise.
8623 (test_range_functions): Rename to ...
8624 (test_bit_in_range): ... this.
8625 (sbitmap_c_tests): Add new test.
8627 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8629 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
8631 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
8632 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
8634 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8636 * config/aarch64/aarch64-builtins.c
8637 (aarch64_types_quadopu_lane_qualifiers): New.
8638 (TYPES_QUADOPU_LANE): New.
8639 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
8640 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
8641 (aarch64_<sur>dot_laneq<vsi2qi>): New.
8642 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
8643 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
8644 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
8645 (Vdottype, DOTPROD): New.
8646 (sur): Add SDOT and UDOT.
8648 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8650 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
8651 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
8652 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
8654 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
8655 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
8656 Enable TARGET_DOTPROD.
8657 (cortex-a75.cortex-a55): Likewise.
8658 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
8660 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8662 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
8663 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
8665 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
8667 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
8668 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
8669 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
8670 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
8671 * config/arm/types.md (neon_dot, neon_dot_q): New.
8672 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
8674 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8676 * config/arm/arm.h (TARGET_DOTPROD): New.
8677 * config/arm/arm.c (arm_arch_dotprod): New.
8678 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
8679 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
8680 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
8681 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
8682 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
8683 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
8684 * doc/invoke.texi (armv8.2-a): Document dotprod
8686 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
8688 * i386.c (ix86_vec_cost): New function.
8689 (ix86_rtx_costs): Handle vector operations better.
8690 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
8691 * x86-tune-costs.h: Add new costs to all tables.
8693 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
8695 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
8697 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
8698 divsd, sqrtss and sqrtsd
8699 * x86-tune-costs.h: Add new entries to all costs.
8700 (znver1_cost): Fix to match real instruction latencies.
8702 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8703 Michael Collison <michael.collison@arm.com>
8705 * compare-elim.c: Include emit-rtl.h.
8706 (can_merge_compare_into_arith): New function.
8707 (try_validate_parallel): Likewise.
8708 (try_merge_compare): Likewise.
8709 (try_eliminate_compare): Call the above when no previous clobber
8711 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
8714 2017-10-14 Jakub Jelinek <jakub@redhat.com>
8718 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
8719 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
8723 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
8724 to be any operand_equal_p operands. For & (B - 1) require
8725 B to be power of 2. Recognize
8726 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
8728 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8731 * optabs.c (expand_memory_blockage): Fix call of
8732 targetm.have_memory_blockage.
8734 2017-10-14 Jakub Jelinek <jakub@redhat.com>
8737 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
8738 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
8739 objects to extra_objs instead of overwriting it.
8741 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8743 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
8744 Use any_fp_register_operand as operand[3] predicate. Simplify
8745 equality test for operands[2] and operands[4] memory location.
8746 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
8747 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
8748 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
8749 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
8750 any_fp_register_operand as operand[1] predicate. Simplify
8751 equality test for operands[0] and operands[3] memory location.
8752 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
8753 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
8754 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
8756 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8758 * target-insns.def: Add memory_blockage.
8759 * optabs.c (expand_memory_blockage): New function.
8760 (expand_asm_memory_barrier): Rename ...
8761 (expand_asm_memory_blockage): ... to this.
8762 (expand_mem_thread_fence): Call expand_memory_blockage
8763 instead of expand_asm_memory_barrier.
8764 (expand_mem_singnal_fence): Ditto.
8765 (expand_atomic_load): Ditto.
8766 (expand_atomic_store): Ditto.
8767 * doc/md.texi (Standard Pattern Names For Generation):
8768 Document memory_blockage instruction pattern.
8770 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
8772 * config/rl78/rl78.c (rl78_emit_libcall): New function.
8773 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
8774 * config/rl78/rl78.md: New define_expand "adddi3".
8776 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
8778 * cfghooks.c (verify_flow_info): Disable check that all probabilities
8781 2017-10-13 Jeff Law <law@redhat.com>
8783 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
8785 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8788 * internal-fn.c (expand_mul_overflow): If both operands have
8789 the same highpart of -1 or 0 and the topmost bit of lowpart
8790 is different, overflow is if res <= 0 rather than res < 0.
8792 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
8794 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
8795 TARGET_P9_VECTOR code for unaligned_load case.
8797 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
8799 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
8801 2017-10-13 Nathan Sidwell <nathan@acm.org>
8803 * tree-core.h (tree_contains_struct): Make bool.
8804 * tree.c (tree_contains_struct): Likewise.
8805 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
8806 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
8807 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
8808 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
8809 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
8811 2017-10-13 Richard Biener <rguenther@suse.de>
8813 * graphite-isl-ast-to-gimple.c
8814 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
8815 parameters and dominance check.
8816 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
8817 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
8818 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
8819 Do not update SSA form here or do intermediate IL verification.
8820 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
8821 (graphite_initialize): Remove check on the number of loops in
8822 the function and inline into graphite_transform_loops.
8823 (graphite_finalize): Inline into graphite_transform_loops.
8824 (graphite_transform_loops): Perform SSA update and IL verification
8826 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
8828 2017-10-13 Richard Biener <rguenther@suse.de>
8830 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
8831 graphite_expression_type_precision): Avoid global constructor
8833 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
8834 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
8835 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
8836 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
8837 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
8838 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
8840 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
8843 * config/i386/i386.h (ix86_red_zone_size): New.
8844 * config/i386/i386.md (push peephole2s): Replace
8845 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
8847 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8848 Alan Hayward <alan.hayward@arm.com>
8849 David Sherwood <david.sherwood@arm.com>
8851 * combine.c (can_change_dest_mode): Reject changes in
8852 REGMODE_NATURAL_SIZE.
8854 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8855 Alan Hayward <alan.hayward@arm.com>
8856 David Sherwood <david.sherwood@arm.com>
8858 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
8859 (expand_debug_source_expr): Likewise.
8860 * combine.c (combine_simplify_rtx): Likewise.
8861 * cse.c (fold_rtx): Likewise.
8862 * fwprop.c (canonicalize_address): Likewise.
8863 * targhooks.c (default_shift_truncation_mask): Likewise.
8865 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8866 Alan Hayward <alan.hayward@arm.com>
8867 David Sherwood <david.sherwood@arm.com>
8869 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
8870 (widened_mode): Likewise.
8871 (expand_unop): Likewise.
8872 * ree.c (transform_ifelse): Likewise.
8873 (merge_def_and_ext): Likewise.
8874 (combine_reaching_defs): Likewise.
8875 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8877 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8878 Alan Hayward <alan.hayward@arm.com>
8879 David Sherwood <david.sherwood@arm.com>
8881 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
8882 * combine.c (gen_lowpart_for_combine): Likewise.
8883 * dwarf2out.c (rtl_for_decl_location): Likewise.
8884 * final.c (alter_subreg): Likewise.
8885 * rtlhooks.c (gen_lowpart_general): Likewise.
8886 (gen_lowpart_if_possible): Likewise.
8888 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8889 Alan Hayward <alan.hayward@arm.com>
8890 David Sherwood <david.sherwood@arm.com>
8892 * calls.c (expand_call): Use subreg_lowpart_offset.
8893 * cse.c (cse_insn): Likewise.
8894 * regcprop.c (copy_value): Likewise.
8895 (copyprop_hardreg_forward_1): Likewise.
8897 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8900 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
8901 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
8902 =Q constraints to +Q and into insn condition add check
8903 that operands[0] and operands[1] are equal.
8904 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
8905 =Q constraints to +Q and into insn condition add check
8906 that operands[0] is equal to either operands[1] or operands[2].
8909 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
8910 instead of handling MINUS_EXPR twice (once for each argument),
8911 canonicalize operand order and handle just once, use rtype where
8912 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
8915 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
8916 any values of __C while still being pattern recognizable as a simple
8919 2017-10-13 Richard Biener <rguenther@suse.de>
8921 PR tree-optimization/82451
8923 2017-10-02 Richard Biener <rguenther@suse.de>
8925 PR tree-optimization/82355
8926 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
8927 a mapping for the enclosing loop but avoid generating one for
8929 (copy_bb_and_scalar_dependences): Remove premature codegen
8930 error on PHIs in blocks duplicated into multiple places.
8931 * graphite-scop-detection.c
8932 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
8933 in the region use it as loop and nest to analyze the DR in.
8934 (try_generate_gimple_bb): Likewise.
8935 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
8936 (add_loop_constraints): For blocks in a loop not in the region
8937 create a dimension with a single iteration.
8938 * sese.h (gbb_loop_at_index): Remove assert.
8940 * cfgloop.c (loop_preheader_edge): For the loop tree root
8941 return the single successor of the entry block.
8942 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8943 Reset the SCEV hashtable and niters.
8944 * graphite-scop-detection.c
8945 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
8946 assert that we only have POLYNOMIAL_CHREC that vary in loops
8947 contained in the region.
8948 (scop_detection::graphite_can_represent_expr): Adjust.
8949 (scop_detection::stmt_has_simple_data_refs_p): For loops
8950 not in the region set loop to NULL. The nest is now the
8951 entry edge to the region.
8952 (try_generate_gimple_bb): Likewise.
8953 * sese.c (scalar_evolution_in_region): Adjust for
8954 instantiate_scev change.
8955 * tree-data-ref.h (graphite_find_data_references_in_stmt):
8956 Make nest parameter the edge into the region.
8957 (create_data_ref): Likewise.
8958 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
8959 entry edge into a region and adjust instantiate_scev calls.
8960 (create_data_ref): Likewise.
8961 (graphite_find_data_references_in_stmt): Likewise.
8962 (find_data_references_in_stmt): Pass the loop preheader edge
8963 from the nest argument.
8964 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
8965 parameter the edge into the region.
8966 (instantiate_parameters): Use the loop preheader edge as entry.
8967 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
8969 (get_instantiated_value_entry): Make instantiate_below parameter
8970 the edge into the region.
8971 (instantiate_scev_name): Likewise. Adjust dominance checks,
8972 when we cannot use loop-based instantiation instantiate by
8973 walking use-def chains.
8974 (instantiate_scev_poly): Adjust.
8975 (instantiate_scev_binary): Likewise.
8976 (instantiate_scev_convert): Likewise.
8977 (instantiate_scev_not): Likewise.
8978 (instantiate_array_ref): Remove.
8979 (instantiate_scev_3): Likewise.
8980 (instantiate_scev_2): Likewise.
8981 (instantiate_scev_1): Likewise.
8982 (instantiate_scev_r): Do not blindly handle N-operand trees.
8983 Do not instantiate array-refs. Handle all constants and invariants.
8984 (instantiate_scev): Make instantiate_below parameter
8985 the edge into the region.
8986 (resolve_mixers): Use the loop preheader edge for the region
8987 parameter to instantiate_scev_r.
8988 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
8990 2017-10-13 Richard Biener <rguenther@suse.de>
8992 PR tree-optimization/82525
8993 * graphite-isl-ast-to-gimple.c
8994 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
8996 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
8997 Fail code generation when we cannot represent the isl integer.
8998 (binary_op_to_tree): Elide modulo operations that are no-ops
8999 in the type we code generate. Remove now superfluous code
9002 2017-10-13 Richard Biener <rguenther@suse.de>
9004 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
9005 (scop_detection::harmful_loop_in_region): Remove premature
9006 IV type restriction.
9007 (scop_detection::graphite_can_represent_scev): We can handle
9008 pointer IVs just fine.
9010 2017-10-13 Alan Modra <amodra@gmail.com>
9012 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
9013 "Clobbers and Scratch Registers". Add paragraph on
9014 alternative to clobbers for scratch registers and OpenBLAS
9017 2017-10-13 Alan Modra <amodra@gmail.com>
9019 * doc/extend.texi (Clobbers): Correct vax example. Delete old
9020 example of a memory input for a string of known length. Move
9021 commentary out of table. Add a number of new examples
9022 covering array memory inputs.
9024 2017-10-12 Martin Liska <mliska@suse.cz>
9026 PR tree-optimization/82493
9027 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
9028 (test_range_functions): New function.
9029 (sbitmap_c_tests): Likewise.
9030 * selftest-run-tests.c (selftest::run_tests): Run new tests.
9031 * selftest.h (sbitmap_c_tests): New function.
9033 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
9035 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
9037 * config/rs6000/amo.h: Fix spacing issue.
9039 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9042 * config/i386/i386.md (*ashl<mode>3_mask_1,
9043 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
9044 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
9047 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
9049 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
9050 (profile_probability): Set max_probability
9051 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
9053 * profile-count.c (profile_probability::differs_from_p): Do not
9054 rely on max_probaiblity == 10000
9056 2017-10-12 Jeff Law <law@redhat.com>
9058 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
9061 2017-10-12 Martin Sebor <msebor@redhat.com>
9065 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
9066 (handle_alias_pairs): Call it.
9067 * common.opt (-Wattribute-alias): New option.
9068 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
9069 * doc/invoke.texi (-Wattribute-alias): Document.
9071 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
9074 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9076 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9078 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9079 (make_hard_regno_born, make_hard_regno_dead): Update
9080 bb_killed_pseudos and bb_gen_pseudos.
9082 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
9084 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
9086 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
9088 * config/alpha/alpha.c (alpha_split_conditional_move):
9089 Use std::swap instead of manually swapping.
9090 (alpha_stdarg_optimize_hook): Ditto.
9091 (alpha_canonicalize_comparison): Ditto.
9093 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9095 * tree-loop-distribution.c (struct builtin_info): New struct.
9096 (struct partition): Refactor fields into struct builtin_info.
9097 (partition_free): Free struct builtin_info.
9098 (build_size_arg_loc, build_addr_arg_loc): Delete.
9099 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
9100 information from struct builtin_info.
9101 (find_single_drs): New function refactored from classify_partition.
9102 Also moved builtin validity checks to this function.
9103 (compute_access_range, alloc_builtin): New functions.
9104 (classify_builtin_st, classify_builtin_ldst): New functions.
9105 (classify_partition): Refactor code into functions find_single_drs,
9106 classify_builtin_st and classify_builtin_ldst.
9107 (distribute_loop): Don't do runtime alias check when distributing
9109 (find_seed_stmts_for_distribution): New function.
9110 (pass_loop_distribution::execute): Refactor code finding seed
9111 stmts into above function. Support distribution for the innermost
9112 two-level loop nest. Adjust dump information.
9114 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9116 * tree-loop-distribution.c: Adjust the general comment.
9117 (NUM_PARTITION_THRESHOLD): New macro.
9118 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
9119 (classify_partition): Skip builtin pattern of loop nest's inner loop.
9120 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
9121 in call to build_partition_graph.
9122 (finalize_partitions): New parameter. Make loop distribution more
9123 conservative by fusing more partitions.
9124 (distribute_loop): Don't do runtime alias check in case of loop nest
9126 (find_seed_stmts_for_distribution): New function.
9127 (prepare_perfect_loop_nest): New function.
9128 (pass_loop_distribution::execute): Refactor code finding seed stmts
9129 and loop nest into above functions. Support loop nest distribution.
9130 Adjust dump information accordingly.
9132 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9134 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
9135 and set PTYPE_SEQUENTIAL for merged partition.
9137 2017-10-12 Richard Biener <rguenther@suse.de>
9139 PR tree-optimization/69728
9141 2017-09-19 Richard Biener <rguenther@suse.de>
9143 PR tree-optimization/69728
9144 * graphite-sese-to-poly.c (schedule_error): New global.
9145 (add_loop_schedule): Handle empty domain by failing the
9147 (build_original_schedule): Handle schedule_error.
9149 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
9150 domain by returning an unchanged schedule.
9152 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9154 * genrecog.c (validate_pattern): For VEC_SELECT verify that
9155 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
9157 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
9159 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
9162 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9165 * expr.c (store_field): Don't optimize away bitsize == 0 store
9166 from CALL_EXPR with addressable return type.
9168 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9170 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
9171 * config/rs6000/rs6000.md (sel): Delete mode attribute.
9172 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
9173 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
9174 TARGET_ISEL instead of TARGET_ISEL<sel>.
9176 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
9178 * config/rs6000/rs6000.c
9179 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
9181 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9183 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
9185 (reg_or_cint_operand): Fix comment.
9186 (reg_or_zero_operand): New predicate.
9187 * config/rs6000/rs6000-protos.h (output_isel): Delete.
9188 * config/rs6000/rs6000.c (output_isel): Delete.
9189 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
9190 instead of reg_or_cint_operand. Output instruction directly (not via
9192 (isel_unsigned_<mode>): Ditto.
9193 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
9194 gpc_reg_operand. Add an instruction alternative for this. Output
9195 instruction directly.
9196 (*isel_reversed_unsigned_<mode>): Ditto.
9198 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9200 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
9201 (TARGET_CANONICALIZE_COMPARISON): Define.
9203 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
9206 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
9207 Check whether the dest is REG before adding REG_EQUIV note.
9209 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9212 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9214 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9215 (make_hard_regno_born, make_hard_regno_dead): Update
9216 bb_killed_pseudos and bb_gen_pseudos.
9218 2017-10-11 Nathan Sidwell <nathan@acm.org>
9220 * incpath.h (enum incpath_kind): Name enum, prefix values.
9221 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
9222 * incpath.c (heads, tails): Use INC_MAX.
9223 (add_env_var_paths, add_standard_paths): Use incpath_kind.
9224 (merge_include_chains, split_quote_chain,
9225 register_include_chains): Update incpath_kind names.
9226 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
9227 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
9229 (add_framework_path, darwin_register_objc_includes): Likewise.
9230 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
9232 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9234 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
9235 Do not use float_operator operator predicate.
9236 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
9237 * config/i386/predicates.md (float_operator): Remove predicate.
9239 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9241 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
9242 (*jccxf_i387): Ditto.
9243 (*jcc<mode>_i387): Ditto.
9244 (*jccu<mode>_i387): Ditto.
9245 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
9246 (*jcc_*_i387 splitters): Remove.
9247 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
9248 * config/i386/i386.c (ix86_split_fp_branch): Remove.
9249 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
9252 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9254 * profile-count.h (slow_safe_scale_64bit): New function.
9255 (safe_scale_64bit): New inline.
9256 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
9257 * profile-count.c: Include wide-int.h
9258 (slow_safe_scale_64bit): New.
9260 2017-10-11 Nathan Sidwell <nathan@acm.org>
9262 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
9263 HAS_DECL_ASSEMBLER_NAME_P.
9264 * gimple-expr.c (gimple_decl_printable_name: Check
9265 HAS_DECL_ASSEMBLER_NAME_P too.
9266 * ipa-utils.h (type_in_anonymous_namespace_p): Check
9267 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
9268 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
9269 * passes.c (rest_of_decl_compilation): Check
9270 HAS_DECL_ASSEMBLER_NAME_P too.
9271 * recog.c (verify_changes): Likewise.
9272 * tree-pretty-print.c (dump_decl_name): Likewise.
9273 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
9275 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
9276 (SET_DECL_ASSEMBLER_NAME): Use it.
9277 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
9278 (COPY_DECL_ASSEMBLER_NAME): Likewise.
9279 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
9281 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9283 * config.gcc (i386, x86_64): Add extra objects.
9284 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
9285 (ix86_min_insn_size): Declare.
9286 (ix86_issue_rate): Declare.
9287 (ix86_adjust_cost): Declare.
9288 (ia32_multipass_dfa_lookahead): Declare.
9289 (ix86_macro_fusion_p): Declare.
9290 (ix86_macro_fusion_pair_p): Declare.
9291 (ix86_bd_has_dispatch): Declare.
9292 (ix86_bd_do_dispatch): Declare.
9293 (ix86_core2i7_init_hooks): Declare.
9294 (ix86_atom_sched_reorder): Declare.
9295 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
9296 (COSTS_N_BYTES): Move to x86-tune-costs.h.
9297 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
9298 (rip_relative_addr_p): Rename to ...
9299 (ix86_rip_relative_addr_p): ... this one; export.
9300 (memory_address_length): Update.
9301 (ix86_issue_rate): Move to x86-tune-sched.c.
9302 (ix86_flags_dependent): Move to x86-tune-sched.c.
9303 (ix86_agi_dependent): Move to x86-tune-sched.c.
9304 (exact_dependency_1): Move to x86-tune-sched.c.
9305 (exact_store_load_dependency): Move to x86-tune-sched.c.
9306 (ix86_adjust_cost): Move to x86-tune-sched.c.
9307 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
9308 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
9309 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
9310 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
9311 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
9312 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
9313 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
9314 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
9315 (min_insn_size): Rename to ...
9316 (ix86_min_insn_size): ... this one; export.
9317 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
9318 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
9319 (core2i7_first_cycle_multipass_backtrack): Move to
9320 x86-tune-sched-core.c.
9321 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
9322 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
9323 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
9324 (ix86_avoid_jump_mispredicts): Update.
9325 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
9326 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
9327 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
9328 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
9329 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
9330 (MAX_INSN): Move to ix86-tune-sched-bd.c.
9331 (MAX_IMM): Move to ix86-tune-sched-bd.c.
9332 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
9333 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
9334 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
9335 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
9336 (MAX_STORE): Move to ix86-tune-sched-bd.c.
9337 (BIG): Move to ix86-tune-sched-bd.c.
9338 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
9339 (enum insn_path): Move to ix86-tune-sched-bd.c.
9340 (get_mem_group): Move to ix86-tune-sched-bd.c.
9341 (is_cmp): Move to ix86-tune-sched-bd.c.
9342 (dispatch_violation): Move to ix86-tune-sched-bd.c.
9343 (is_branch): Move to ix86-tune-sched-bd.c.
9344 (is_prefetch): Move to ix86-tune-sched-bd.c.
9345 (init_window): Move to ix86-tune-sched-bd.c.
9346 (allocate_window): Move to ix86-tune-sched-bd.c.
9347 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
9348 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
9349 (process_end_window): Move to ix86-tune-sched-bd.c.
9350 (allocate_next_window): Move to ix86-tune-sched-bd.c.
9351 (find_constant): Move to ix86-tune-sched-bd.c.
9352 (get_num_immediates): Move to ix86-tune-sched-bd.c.
9353 (has_immediate): Move to ix86-tune-sched-bd.c.
9354 (get_insn_path): Move to ix86-tune-sched-bd.c.
9355 (get_insn_group): Move to ix86-tune-sched-bd.c.
9356 (count_num_restricted): Move to ix86-tune-sched-bd.c.
9357 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
9358 (add_insn_window): Move to ix86-tune-sched-bd.c.
9359 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
9360 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
9361 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
9362 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
9363 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
9364 (do_dispatch): Move to ix86-tune-sched-bd.c.
9365 (has_dispatch): Move to ix86-tune-sched-bd.c.
9366 * config/i386/t-i386: Add new object files.
9367 * config/i386/x86-tune-costs.h: New file.
9368 * config/i386/x86-tune-sched-atom.c: New file.
9369 * config/i386/x86-tune-sched-bd.c: New file.
9370 * config/i386/x86-tune-sched-core.c: New file.
9371 * config/i386/x86-tune-sched.c: New file.
9373 2017-10-11 Liu Hao <lh_mouse@126.com>
9375 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
9376 the generic version below instead.
9377 (should_colorize): Recognize Windows consoles as terminals
9379 * pretty-print.c [__MINGW32__] (write_all): New function.
9380 [__MINGW32__] (find_esc_head): Likewise.
9381 [__MINGW32__] (find_esc_terminator): Likewise.
9382 [__MINGW32__] (eat_esc_sequence): Likewise.
9383 [__MINGW32__] (mingw_ansi_fputs): New function that handles
9385 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
9388 2017-10-11 Richard Biener <rguenther@suse.de>
9390 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
9391 Properly call analyze_scalar_evolution with the loop of the stmt.
9393 2017-10-11 Richard Biener <rguenther@suse.de>
9395 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
9396 * tree-core.h (tree_base): Add chrec_var union member.
9397 * tree.h (CHREC_VAR): Remove.
9398 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
9399 * tree-chrec.h (build_polynomial_chrec): Adjust.
9400 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
9401 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
9403 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
9405 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
9406 * match.pd: ... here.
9407 ((T) X == (T) Y): Relax condition.
9409 2017-10-11 Bin Cheng <bin.cheng@arm.com>
9411 PR tree-optimization/82472
9412 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
9414 (break_alias_scc_partitions): Update postorder number.
9416 2017-10-11 Martin Liska <mliska@suse.cz>
9419 * opts.c (parse_no_sanitize_attribute): Do not use error_value
9421 * opts.h (parse_no_sanitize_attribute): Remove last argument.
9423 2017-10-11 Martin Liska <mliska@suse.cz>
9425 * print-rtl.c (print_insn): Move declaration of idbuf
9426 to same scope as name.
9428 2017-10-11 Martin Liska <mliska@suse.cz>
9433 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9436 2017-10-11 Martin Liska <mliska@suse.cz>
9439 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9442 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
9444 * config.gcc (default_gnu_indirect_function): Default to yes for
9445 arm*-*-linux* with glibc.
9447 2017-10-11 Richard Biener <rguenther@suse.de>
9449 * tree-scalar-evolution.c (get_scalar_evolution): Handle
9450 default-defs and types we do not want to analyze.
9451 (interpret_loop_phi): Replace unreachable code with an assert.
9452 (compute_scalar_evolution_in_loop): Remove and inline ...
9453 (analyze_scalar_evolution_1): ... here, replacing condition with
9454 what makes the intent clearer. Remove handling of cases
9455 get_scalar_evolution now handles.
9457 2017-10-10 Jim Wilson <wilson@tuliptree.org>
9459 PR rtl-optimization/81434
9460 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
9461 comment for main loop. In sched_group_found if, also add checks for
9462 pass and min_cost_group.
9464 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
9466 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
9467 (rs6000_insn_cost): New function.
9468 * config/rs6000/rs6000.md (cost): New attribute.
9470 2017-10-10 Jakub Jelinek <jakub@redhat.com>
9471 H.J. Lu <hongjiu.lu@intel.com>
9475 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
9476 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
9477 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
9478 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
9479 to OPTION_MASK_ISA_AVX512VL - builtins that have both
9480 OPTION_MASK_ISA_MMX and some other bit set require both
9481 mmx and the ISAs without the mmx bit.
9482 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
9483 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
9484 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
9485 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
9486 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
9487 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
9488 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
9489 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
9490 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
9491 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
9492 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
9493 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
9494 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
9495 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
9496 Add OPTION_MASK_ISA_MMX.
9498 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
9500 * config.gcc (armv7*-*-freebsd*): New target.
9501 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
9503 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
9505 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
9506 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
9509 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
9511 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
9512 a template parameter.
9513 (WIDE_INT_REF_FOR): Update accordingly.
9514 * tree.h (wi::int_traits <const_tree>): Delete.
9515 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
9516 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
9517 (wi::tree_to_wide_ref): New typedef.
9518 (wi::to_wide): New function.
9519 * calls.c (get_size_range): Use wi::to_wide when operating on
9521 * cgraph.c (cgraph_node::create_thunk): Likewise.
9522 * config/i386/i386.c (ix86_data_alignment): Likewise.
9523 (ix86_local_alignment): Likewise.
9524 * dbxout.c (stabstr_O): Likewise.
9525 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
9526 * expr.c (const_vector_from_tree): Likewise.
9527 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
9528 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
9529 (fold_negate_expr_1, int_const_binop_1, const_binop)
9530 (fold_convert_const_int_from_real, optimize_bit_field_compare)
9531 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
9532 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
9533 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
9534 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
9535 (fold_not_const, round_up_loc): Likewise.
9536 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
9537 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
9538 (alloca_call_type): Likewise.
9539 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
9540 * godump.c (go_output_typedef): Likewise.
9541 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
9542 * internal-fn.c (get_min_precision): Likewise.
9543 * ipa-cp.c (ipcp_store_vr_results): Likewise.
9544 * ipa-polymorphic-call.c
9545 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
9546 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
9547 (ipa_modify_call_arguments): Likewise.
9548 * match.pd: Likewise.
9549 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
9550 * print-tree.c (print_node_brief, print_node): Likewise.
9551 * stmt.c (expand_case): Likewise.
9552 * stor-layout.c (layout_type): Likewise.
9553 * tree-affine.c (tree_to_aff_combination): Likewise.
9554 * tree-cfg.c (group_case_labels_stmt): Likewise.
9555 * tree-data-ref.c (dr_analyze_indices): Likewise.
9556 (prune_runtime_alias_test_list): Likewise.
9557 * tree-dump.c (dequeue_and_dump): Likewise.
9558 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
9559 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
9560 * tree-pretty-print.c (dump_generic_node): Likewise.
9561 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
9562 (simple_iv_with_niters): Likewise.
9563 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
9564 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
9565 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
9566 * tree-ssa-loop-niter.c (split_to_var_and_offset)
9567 (refine_value_range_using_guard, number_of_iterations_ne_max)
9568 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
9569 (get_cst_init_from_scev, record_nonwrapping_iv)
9570 (scev_var_range_cant_overflow): Likewise.
9571 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
9572 * tree-ssa-pre.c (compute_avail): Likewise.
9573 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
9574 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
9575 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
9576 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
9577 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
9578 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
9579 (dump_case_nodes, try_switch_expansion): Likewise.
9580 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
9581 (vect_do_peeling): Likewise.
9582 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
9583 * tree-vect-stmts.c (vectorizable_load): Likewise.
9584 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
9585 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
9586 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
9587 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
9588 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
9589 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
9590 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
9591 (evrp_dom_walker::before_dom_children): Likewise.
9592 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
9593 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
9594 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
9595 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
9596 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
9597 (get_range_pos_neg): Likewise.
9598 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
9599 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
9600 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
9601 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
9602 * config/avr/avr.c (avr_fold_builtin): Likewise.
9603 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
9604 * config/msp430/msp430.c (msp430_attr): Likewise.
9605 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
9606 * config/powerpcspe/powerpcspe-c.c
9607 (altivec_resolve_overloaded_builtin): Likewise.
9608 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
9609 (rs6000_expand_ternop_builtin): Likewise.
9610 * config/rs6000/rs6000-c.c
9611 (altivec_resolve_overloaded_builtin): Likewise.
9612 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
9613 (rs6000_expand_ternop_builtin): Likewise.
9614 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
9616 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9618 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
9619 when copying loop nest with only one inner loop.
9621 2017-10-10 Richard Biener <rguenther@suse.de>
9623 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
9624 blocks if SCEV is active.
9625 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
9627 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
9628 (scev_initialize): Assert we are not yet initialized.
9630 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9632 * tree-loop-distribution.c (generate_loops_for_partition): Remove
9633 inner loop's exit stmt by making it always exit the loop, otherwise
9634 we would generate an infinite empty loop.
9636 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9638 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
9639 renaming variables in new preheader if it's deleted.
9641 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9643 * tree-loop-distribution.c (struct partition): Remove unused field
9644 loops of the structure.
9645 (partition_alloc, partition_free): Ditto.
9646 (build_rdg_partition_for_vertex): Ditto.
9648 2017-10-09 Jeff Law <law@redhat.com>
9650 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
9651 return type to match prototype and documentation.
9653 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9655 * config/rs6000/rs6000.c (processor_costs): Move to ...
9656 * config/rs6000/rs6000.h: ... here.
9657 (rs6000_cost): Declare.
9659 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
9661 * except.c (setjmp_fn): New global variable.
9662 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
9663 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
9664 if DONT_USE_BUILTIN_SETJMP is defined.
9666 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9668 * target.def (insn_cost): New hook.
9669 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
9670 * doc/tm.texi: Regenerate.
9671 * rtlanal.c (insn_cost): Use the new hook.
9673 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9675 * combine.c (combine_validate_cost): Compute the new insn_cost,
9676 not just pattern_cost.
9677 (try_combine): Adjust comment.
9679 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9681 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
9683 * combine.c (uid_insn_cost): Adjust comment.
9684 (combine_validate_cost): Adjust comment. Use pattern_cost instead
9686 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
9687 * dse.c (find_shift_sequence): Ditto.
9688 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
9689 (bb_valid_for_noce_process_p): Use pattern_cost.
9690 * rtl.h (insn_rtx_cost): Delete.
9691 (pattern_cost): New prototype.
9692 (insn_cost): New prototype.
9693 * rtlanal.c (insn_rtx_cost): Rename to...
9694 (pattern_cost): ... this.
9697 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
9699 * config/i386/i386.md (*jcc_2): Remove insn pattern.
9700 (*jcc<mode>_0_r_i387): Ditto.
9701 (*jccxf_r_i387): Ditto.
9702 (*jcc<mode>_r_i387): Ditto.
9703 (*jccu<mode>_r_i387): Ditto.
9704 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
9705 (*jcc): Rename from *jcc_1.
9707 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9709 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
9710 deferred rescans after the lvx/stvx recombination pre-pass.
9712 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
9714 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
9715 memory operation instruction support.
9716 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
9717 (rs6000-ibm-aix[789]*): Likewise.
9718 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
9719 Document new functions.
9721 2017-10-09 Richard Biener <rguenther@suse.de>
9723 PR tree-optimization/82397
9724 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
9725 equality only for semantically equal trees.
9727 2017-10-09 Richard Biener <rguenther@suse.de>
9729 PR tree-optimization/82449
9730 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
9731 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
9732 allow constant addresses.
9733 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
9736 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9738 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
9741 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9744 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
9747 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9750 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
9752 2017-10-09 Jakub Jelinek <jakub@redhat.com>
9755 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
9756 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
9758 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
9760 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
9761 (WI_BINARY_PREDICATE_RESULT): Likewise.
9762 (wi::binary_traits::operator_result): New type.
9763 (wi::binary_traits::predicate_result): Likewise.
9764 (generic_wide_int::operator~, unary generic_wide_int::operator-)
9765 (generic_wide_int::operator==, generic_wide_int::operator!=)
9766 (generic_wide_int::operator&, generic_wide_int::and_not)
9767 (generic_wide_int::operator|, generic_wide_int::or_not)
9768 (generic_wide_int::operator^, generic_wide_int::operator+
9769 (binary generic_wide_int::operator-, generic_wide_int::operator*):
9771 (operator~, unary operator-, operator==, operator!=, operator&)
9772 (operator|, operator^, operator+, binary operator-, operator*): New
9774 * expr.c (get_inner_reference): Use wi::bit_and_not.
9775 * fold-const.c (fold_binary_loc): Likewise.
9776 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
9777 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
9778 (bit_value_binop): Likewise.
9779 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
9780 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
9781 (extract_range_from_binary_expr_1): Likewise.
9782 (masked_increment): Likewise.
9783 (simplify_bit_ops_using_ranges): Likewise.
9785 2017-10-09 Martin Jambor <mjambor@suse.cz>
9788 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
9789 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
9790 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
9792 (hsa_fixup_mov_insn_type): New function.
9793 (hsa_op_with_type::get_in_type): Use it.
9794 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
9795 immediates in an assert.
9796 (hsa_op_with_type::extend_int_to_32bit): New method.
9797 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
9798 types. Convert to dest type if necessary.
9799 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
9800 (reg_for_gimple_ssa): Pass false as min32int to
9801 hsa_type_for_scalar_tree_type.
9802 (gen_hsa_addr): Fixup type when creating addresable temporary.
9803 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
9804 (gen_hsa_unary_operation): Extend operands and convert to dest type if
9805 necessary. Call hsa_fixup_mov_insn_type.
9806 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
9807 extend operands and convert to dest type if necessary.
9808 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
9809 to dest type if necessary.
9810 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
9811 if conversion nt necessary and size matches.
9812 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
9813 to dest type if necessary.
9814 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
9815 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
9817 (gen_hsa_clrsb): Likewise.
9818 (gen_hsa_ffs): Likewise.
9819 (gen_hsa_divmod): Extend operands and convert to dest type if
9821 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
9823 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
9825 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
9826 Remove empty default arguments. Use a brace block as output
9828 (conditional return): Ditto.
9830 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
9831 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
9833 (group_ending_nop): Ditto.
9834 (doloop_end): Ditto.
9835 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
9836 (splitters for those): Ditto.
9838 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
9840 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
9841 a conditional jump (and the compare for it) so that pc_rtx is the
9843 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
9844 for the deleted and renamed ctr<mode>_internal[234] patterns.
9845 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
9846 Delete second conditional return pattern.
9847 (ctr<mode>_internal2): Delete this second bdnz pattern.
9848 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
9849 (ctr<mode>_internal4): Delete this second bdz pattern.
9851 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9853 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
9854 (always_initialized_rtx_for_ssa_name_p): New predicate.
9855 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
9856 (finish_out_of_ssa): Free new field of SA.
9857 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
9858 * tree-ssa-coalesce.c: Include tree-ssa.h.
9859 (get_parm_default_def_partitions): Remove extern keyword.
9860 (get_undefined_value_partitions): New function.
9861 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
9862 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
9864 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
9866 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9868 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
9870 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9872 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
9873 for targets that preffer 128bit.
9875 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9877 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
9879 2017-10-08 Olivier Hainque <hainque@adacore.com>
9881 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
9882 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
9884 (thumb_set_return_address): Likewise.
9886 2017-10-08 Olivier Hainque <hainque@adacore.com>
9888 * common/config/arm/arm-common.c (arm_except_unwind_info):
9889 Handle DWARF2_UNWIND_INFO.
9891 2017-10-07 Michael Collison <michael.collison@arm.com>
9893 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
9896 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
9898 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
9899 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
9900 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
9901 defined, force the creation of a new block for a dispatch label.
9903 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
9905 * invoke.texi (Wsuggest-attribute=cold): Document.
9906 * common.opt (Wsuggest-attribute=cold): New
9907 * ipa-pure-const.c (warn_function_cold): New function.
9908 * predict.c (compute_function_frequency): Use it.
9909 * predict.h (warn_function_cold): Declare.
9911 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
9913 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
9916 2017-10-06 Martin Liska <mliska@suse.cz>
9918 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
9919 keyword for member functions.
9920 (struct sanopt_tree_couple): New struct.
9921 (struct sanopt_tree_couple_hash): New function.
9922 (struct sanopt_ctx): Add new hash_map.
9923 (has_dominating_ubsan_ptr_check): New function.
9924 (record_ubsan_ptr_check_stmt): Likewise.
9925 (maybe_optimize_ubsan_ptr_ifn): Likewise.
9926 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
9927 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
9929 2017-10-06 Sudakshina Das <sudi.das@arm.com>
9932 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
9933 aarch64_simd_valid_immediate on CONST_VECTORs.
9934 (aarch64_reg_or_bic_imm): Likewise.
9936 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
9938 PR rtl-optimization/82396
9939 * haifa-sched.c (ready_sort_real): Disable qsort checking.
9941 2017-10-06 Sebastian Pop <sebpop@gmail.com>
9943 * graphite-dependences.c (scop_get_reads): Move code to...
9944 (scop_get_must_writes): Move code to...
9945 (scop_get_may_writes): Move code to...
9946 (scop_get_reads_and_writes): ... here.
9947 (scop_get_dependences): Call scop_get_reads_and_writes.
9949 2017-10-06 Jakub Jelinek <jakub@redhat.com>
9951 PR tree-optimization/82434
9952 * fold-const.h (can_native_encode_type_p,
9953 can_native_encode_string_p): Remove.
9954 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
9955 don't encode anything, just return what would be otherwise returned.
9956 (native_encode_fixed, native_encode_complex, native_encode_vector):
9958 (native_encode_string): Likewise. Inline by hand
9959 can_native_encode_string_p.
9960 (can_native_encode_type_p): Remove.
9961 (can_native_encode_string_p): Remove.
9962 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
9963 STRING_CSTs using can_native_encode_string_p, test all
9964 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
9965 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
9966 argument from native_encode_expr.
9967 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
9968 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
9971 2017-10-06 Richard Biener <rguenther@suse.de>
9973 PR tree-optimization/82397
9974 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
9975 operand_equal_p but rely on data_ref_compare_tree for detecting
9977 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
9978 to match up with dr_group_sort_cmp.
9980 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9983 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
9985 * config/s390/s390-builtin-types.def: Regenerate.
9987 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9990 * config/s390/s390-builtin-types.def: Regenerate.
9991 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
9992 Change flag from B_VXE to B_VX.
9993 (s390_vec_min_dbl): Remove B_VXE flag.
9995 2017-10-06 Richard Biener <rguenther@suse.de>
9997 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
9998 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
9999 translate_isl_ast_to_gimple::is_valid_rename,
10000 translate_isl_ast_to_gimple::get_rename,
10001 translate_isl_ast_to_gimple::get_def_bb_for_const,
10002 translate_isl_ast_to_gimple::get_new_name,
10003 translate_isl_ast_to_gimple::collect_all_ssa_names,
10004 translate_isl_ast_to_gimple::copy_loop_phi_args,
10005 translate_isl_ast_to_gimple::collect_all_ssa_names,
10006 translate_isl_ast_to_gimple::copy_loop_phi_args,
10007 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
10008 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
10009 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
10010 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
10011 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
10012 translate_isl_ast_to_gimple::copy_cond_phi_args,
10013 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
10014 translate_isl_ast_to_gimple::edge_for_new_close_phis,
10015 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
10016 translate_isl_ast_to_gimple::rename_uses,
10017 translate_isl_ast_to_gimple::rename_all_uses): Remove.
10018 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
10019 (set_rename_for_each_def): Likewise.
10020 (graphite_copy_stmts_from_block): Handle debug stmt resetting
10021 here. Handle rewriting SCEV analyzable uses here.
10022 (copy_bb_and_scalar_dependences): Generate code for PHI
10024 (graphite_regenerate_ast_isl): Adjust.
10025 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
10026 (add_write, add_read): New functions.
10027 (build_cross_bb_scalars_def): Use it and simplify.
10028 (build_cross_bb_scalars_use): Likewise.
10029 (graphite_find_cross_bb_scalar_vars): Inline into...
10030 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
10031 simulating out-of-SSA. Compute liveout and add dependencies.
10032 (build_scops): Force an empty entry block.
10033 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
10035 (sese_build_liveouts): Declare.
10036 (sese_trivially_empty_bb_p): Likewise.
10037 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
10038 compute liveout and debug_liveout.
10039 (sese_bad_liveouts_use): Remove.
10040 (sese_reset_debug_liveouts_bb): Likewise.
10041 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
10042 (sese_build_liveouts): Build liveout and debug_liveout and store
10044 (new_sese_info): Adjust.
10045 (free_sese_info): Likewise.
10046 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
10047 do not build liveout here.
10048 (move_sese_in_condition): Adjust region entry.
10049 (scev_analyzable_p): Match up with chrec_apply requirements.
10050 (sese_trivially_empty_bb_p): New.
10051 * tree-into-ssa.c (get_reaching_def): Properly support generating
10052 default-defs for incremental rewrite of anonymous names.
10054 2017-10-06 Richard Biener <rguenther@suse.de>
10056 * graphite-sese-to-poly.c (extract_affine): For casts increasing
10057 precision do not perform modulo reduction.
10059 2017-10-06 Richard Biener <rguenther@suse.de>
10061 PR tree-optimization/82436
10062 * tree-vect-slp.c (vect_supported_load_permutation_p): More
10063 conservatively choose the vectorization factor when checking
10064 whether we can perform the required load permutation.
10065 (vect_transform_slp_perm_load): Assert when we may not fail.
10067 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
10069 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
10070 message for incompatible -msdata=* and -mcall-* options.
10072 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10074 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
10075 rate for post-reload scheduling.
10077 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10079 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
10081 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10083 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
10084 to improve monte carlo in scimark.
10086 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10088 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
10089 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
10090 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
10091 pentium4_cost, nocona_cost): Set reassociation width to 1.
10092 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
10093 width to 2 for fp operations and 1 otherwise.
10094 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
10096 (atom_cost): Set reassociation width to 2.
10097 (slm_cost, generic_cost): Set fp reassociation width
10098 to 2 and 1 otherwise.
10099 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
10100 (core_cost): Set fp reassociation width to 4 and vector to 2.
10101 (ix86_reassociation_width): Rewrite using cost table; special case
10102 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
10103 and TARGET_AVX128_OPTIMAL.
10104 * config/i386/i386.h (processor_costs): Add
10105 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
10106 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
10107 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
10108 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
10109 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
10110 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
10112 2017-10-05 Nathan Sidwell <nathan@acm.org>
10114 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
10116 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10118 * config/arm/arm.c (arm_test_fpu_data): New.
10119 (arm_run_selftests): Call arm_test_fpu_data.
10121 2017-10-04 Nathan Sidwell <nathan@acm.org>
10123 * toplev.c (toplev::main): Remove excess parens on pretty_printer
10125 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
10127 2017-10-04 Sudakshina Das <sudi.das@arm.com>
10129 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
10130 check type for aarch64_simd_valid_immediate.
10131 (aarch64_output_simd_mov_immediate): Update prototype.
10132 (aarch64_simd_valid_immediate): Update prototype.
10133 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
10134 support for ORR-immediate.
10135 (and<mode>3): modified pattern to add support for BIC-immediate.
10136 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
10137 now checks for valid immediate for BIC and ORR based on new enum
10139 (aarch64_output_simd_mov_immediate): Function now used to output
10140 BIC/ORR imm as well based on new enum argument.
10141 * config/aarch64/constraints.md (Do): New vector immediate constraint.
10143 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
10144 (aarch64_reg_or_bic_imm): Likewise.
10146 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10148 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
10149 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
10151 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10155 PR rtl-optimization/82396
10156 * haifa-sched.c (autopref_multipass_init): Simplify
10158 (autopref_rank_data): Simplify sort order.
10159 * sched-int.h (autopref_multipass_data_): Remove
10160 multi_mem_insn_p, min_offset and max_offset.
10162 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10164 * doc/sourcebuild.texi: Document vect_peeling_profitable.
10166 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10168 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
10169 vect_doubleint_cvt.
10171 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10173 * doc/sourcebuild.texi: Document vect_long_mult.
10175 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
10177 PR tree-optimization/82413
10178 * fold-const.c (build_range_check): Use widest_int when comparing
10179 the maximum ETYPE value with HIGH.
10181 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10183 PR rtl-optimization/82396
10184 * haifa-sched.c (autopref_multipass_init): Simplify
10186 (autopref_rank_data): Simplify sort order.
10187 * sched-int.h (autopref_multipass_data_): Remove
10188 multi_mem_insn_p, min_offset and max_offset.
10190 2017-10-04 Jakub Jelinek <jakub@redhat.com>
10192 PR tree-optimization/82381
10193 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
10194 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
10197 PR tree-optimization/82374
10198 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
10199 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
10200 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
10201 current_function_decl to the new decl.
10203 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
10205 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
10206 helper macro for IEEE float128 hardware built-in functions.
10207 (SQRTF128_ODD): Add built-in functions with the round-to-odd
10209 (TRUNCF128_ODD): Likewise.
10210 (ADDF128_ODD): Likewise.
10211 (SUBF128_ODD): Likewise.
10212 (MULF128_ODD): Likewise.
10213 (DIVF128_ODD): Likewise.
10214 (FMAF128_ODD): Likewise.
10215 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
10216 UNSPEC_TRUNC_ROUND_TO_ODD.
10217 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
10218 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
10219 floating point round to odd instructions.
10220 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
10221 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
10222 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
10223 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
10224 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
10225 (trunc<mode>sf2_hw): Change the truncate with round to odd
10226 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
10227 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
10228 to odd hardware instructions.
10229 (sub<mode>3_odd): Likewise.
10230 (mul<mode>3_odd): Likewise.
10231 (div<mode>3_odd): Likewise.
10232 (sqrt<mode>2_odd): Likewise.
10233 (fma<mode>4_odd): Likewise.
10234 (fms<mode>4_odd): Likewise.
10235 (nfma<mode>4_odd): Likewise.
10236 (nfms<mode>4_odd): Likewise.
10237 (trunc<mode>df2_odd): Change the truncate with round to odd
10238 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
10240 * doc/extend.texi (PowerPC built-in functions): Update documentation
10241 for existing IEEE float128-bit built-in functions. Add built-in
10242 functions that generate the IEEE 128-bit floating point round to
10245 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
10247 PR rtl-optimization/77729
10248 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
10249 to (X&(C1&~C2))|C2 transformations.
10251 2017-10-03 Martin Jambor <mjambor@suse.cz>
10253 PR tree-optimization/82363
10254 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
10255 mismatch, mark lacc written regardless of racc.
10257 2017-10-03 Jakub Jelinek <jakub@redhat.com>
10259 PR tree-optimization/82381
10260 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
10261 stmt_to_insert nor wheather SSA_NAMEs are default defs.
10262 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
10263 fallthrough into reassoc_stmt_dominates_stmt_p.
10266 * combine.c (combine_instructions): Don't combine in unreachable
10269 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
10272 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
10273 function to not use the have_cpu variable. Do not set cpu_index,
10274 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
10275 or the default cpu.
10276 (rs6000_valid_attribute_p): Remove duplicate initializations of
10277 old_optimize and func_optimize.
10278 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
10279 (rs6000_activate_target_options): Make global.
10280 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
10283 2017-10-02 Jakub Jelinek <jakub@redhat.com>
10285 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
10286 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
10287 Set *poffset to 0 and *psize and *pmax_size to -1 if
10288 *poffset + *psize overflows in HOST_WIDE_INT.
10290 PR tree-optimization/82387
10291 PR tree-optimization/82388
10292 PR tree-optimization/82389
10293 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
10294 instead of live_bytes non-NULL.
10296 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
10299 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
10302 2017-10-02 Richard Biener <rguenther@suse.de>
10304 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
10305 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
10306 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
10308 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10310 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
10311 requested precision matches the type's.
10312 * calls.c (alloc_max_size): Calculate the new candidate size as
10313 a widest_int and use wi::to_widest when comparing it with the
10314 current candidate size.
10315 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
10316 zero rather than integer_zero_node.
10317 * match.pd: Check for a no-op conversion before using wi::add
10318 rather than after. Use tree_to_uhwi when summing small shift
10319 counts into an unsigned int.
10321 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10322 Alan Hayward <alan.hayward@arm.com>
10323 David Sherwood <david.sherwood@arm.com>
10326 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
10327 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
10328 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
10329 POINTER_AND_FP_REGS.
10331 2017-10-02 Richard Biener <rguenther@suse.de>
10333 PR tree-optimization/82355
10334 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
10335 a mapping for the enclosing loop but avoid generating one for
10336 the loop tree root.
10337 (copy_bb_and_scalar_dependences): Remove premature codegen
10338 error on PHIs in blocks duplicated into multiple places.
10339 * graphite-scop-detection.c
10340 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
10341 in the region use it as loop and nest to analyze the DR in.
10342 (try_generate_gimple_bb): Likewise.
10343 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
10344 (add_loop_constraints): For blocks in a loop not in the region
10345 create a dimension with a single iteration.
10346 * sese.h (gbb_loop_at_index): Remove assert.
10348 2017-10-01 Kevin Buettner <kevinb@redhat.com>
10350 * omp-expand.c (adjust_context_scope): New function.
10351 (expand_parallel_call): Call adjust_context_scope.
10353 2017-10-01 Jeff Law <law@redhat.com>
10355 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
10356 dom_opt_dom_walker class with direct access to private members.
10357 Add comments. Call test_for_singularity.
10358 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
10359 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
10360 m_dummy_cond anymore.
10361 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
10363 (pass_dominator:execute): Build the dummy_cond here and pass it
10364 to the dom_opt_dom_walker ctor.
10365 (test_for_singularity): New function.
10367 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
10368 Maya Rashish <coypu@sdf.org>
10370 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
10371 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
10372 (alpha*-*-netbsd*) Use nbsd_tm_file.
10373 (arm*-*-netbsdelf*) Likewise.
10374 (i[34567]86-*-netbsdelf*) Likewise.
10375 (x86_64-*-netbsd*) Likewise.
10376 (mips*-*-netbsd*) Likewise.
10377 (powerpc-*-netbsd*) Likewise.
10378 (sh*-*-netbsd*) Likewise.
10379 (sparc-*-netbsdelf*) Likewise.
10380 (sparc64-*-netbsd*) Likewise.
10381 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
10383 (vax-*-netbsdelf*) Likewise.
10384 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
10385 (UINT_FAST8_TYPE) Likewise.
10386 (INT_FAST16_TYPE) Check CHAR_FAST16.
10387 (UINT_FAST16_TYPE) Likewise.
10389 2017-09-30 Jakub Jelinek <jakub@redhat.com>
10392 * config/i386/i386.md
10393 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
10394 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
10395 *divmodsi4_zext_2): New define_insn_and_split.
10396 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
10397 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
10398 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
10399 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
10400 New define_insn_and_split.
10401 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
10402 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
10403 operands[1] having DImode when mode is SImode.
10405 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
10406 always SImode for DIV and MOD in REG_EQUAL notes.
10408 2017-09-29 Yury Gribov <tetra2005@gmail.com>
10410 PR middle-end/82319
10411 * match.pd: Fix handling of NaNs in pattern.
10413 2017-09-29 Jeff Law <law@redhat.com>
10415 * sbitmap.c (bitmap_bit_in_range_p): New function.
10416 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
10417 * tree-ssa-dse.c (live_bytes_read): New function.
10418 (dse_classify_store): Ignore reads of dead bytes.
10420 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
10421 typos and whitespace errors.
10422 * config/i386/predicates.md (address_no_seg_operand): Likewise.
10423 * config/s390/s390.c (s390_emit_prologue): Likewise.
10425 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10428 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
10429 with a symbol for LRA.
10431 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10433 PR rtl-optimization/82338
10434 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
10436 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
10438 * genmodes.c (calc_wider_mode): Suppress qsort macro.
10439 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
10440 (qsort_chk): Declare.
10441 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
10442 (qsort_chk): New function.
10444 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10446 PR tree-optimization/82337
10447 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
10448 phi definition if the PHI result appears in an abnormal PHI.
10449 (find_basis_for_base_expr): Don't record a basis if the LHS of the
10450 basis appears in an abnormal PHI.
10452 2017-09-29 Richard Biener <rguenther@suse.de>
10454 * graphite-isl-ast-to-gimple.c
10455 (translate_isl_ast_to_gimple::set_codegen_error): New function.
10456 (binary_op_to_tree): Use it.
10457 (get_rename_from_scev): Likewise.
10458 (copy_loop_phi_nodes): Likewise.
10459 (copy_bb_and_scalar_dependences): Likewise.
10460 (translate_pending_phi_nodes): Likewise.
10462 2017-09-29 Jakub Jelinek <jakub@redhat.com>
10465 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
10466 for movabsq $(i32 << shift), r64.
10468 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
10470 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
10471 index when encoding %esp as %rsp to avoid 0x67 prefix.
10473 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
10475 * config/i386/i386.md (*movsf_internal, *movdf_internal):
10476 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
10478 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
10480 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
10481 Extensions with more than 16 double VFP registers.
10482 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
10483 to_clear_mask and all code related to it. Replace the remaining
10484 entry by a sbitmap and adapt code accordingly.
10486 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
10488 * brig-builtins.def: Change pure attributes to const.
10490 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10492 * config.gcc (default_gnu_indirect_function): Default to yes for
10493 sparc*-*-linux* with glibc.
10495 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10497 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
10498 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
10499 when creating .init_array and .fini_array sections with priority
10502 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
10505 * config/aarch64/aarch64.c
10506 (aarch64_builtin_support_vector_misalignment): Always return false
10507 when misalignment is unknown.
10509 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
10511 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
10512 this function to return false if the definition used by the swap
10513 instruction is artificial, or if the memory address from which the
10514 constant value is loaded is not represented by a base address held
10515 in a register or if the base address register is a frame or stack
10516 pointer. Additionally, return false if the base address of the
10517 loaded constant is a SYMBOL_REF but is not considered to be a
10519 (replace_swapped_load_constant): New function.
10520 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
10521 loaded constant vector with a load of a swapped constant vector.
10523 2017-09-27 Carl Love <cel@us.ibm.com>
10525 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
10526 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
10527 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
10530 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
10532 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
10533 first, always call autopref_rank_data otherwise.
10535 2017-09-27 Richard Biener <rguenther@suse.de>
10537 * graphite-scop-detection.c (find_scop_parameters): Move
10538 loop bound handling ...
10539 (gather_bbs::before_dom_children): ... here, avoiding the need
10540 to build scop_info->loop_nest.
10541 (record_loop_in_sese): Remove.
10542 * sese.h (sese_info_t::loop_nest): Remove.
10543 * sese.c (new_sese_info): Do not allocate loop_nest.
10544 (free_sese_info): Do not free loop_nest.
10546 2017-09-27 Jakub Jelinek <jakub@redhat.com>
10549 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
10550 lhs from calls if the lhs has addressable type.
10552 2017-09-27 Richard Biener <rguenther@suse.de>
10554 * graphite.h (scop::max_alias_set): New member.
10555 * graphite-scop-detection.c: Remove references to non-existing
10556 --param in comments.
10557 (build_alias_sets): Record the maximum alias set used for drs.
10558 (build_scops): Support zero as unlimited for
10559 --param graphite-max-arrays-per-scop.
10560 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
10561 and inline into ...
10562 (build_poly_sr_1): ... here. Compute alias set based on the
10563 maximum alias set used for drs rather than
10564 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
10566 2017-09-27 Richard Biener <rguenther@suse.de>
10568 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
10569 --param loop-block-tile-size=0 to disable tiling.
10571 2017-09-27 Richard Biener <rguenther@suse.de>
10573 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
10574 (graphite-max-nb-scop-params): Document special value zero.
10575 * domwalk.h (dom_walker::STOP): New symbolical constant.
10576 (dom_walker::dom_walker): Add optional parameter for bb to
10578 (dom_walker::~dom_walker): Declare.
10579 (dom_walker::before_dom_children): Document STOP return value.
10580 (dom_walker::m_user_bb_to_rpo): New member.
10581 (dom_walker::m_bb_to_rpo): Likewise.
10582 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
10583 mapping here if not provided by the user.
10584 (dom_walker::~dom_walker): Free bb to RPO mapping if not
10585 provided by the user.
10586 (dom_walker::STOP): Define.
10587 (dom_walker::walk): Do not compute bb to RPO mapping here.
10588 Support STOP return value from before_dom_children to stop
10590 * graphite-optimize-isl.c (optimize_isl): If the schedule
10591 is the same still generate code if -fgraphite-identity
10592 or -floop-parallelize-all are given.
10593 * graphite-scop-detection.c: Include cfganal.h.
10594 (gather_bbs::gather_bbs): Get and pass through bb to RPO
10596 (gather_bbs::before_dom_children): Return STOP for BBs
10598 (build_scops): Compute bb to RPO mapping and pass it to
10599 the domwalk. Treat --param graphite-max-nb-scop-params=0
10600 as not limiting the number of params.
10601 * graphite.c (graphite_initialize): Remove limit on the
10602 number of basic-blocks in a function.
10603 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
10604 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
10605 default value of 10.
10607 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10609 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
10610 Adjust code to eliminate needing to do the shift right 32-bits
10611 operation after XSCVDPSPN.
10613 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10615 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
10616 ((X / Y) != 0 -> X >= Y): Likewise.
10618 2017-09-26 Carl Love <cel@us.ibm.com>
10620 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
10621 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
10622 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
10623 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
10624 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
10625 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
10626 definitions and overloading.
10627 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
10628 statement for P9V_BUILTIN_XST_LEN_R.
10629 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
10630 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
10631 define_expand and define_insn for the instructions and builtins.
10632 * doc/extend.texi: Update the built-in documentation file for the new
10633 built-in functions.
10634 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
10635 define_insn for the instructions
10637 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
10640 * gcc/config/netbsd-protos.h: New file.
10641 * gcc/config/netbsd.c: New file.
10642 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
10643 * gcc/config/t-netbsd: New file.
10644 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
10645 (tmake_file) Add t-netbsd.
10646 (extra_objs) Add netbsd.o.
10648 2017-09-26 Janus Weil <janus@gcc.gnu.org>
10652 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
10654 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10656 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
10657 sign extension from a vector register to a GPR by doing a 32-bit
10658 direct move and then an EXTSW.
10659 (extendsi<mode>2 splitter): Likewise.
10660 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
10661 right or vector extract after doing XSCVDPSPN. Use
10662 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
10664 (movdi_from_sf_zero_ext): Likewise.
10665 (reload_gpr_from_vsxsf): Likewise.
10666 (p8_mfvsrd_4_disf): Delete, no longer used.
10667 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
10668 then needing to move the SFmode to a GPR to use the XSCVDPSP
10669 instruction instead of FRSP and XSCVDPSPN.
10670 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
10671 it is adjacent to the other XSCVSPDP insns.
10672 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
10673 SFmode to be in traditional Altivec registers.
10674 (vsx_xscvdpspn): Eliminate useless alternative constraint.
10675 (vsx_xscvspdpn): Likewise.
10676 (vsx_xscvspdpn_scalar): Likewise.
10678 2017-09-26 Martin Jambor <mjambor@suse.cz>
10680 * tree-sra.c (compare_access_positions): Put integral types first,
10681 stabilize sorting of integral types, remove conditions putting
10682 non-full-precision integers last.
10683 (sort_and_splice_var_accesses): Disable scalarization if a
10684 non-integert would be represented by a non-full-precision integer.
10686 2017-09-26 Joseph Myers <joseph@codesourcery.com>
10688 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
10689 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
10690 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
10691 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
10692 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
10693 conditionals inside the function instead of around it. Call
10694 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
10695 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
10697 2017-09-26 Richard Biener <rguenther@suse.de>
10699 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
10701 (scop_detection::build_scop_breadth): ... this. Removed.
10702 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
10703 (scop_detection::harmful_stmt_in_bb): Likewise.
10704 (scop_detection::graphite_can_represent_stmt): Likewise.
10705 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
10706 (scop_detection::can_represent_loop): Remove recursion, fold in ...
10707 (scop_detection::can_represent_loop_1): ... this. Removed.
10708 (scop_detection::harmful_loop_in_region): Simplify after inlining
10709 the above and remove more quadraticness.
10710 (build_scops): Adjust.
10711 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
10714 2017-09-26 Jakub Jelinek <jakub@redhat.com>
10717 * config/i386/i386.c (ix86_print_operand_address_as): Only test
10718 REGNO (base) == SP_REG if base is a REG.
10720 PR middle-end/35691
10721 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
10722 if it is different SSA_NAME.
10723 (optimize_range_tests_cmp_bitwise): New function.
10724 (optimize_range_tests): Call it.
10726 2017-09-26 Richard Biener <rguenther@suse.de>
10728 PR tree-optimization/82321
10729 * graphite.c (canonicalize_loop_closed_ssa): Properly check
10730 for the def being inside the loop.
10732 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10734 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
10736 * config/s390/s390-builtins.def: Fix constraint on op4.
10738 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10740 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
10741 independent expanders.
10742 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
10743 ("vec_ordered", "vec_unordered"): New expanders.
10745 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10747 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
10750 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10752 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
10753 vec_unpacks_lo_v16qi.
10754 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
10756 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10758 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
10759 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
10760 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
10762 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10764 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
10766 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
10768 ("*vec_slb<mode>"): New insn pattern.
10769 ("vec_shr_<mode>"): New expander.
10770 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
10771 and force the shift count operand to V16QImode.
10772 ("vec_srb<mode>"): Set shift count mode to V16QI.
10774 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10776 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
10777 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
10778 ("vec_widen_smult_hi_<mode>"): New expander definitions.
10780 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
10783 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
10785 2017-09-26 Richard Biener <rguenther@suse.de>
10787 PR tree-optimization/82320
10788 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
10791 2017-09-25 Jeff Law <law@redhat.com>
10793 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
10794 prototype for new argument.
10795 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
10796 mostly extracted from rs6000_emit_allocate_stack.
10797 (rs6000_emit_probe_stack_range_stack_clash): New function.
10798 (rs6000_emit_allocate_stack): Call
10799 rs6000_emit_probe_stack_range_stack_clash as needed.
10800 (rs6000_emit_probe_stack_range): Add additional argument
10801 to call to gen_probe_stack_range{si,di}.
10802 (output_probe_stack_range): New.
10803 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
10804 (output_probe_stack_range_stack_clash): New.
10805 (rs6000_emit_prologue): Emit notes into dump file as requested.
10806 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
10807 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
10808 Add additional operand and pass it to output_probe_stack_range.
10810 2017-09-25 Bin Cheng <bin.cheng@arm.com>
10812 PR tree-optimization/82163
10813 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
10814 (checking_verify_loop_closed_ssa): New parameter.
10815 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
10816 (check_loop_closed_ssa_stmt): Delete.
10817 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
10818 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
10819 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
10822 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
10824 * brig-builtins.def: Treat HSAIL barrier builtins as
10825 setjmp/longjump style functions.
10827 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10829 * target.def (constant_alignment): New hook.
10830 * defaults.h (CONSTANT_ALIGNMENT): Delete.
10831 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
10832 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
10833 * doc/tm.texi: Regenerate.
10834 * targhooks.h (default_constant_alignment): Declare.
10835 (constant_alignment_word_strings): Likewise.
10836 * targhooks.c (default_constant_alignment): New function.
10837 (constant_alignment_word_strings): Likewise.
10838 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
10839 instead of CONSTANT_ALIGNMENT.
10840 * varasm.c (align_variable, get_variable_align, build_constant_desc)
10841 (force_const_mem): Likewise.
10842 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
10843 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
10844 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
10845 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10846 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
10848 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
10849 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10850 constant_alignment_word_strings.
10851 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
10852 (CONSTANT_ALIGNMENT): Likewise.
10853 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10854 (arm_constant_alignment): New function.
10855 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
10856 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10857 constant_alignment_word_strings.
10858 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
10859 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10860 constant_alignment_word_strings.
10861 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
10862 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10863 (cris_constant_alignment): New function.
10864 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
10865 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10866 (epiphany_constant_alignment): New function.
10867 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
10868 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10869 constant_alignment_word_strings.
10870 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
10871 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10872 constant_alignment_word_strings.
10873 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
10874 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10875 constant_alignment_word_strings.
10876 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
10877 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
10878 * config/i386/i386.c (ix86_constant_alignment): Make static.
10879 Use the same interface as the target hook.
10880 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10881 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
10882 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10883 constant_alignment_word_strings.
10884 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
10885 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
10886 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10887 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
10888 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10889 constant_alignment_word_strings.
10890 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
10891 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10892 constant_alignment_word_strings.
10893 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
10894 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10895 constant_alignment_word_strings.
10896 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
10897 * config/microblaze/microblaze.c (microblaze_constant_alignment):
10899 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10900 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
10901 * config/mips/mips.c (mips_constant_alignment): New function.
10902 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10903 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
10904 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
10905 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10906 (mmix_constant_alignment): Make static. Use the same interface
10907 as the target hook.
10908 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
10909 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10910 constant_alignment_word_strings.
10911 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
10912 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10913 constant_alignment_word_strings.
10914 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
10915 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10916 constant_alignment_word_strings.
10917 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
10918 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10919 (rs6000_constant_alignment): New function.
10920 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
10921 * config/riscv/riscv.c (riscv_constant_alignment): New function.
10922 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10923 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
10924 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10925 (rs6000_constant_alignment): New function.
10926 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
10927 * config/s390/s390.c (s390_constant_alignment): New function.
10928 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10929 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
10930 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10931 constant_alignment_word_strings.
10932 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
10933 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10934 (sparc_constant_alignment): New function.
10935 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
10936 * config/spu/spu.c (spu_constant_alignment): New function.
10937 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10938 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
10939 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10940 constant_alignment_word_strings.
10941 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
10942 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10943 constant_alignment_word_strings.
10944 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
10945 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10946 constant_alignment_word_strings.
10947 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
10948 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10949 (visium_constant_alignment): New function.
10950 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
10951 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10952 (xtensa_constant_alignment): New function.
10953 * system.h (CONSTANT_ALIGNMENT): Poison.
10955 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
10957 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10958 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
10959 (rs6000_builtin_valid_without_lhs): New helper function.
10960 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10961 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
10963 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10965 * target.h (vec_perm_indices): Use unsigned short rather than
10967 (auto_vec_perm_indices): Likewise.
10968 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
10969 Use unsigned int rather than unsigned char.
10970 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
10972 2017-09-25 Richard Biener <rguenther@suse.de>
10974 * cfgloop.h (sort_sibling_loops): Declare.
10975 * cfgloop.c (sort_sibling_loops_cmp): New helper.
10976 (sort_sibling_loops): New function sorting the sibling loop list
10978 * graphite.c (graphite_transform_loops): Sort sibling loops.
10980 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
10982 * target.def (vec_perm_const_ok): Change sel parameter to
10984 * optabs-query.c (can_vec_perm_p): Update accordingly.
10985 * doc/tm.texi: Regenerate.
10986 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
10987 auto_vec_perm_indices and remove separate nelt field.
10988 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
10989 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
10990 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
10991 (aarch64_expand_vec_perm_const): Update accordingly.
10992 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
10993 to vec_perm_indices.
10994 * config/arm/arm.c (expand_vec_perm_d): Change perm to
10995 auto_vec_perm_indices and remove separate nelt field.
10996 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
10997 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
10998 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
11000 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
11001 to vec_perm_indices.
11002 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
11003 sel to vec_perm_indices.
11004 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
11005 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
11006 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
11008 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
11011 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
11014 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
11015 on the FUNCTION_DECL function context if it has a DIE that is a
11018 2017-09-25 Richard Biener <rguenther@suse.de>
11020 PR tree-optimization/82285
11021 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
11024 2017-09-25 Tom de Vries <tom@codesourcery.com>
11028 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
11031 2017-09-25 Richard Biener <rguenther@suse.de>
11033 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
11034 ISL errors other than isl_error_quota happen. Dump if the
11035 schedule is the same.
11036 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
11037 errors instead of aborting inside ISL.
11039 2017-09-25 Iain Sandoe <iain@codesourcery.com>
11042 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
11043 of libgcc_eh for m64.
11044 * config/i386/darwin64.h: Likewise.
11046 2017-09-25 Richard Biener <rguenther@suse.de>
11048 PR middle-end/82144
11049 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
11050 attribute for incomplete types nor twice for complete ones.
11052 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
11055 * config/i386/i386.c (ix86_print_operand_address_as): Encode
11056 %esp as %rsp to avoid 0x67 prefix if there is no index or base
11059 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
11062 * config/i386/i386.opt (mprefer-avx256): Use
11063 ix86_target_flags variable.
11064 * config/i386/i386.c (ix86_target_string): Move
11065 -mprefer-avx256 to flag2_opts.
11067 2017-09-22 Jakub Jelinek <jakub@redhat.com>
11069 PR middle-end/35691
11070 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
11071 and x != -1 | y != -1 into (x & y) != -1.
11073 2017-09-22 Steve Ellcey <sellcey@cavium.com>
11075 * config.gcc: Add new case statement to set
11076 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
11077 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
11078 s390x-*-linux* case statements. Added aarch64 to the list of
11079 supported architectures.
11081 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11083 PR tree-optimization/82289
11084 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
11085 STMT_VINFO_RELEVANT_P.
11087 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11088 Alan Hayward <alan.hayward@arm.com>
11089 David Sherwood <david.sherwood@arm.com>
11091 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
11092 for VR_RANGE only; don't allow VR_ANTI_RANGE.
11093 (extract_range_from_binary_expr_1): Don't call
11094 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
11096 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11097 Alan Hayward <alan.hayward@arm.com>
11098 David Sherwood <david.sherwood@arm.com>
11100 * target.def (preferred_vector_alignment): New hook.
11101 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
11103 * doc/tm.texi: Regenerate.
11104 * targhooks.h (default_preferred_vector_alignment): Declare.
11105 * targhooks.c (default_preferred_vector_alignment): New function.
11106 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
11108 (DR_TARGET_ALIGNMENT): New macro.
11109 (aligned_access_p): Update commentary.
11110 (vect_known_alignment_in_bytes): New function.
11111 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
11113 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
11114 Calculate the misalignment based on the target alignment rather than
11116 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
11117 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
11118 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
11119 the target alignment, rather than masking the element misalignment
11120 with the number of elements in a vector. Also use the target
11121 alignment when calculating the maximum number of peels.
11122 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
11123 instead of TYPE_ALIGN_UNIT.
11124 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
11125 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
11126 (vect_create_addr_base_for_vector_ref): Update call accordingly.
11127 (vect_create_data_ref_ptr): Likewise.
11128 (vect_setup_realignment): Realign by ANDing with
11129 -DR_TARGET_MISALIGNMENT.
11130 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
11131 the number of peels based on DR_TARGET_ALIGNMENT.
11132 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
11133 with the guaranteed alignment boundary when deciding whether
11135 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
11136 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
11137 (ensure_base_align): Remove stmt_info parameter. Get the
11138 target base alignment from DR_TARGET_ALIGNMENT.
11139 (vectorizable_store): Update call accordingly. Interpret
11140 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
11142 (vectorizable_load): Likewise.
11144 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11145 Alan Hayward <alan.hayward@arm.com>
11146 David Sherwood <david.sherwood@arm.com>
11148 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
11149 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
11150 (vect_enhance_data_refs_alignment): Likewise.
11152 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11154 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
11155 error. Only quit immediately if parsing is complete.
11156 (BEGIN): Initialize fatal_err and parse_done.
11157 (begin fpu, end fpu): Check number of arguments.
11158 (begin arch, end arch): Likewise.
11159 (begin cpu, end cpu): Likewise.
11160 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
11161 (optalias): Likewise.
11163 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11165 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
11166 * config/arm/arm-isa.h: Delete. Move definitions to ...
11167 * arm-cpus.in: ... here. Use new feature and fgroup values.
11168 * config/arm/arm.c (arm_option_override): Use lower case for feature
11170 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
11171 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
11172 * config/arm/parsecpu.awk (END): Add new command 'isa'.
11174 (print_isa_bits_for): New function.
11175 (gen_isa): New function.
11176 (gen_comm_data): Use print_isa_bits_for.
11177 (define feature): New keyword.
11178 (define fgroup): New keyword.
11179 * config/arm/t-arm (TM_H): Remove.
11180 (GTM_H): Add arm-isa.h.
11181 (arm-isa.h): Add rule to generate file.
11182 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
11183 case for feature bit names.
11185 2017-09-22 Richard Biener <rguenther@suse.de>
11187 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
11189 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
11190 print of no dependency loops ...
11191 * graphite.c (graphite_transform_loops): ... here.
11192 (canonicalize_loop_closed_ssa_form): Work from inner to outer
11194 (same_close_phi_node, remove_duplicate_close_phi,
11195 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
11196 (canonicalize_loop_closed_ssa): ... here and simplify.
11197 * graphite-optimize-isl.c: Include tree-vectorizer.h.
11198 (optimize_isl): Use dump_printf_loc to tell when we stopped
11199 optimizing because of an ISL timeout.
11201 2017-09-22 Richard Biener <rguenther@suse.de>
11203 PR tree-optimization/82291
11204 * tree-if-conv.c (predicate_mem_writes): Make sure to
11205 remove writes in blocks predicated with false.
11207 2017-09-22 Richard Biener <rguenther@suse.de>
11209 * sese.c: Include cfganal.h.
11210 (if_region_set_false_region): Remove.
11211 (create_if_region_on_edge): Likewise.
11212 (move_sese_in_condition): Re-implement without destroying
11215 2017-09-22 Richard Biener <rguenther@suse.de>
11217 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
11218 Verify both BBs contain loop PHI nodes before dispatching to
11219 copy_loop_phi_args.
11220 (graphite_regenerate_ast_isl): Do not recompute dominators,
11221 do not verify three times. Restructure for clarity.
11222 * graphite-scop-detection.c (same_close_phi_node,
11223 remove_duplicate_close_phi, make_close_phi_nodes_unique,
11224 defined_in_loop_p, canonicalize_loop_closed_ssa,
11225 canonicalize_loop_closed_ssa_form): Simplify, remove excess
11226 checking and SSA rewrite, move to ...
11227 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
11228 (graphite_initialize): Do not pass in ctx, do not reset the
11229 SCEV cache, compute only dominators.
11230 (graphite_transform_loops): Allocate ISL ctx after
11231 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
11232 Maintain post-dominators only around build_scops.
11233 * sese.c (if_region_set_false_region): Make static. Free
11234 and recompute dominators.
11235 (move_sese_in_condition): Assert we don't get called with
11236 post-dominators computed.
11237 * sese.h (if_region_set_false_region): Remove.
11239 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
11241 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
11242 mode attribute for TARGET_AVX512VL.
11244 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
11246 * config/i386/i386.opt (mprefer-avx256): New option.
11247 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
11249 (ix86_preferred_simd_mode): Return 256-bit AVX modes
11250 for TARGET_PREFER_AVX256.
11251 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
11253 2017-09-21 Jeff Law <law@redhat.com>
11255 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
11256 Fix dump output if the only stack space is for pushed registers.
11258 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11260 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
11263 2017-09-21 Martin Sebor <msebor@redhat.com>
11266 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
11267 code (in C++) or code that triggers warnings.
11269 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
11271 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
11273 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
11275 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
11276 * sched-rgn.c: Ditto.
11277 * sel-sched-ir.c: Ditto.
11279 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
11281 * toplev.h (set_random_seed): Adjust return type.
11282 * toplev.c (init_local_tick): Move eager initialization of random_seed
11283 to get_random_seed. Adjust comment.
11284 (init_random_seed): Inline to get_random_seed, delete.
11285 (get_random_seed): Initialize random_seed lazily.
11286 (set_random_seed): Do not return previous value.
11287 (print_switch_value): Do not call get_random_seed.
11289 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
11291 * cgraph.c (delete_function_version): New, broken out from...
11292 (cgraph_node::delete_function_version): ...here. Rename to
11293 cgraph_node::delete_function_version_by_decl. Update all uses.
11294 (cgraph_node::remove): Call delete_function_version.
11296 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11299 * tree-inline.c (expand_call_inline): Emit clobber stmts for
11300 VAR_DECLs to which addressable non-volatile parameters are mapped
11301 and for id->retvar after the return value assignment. Clear
11302 id->retval and id->retbnd after inlining.
11304 2017-09-21 Richard Biener <rguenther@suse.de>
11306 PR tree-optimization/82276
11307 PR tree-optimization/82244
11308 * tree-vrp.c (build_assert_expr_for): Set
11309 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
11311 (remove_range_assertions): Revert earlier change.
11313 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
11316 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
11318 2017-09-21 Richard Biener <rguenther@suse.de>
11320 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11321 Restore valid IL after code generation errors.
11322 * graphite.c (graphite_transform_loops): Diagnose code
11323 generation issues as MSG_MISSED_OPTIMIZATION and continue
11324 with processing SCOPs.
11326 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11327 Alan Hayward <alan.hayward@arm.com>
11328 David Sherwood <david.sherwood@arm.com>
11330 * calls.c (compute_argument_addresses): Use simplify_gen_binary
11331 rather than choosing between plus_constant and gen_rtx_<CODE>.
11332 * expr.c (emit_push_insn): Likewise.
11333 (expand_expr_real_2): Likewise.
11335 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11336 Alan Hayward <alan.hayward@arm.com>
11337 David Sherwood <david.sherwood@arm.com>
11339 * loop-unroll.c (split_iv): Call copy_rtx on the step.
11341 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11342 Alan Hayward <alan.hayward@arm.com>
11343 David Sherwood <david.sherwood@arm.com>
11345 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
11346 calling tree_to_uhwi.
11348 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11349 Alan Hayward <alan.hayward@arm.com>
11350 David Sherwood <david.sherwood@arm.com>
11352 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
11353 INTEGER_CST rather than a negative test for ADDR_EXPR.
11355 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11356 Alan Hayward <alan.hayward@arm.com>
11357 David Sherwood <david.sherwood@arm.com>
11359 * tree-vrp.c (extract_range_from_binary_expr_1): Check
11360 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
11362 2017-09-21 Richard Biener <rguenther@suse.de>
11364 PR tree-optimization/71351
11365 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
11366 graphite_create_new_loop_guard): Remove, fold remaining parts
11368 (translate_isl_ast_node_for): ... here and simplify.
11370 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11373 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
11374 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
11375 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
11376 alternative always use QI mode, for -Os imov (=R,R) alternative
11377 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
11380 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11381 Jeff Law <law@redhat.com>
11383 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
11384 (allocate_stack_space): New function, partially extracted from
11385 s390_emit_prologue.
11386 (s390_emit_prologue): Track offset to most recent stack probe.
11387 Code to allocate space moved into allocate_stack_space.
11388 Dump actions when no stack is allocated.
11389 (s390_prologue_plus_offset): New function.
11390 (s390_emit_stack_probe): Likewise.
11392 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
11394 * common.opt (Wa, Wl, Wp, g, gz=): Add
11396 (gno-column-info): Remove.
11397 (gcolumn-info): Drop RejectNegative.
11398 (gno-): New prefix.
11399 (gno-record-gcc-switches): Remove.
11400 (grecord-gcc-switches): Drop RejectNegative.
11401 (gno-split-dwarf): Remove.
11402 (gsplit-dwarf): Drop RejectNegative.
11403 (gno-strict-dwarf): Remove.
11404 (gstrict-dwarf): Drop RejectNegative.
11405 * config/darwin.opt (gfull, gused): Add RejectNegative.
11406 * dwarf2out.c (gen_producer_string): Drop
11407 gno-record-gcc-switches handler.
11408 * optc-gen.awk: Add g to prefixes with negative forms.
11409 * opts-common.c (remapping_prefix_p): New.
11410 (find_opt): Check it.
11411 (generate_canonical_option): Test g prefix.
11412 (option_map): Add -gno- mapping.
11413 (add_misspelling_candidates): Check remapping_prefix_p.
11415 2017-09-20 Jeff Law <law@redhat.com>
11417 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
11418 thinko in stack clash protection support.
11420 * explow.c (compute_stack_clash_protection_loop_data): Use
11421 CONST_INT_P instead of explicit test. Verify object is a
11422 CONST_INT_P before looking at INTVAL.
11423 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
11424 instead of explicit test.
11426 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
11429 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
11430 address instead of to r1 and r11.
11432 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
11434 * config.gcc: Support "knm".
11435 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
11436 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
11438 * config/i386/i386.c (m_KNM): Define.
11439 (processor_target_table): Add "knm".
11441 (ix86_option_override_internal): Add "knm".
11442 (ix86_issue_rate): Add PROCESSOR_KNM.
11443 (ix86_adjust_cost): Ditto.
11444 (ia32_multipass_dfa_lookahead): Ditto.
11445 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
11446 (fold_builtin_cpu): Add M_INTEL_KNM.
11447 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
11448 (processor_type): Add PROCESSOR_KNM.
11449 * config/i386/x86-tune.def: Add m_KNM.
11450 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
11452 2017-09-20 Richard Biener <rguenther@suse.de>
11454 PR tree-optimization/80213
11455 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
11456 are allowed in empty BBs as well.
11457 (canonicalize_loop_closed_ssa): Also look for other complex
11459 (scop_detection::get_sese): Include the loop-closed PHI block
11461 (scop_detection::merge_sese): Remove code adding extra blocks.
11462 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
11463 (build_scops): Assert the final returned scop is invalid.
11465 2017-09-20 Richard Biener <rguenther@suse.de>
11467 PR tree-optimization/82264
11468 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
11470 (vn_phi_lookup): Likewise.
11471 (vn_phi_insert): Likewise.
11473 2017-09-20 Jakub Jelinek <jakub@redhat.com>
11475 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
11476 that fits into uhwi or shwi, add DW_AT_const_value regardless
11477 of early_dwarf without going through RTL, using add_AT_unsigned
11480 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
11481 (DEBUG_LTO_ABBREV_SECTION): Likewise.
11482 (DEBUG_LTO_MACINFO_SECTION): Likewise.
11483 (DEBUG_MACRO_SECTION): Likewise.
11484 (DEBUG_LTO_MACRO_SECTION): Likewise.
11485 (DEBUG_STR_DWO_SECTION): Likewise.
11486 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
11487 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
11488 (DEBUG_LTO_DWO_LINE_SECTION): Define.
11489 (DEBUG_LTO_LINE_STR_SECTION): Define.
11490 (init_sections_and_labels): Initialize debug_line_str_section
11491 variable. Initialize debug_loc_section for -gdwarf-5 to
11492 DEBUG_LOCLISTS_SECTION. Formatting fixes.
11494 2017-09-20 Richard Biener <rguenther@suse.de>
11496 * graphite-sese-to-poly.c (extract_affine): Properly handle
11497 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
11499 2017-09-20 Richard Biener <rguenther@suse.de>
11501 PR tree-optimization/81373
11502 * graphite-scop-detection.c (build_cross_bb_scalars_def):
11503 Force SESE live-out defs to be handled even if they are
11506 2017-09-19 Jeff Law <law@redhat.com>
11508 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
11509 nothing for stack adjustments with REG_STACK_CHECK.
11510 * sched-deps.c (parse_add_or_inc): Reject insns with
11511 REG_STACK_CHECK from dependency breaking.
11512 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
11513 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
11514 * reg-notes.def (STACK_CHECK): New note.
11516 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
11517 (ix86_expand_prologue): Dump stack clash info as needed.
11518 Call ix86_adjust_stack_and_probe_stack_clash as needed.
11520 * function.c (dump_stack_clash_frame_info): New function.
11521 * function.h (dump_stack_clash_frame_info): Prototype.
11522 (enum stack_clash_probes): New enum.
11524 * config/alpha/alpha.c (alpha_expand_prologue): Also check
11525 flag_stack_clash_protection.
11526 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
11527 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
11528 (arm_frame_pointer_required): Likewise.
11529 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
11530 (ia64_expand_prologue): Likewise.
11531 * config/mips/mips.c (mips_expand_prologue): Likewise.
11532 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
11533 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11534 (sparc_flat_expand_prologue): Likewise.
11535 * config/spu/spu.c (spu_expand_prologue): Likewise.
11537 * explow.c: Include "params.h".
11538 (anti_adjust_stack_and_probe_stack_clash): New function.
11539 (get_stack_check_protect): Likewise.
11540 (compute_stack_clash_protection_loop_data): Likewise.
11541 (emit_stack_clash_protection_loop_start): Likewise.
11542 (emit_stack_clash_protection_loop_end): Likewise.
11543 (allocate_dynamic_stack_space): Use get_stack_check_protect.
11544 Use anti_adjust_stack_and_probe_stack_clash.
11545 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
11546 (emit_stack_clash_protection_loop_start): Likewise.
11547 (emit_stack_clash_protection_loop_end): Likewise.
11548 * rtl.h (get_stack_check_protect): Prototype.
11549 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
11550 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
11551 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
11553 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
11555 * doc/tm.texi: Rebuilt.
11556 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
11557 get_stack_check_protect.
11558 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
11559 * config/arm/arm.c (arm_expand_prologue): Likewise.
11560 (arm_frame_pointer_required): Likewise.
11561 * config/i386/i386.c (ix86_expand_prologue): Likewise.
11562 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
11563 * config/mips/mips.c (mips_expand_prologue): Likewise.
11564 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
11565 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
11566 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11567 (sparc_flat_expand_prologue): Likewise.
11569 * common.opt (-fstack-clash-protection): New option.
11570 * flag-types.h (enum stack_check_type): Note difference between
11571 -fstack-check= and -fstack-clash-protection.
11572 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
11573 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
11574 * toplev.c (process_options): Issue warnings/errors for cases
11575 not handled with -fstack-clash-protection.
11576 * doc/invoke.texi (-fstack-clash-protection): Document new option.
11577 (-fstack-check): Note additional problem with -fstack-check=generic.
11578 Note that -fstack-check is primarily for Ada and refer users
11579 to -fstack-clash-protection for stack-clash-protection.
11580 Document new params for stack clash protection.
11582 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
11584 * config/i386/i386.c (ix86_split_long_move): Do not handle
11585 address used for LEA in a special way.
11587 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
11589 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
11591 2017-09-19 Martin Sebor <msebor@redhat.com>
11594 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
11595 of incompatible types.
11597 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
11599 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11600 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
11601 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11602 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
11604 2017-09-19 Richard Biener <rguenther@suse.de>
11606 PR tree-optimization/82244
11607 * tree-vrp.c (remove_range_assertions): Do not propagate
11608 a constant to abnormals but replace the assert with a copy.
11610 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
11612 PR rtl-optimization/57878
11613 PR rtl-optimization/68988
11614 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
11615 avoidance test involving non_reload_pseudos. Move frequency test
11616 below the general fragmentation avoidance test.
11618 2017-09-19 Richard Biener <rguenther@suse.de>
11620 PR tree-optimization/69728
11621 * graphite-sese-to-poly.c (schedule_error): New global.
11622 (add_loop_schedule): Handle empty domain by failing the
11624 (build_original_schedule): Handle schedule_error.
11626 2017-09-19 Richard Biener <rguenther@suse.de>
11628 * graphite-scop-detection.c (scop_detection::can_represent_loop):
11629 Do not iterate to sibling loops but only to siblings of inner
11632 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
11635 * config/m68k/m68k.md (moveq feeding equality comparison): Check
11636 that the registers are different.
11638 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
11640 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
11641 to processor_model and "amdfam17h" to arch_names_table.
11642 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
11644 2017-09-18 Jakub Jelinek <jakub@redhat.com>
11647 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
11649 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11650 Alan Hayward <alan.hayward@arm.com>
11651 David Sherwood <david.sherwood@arm.com>
11653 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
11655 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
11657 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
11658 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
11659 vect_schedule_slp_instance.
11660 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
11661 Update call to vect_slp_analyze_node_operations. Simplify return
11663 (vect_slp_analyze_bb_1): Update call accordingly.
11664 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
11665 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
11666 (vect_schedule_slp): Update call accordingly.
11668 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11669 Alan Hayward <alan.hayward@arm.com>
11670 David Sherwood <david.sherwood@arm.com>
11672 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
11673 with types that aren't in fact scalar.
11675 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11677 * tree-vect-slp.c (vect_record_max_nunits): New function,
11679 (vect_build_slp_tree_1): ...here.
11680 (vect_build_slp_tree_2): Call it for phis too.
11682 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11684 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
11685 to vect_get_vec_def_for_operand when getting the mask operand.
11687 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11688 Alan Hayward <alan.hayward@arm.com>
11689 David Sherwood <david.sherwood@arm.com>
11691 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
11694 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11695 Alan Hayward <alan.hayward@arm.com>
11696 David Sherwood <david.sherwood@arm.com>
11698 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
11699 calculation for vector booleans.
11701 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11702 Alan Hayward <alan.hayward@arm.com>
11703 David Sherwood <david.sherwood@arm.com>
11705 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
11707 (vect_transform_stmt): ...here.
11708 (vect_analyze_stmt): Use it instead of calling
11709 vectorizable_live_operation directly.
11711 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
11713 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
11714 non-SIMT targets in acc vector loops.
11716 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
11718 * configure.ac: Add arc and check if assembler supports gdwarf2.
11719 * configure: Regenerate.
11721 2017-09-18 Richard Biener <rguenther@suse.de>
11723 PR tree-optimization/82220
11724 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
11725 epilogue niters from the min_profitable_iters compute.
11727 2017-09-18 Jakub Jelinek <jakub@redhat.com>
11730 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
11731 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
11732 (ix86_init_pic_reg): Revert 2017-09-01 changes.
11734 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
11737 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
11738 switching to a new text section.
11740 2017-09-18 Richard Biener <rguenther@suse.de>
11742 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
11744 (build_alias_set): Reject aliases with no access function.
11746 2017-09-18 Richard Biener <rguenther@suse.de>
11748 PR tree-optimization/79622
11749 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
11751 (build_cross_bb_scalars_use): Likewise.
11753 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
11755 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
11757 2017-09-18 Alan Modra <amodra@gmail.com>
11760 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
11761 stack_pointer_rtx for count 0. Update comments. Break up
11762 large rtl expression.
11764 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
11766 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
11767 Increase to 20 bytes.
11768 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
11769 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
11770 or avx version of the stub.
11772 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
11775 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
11776 compute the minimum stack alignment. Also update preferred stack
11777 boundary for leaf functions.
11779 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
11781 PR tree-optimization/82228
11782 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
11785 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
11787 * common/config/nds32/nds32-common.c
11788 (nds32_option_optimization_table): Refine formatting.
11789 (nds32_option_optimization_table): Use -fsched-pressure and
11790 -fomit-frame-pointer for specific optimization level.
11792 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
11794 * config/nds32/nds32.c: Refine formatting and comments.
11795 * config/nds32/nds32.h: Likewise.
11796 * config/nds32/nds32.md: Likewise.
11797 * config/nds32/nds32-cost.c: Likewise.
11798 * config/nds32/nds32-isr.c: Likewise.
11799 * config/nds32/nds32-md-auxiliary.c: Likewise.
11800 * config/nds32/nds32-multiple.md: Likewise.
11801 * config/nds32/nds32-predicates.c: Likewise.
11803 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
11804 Jakub Jelinek <jakub@redhat.com>
11806 Add support for -std=c++2a.
11807 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
11809 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
11811 2017-09-15 Steve Ellcey <sellcey@cavium.com>
11814 * doc/extend.texi (Common Function Attributes): Add
11815 references to ARM, AArch64, and S/390 specific attributes.
11816 (Function Specific Option Pragmas): Add AArch64 and S/390
11817 to list of back ends that support the target pragma.
11819 2017-09-15 Nathan Sidwell <nathan@acm.org>
11821 * doc/standards.texi: Fix C++17 description. Update URLs for
11824 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
11826 * common.opt (Wcast-align=strict): New warning option.
11827 * doc/invoke.texi: Document -Wcast-align=strict.
11829 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
11831 * cgraph.h (cgraph_thunk_info): Add comments.
11832 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
11833 assert for VIRTUAL_* arguments stricter.
11835 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
11837 PR tree-optimization/71026
11838 * match.pd: Move RDIV patterns from fold-const.c
11839 * fold-const.c (distribute_real_division): Removed.
11840 (fold_binary_loc): Remove calls to distribute_real_divison.
11842 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11844 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
11845 c++1z and gnu++1z as deprecated. Change other references to
11846 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
11847 Change -Wc++1z-compat to -Wc++17-compat.
11848 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
11849 * dwarf2out.c (highest_c_language): Handle C++17.
11850 (gen_compile_unit_die): Likewise.
11852 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11854 PR rtl-optimization/82192
11855 * combine.c (make_extraction): Don't look through non-paradoxical
11856 SUBREGs or TRUNCATE if pos + len is or might be bigger than
11859 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11860 Alan Hayward <alan.hayward@arm.com>
11861 David Sherwood <david.sherwood@arm.com>
11863 * target.def (function_arg_offset): New hook.
11864 * targhooks.h (default_function_arg_offset): Declare.
11865 * targhooks.c (default_function_arg_offset): New function.
11866 * function.c (locate_and_pad_parm): Use
11867 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
11868 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
11869 (TARGET_FUNCTION_ARG_OFFSET): ...this.
11870 * doc/tm.texi: Regenerate.
11871 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
11872 * config/spu/spu.c (spu_function_arg_offset): New function.
11873 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
11874 * system.h (FUNCTION_ARG_OFFSET): Poison.
11876 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11877 Alan Hayard <alan.hayward@arm.com>
11878 David Sherwood <david.sherwood@arm.com>
11880 * target.def (truly_noop_truncation): New hook.
11881 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
11882 than TRULY_NOOP_TRUNCATION.
11883 * hooks.h (hook_bool_uint_uint_true): Declare.
11884 * hooks.c (hook_bool_uint_uint_true): New function.
11885 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
11886 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
11887 * doc/tm.texi: Regenerate.
11888 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
11889 rather than TRULY_NOOP_TRUNCATION in comments.
11890 (simplify_comparison): Likewise.
11891 (record_truncated_value): Likewise.
11892 * expmed.c (extract_bit_field_1): Likewise.
11893 (extract_split_bit_field): Likewise.
11894 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
11895 instead of TRULY_NOOP_TRUNCATION.
11896 * function.c (assign_parm_setup_block): Likewise.
11897 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
11898 * rtlhooks.c: Include target.h.
11899 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
11900 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
11901 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
11902 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
11903 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
11904 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
11905 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
11906 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
11907 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
11908 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
11909 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
11910 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
11911 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
11912 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
11913 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
11914 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
11915 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
11916 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
11917 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
11918 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
11919 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
11920 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
11921 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
11922 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
11923 * config/mips/mips.c (mips_truly_noop_truncation): New function.
11924 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11925 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11926 rather than TRULY_NOOP_TRUNCATION in comments.
11927 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
11928 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
11929 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
11930 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
11931 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
11932 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
11933 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
11934 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
11935 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
11936 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
11937 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
11938 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11939 rather than TRULY_NOOP_TRUNCATION in comments.
11940 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
11941 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
11942 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
11943 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
11944 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
11945 TRULY_NOOP_TRUNCATION condition.
11946 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
11947 (TRULY_NOOP_TRUNCATION): Delete.
11948 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
11949 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
11950 * config/spu/spu.c (spu_truly_noop_truncation): New function.
11951 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11952 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
11953 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
11954 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
11955 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11956 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11957 rather than TRULY_NOOP_TRUNCATION in comments.
11958 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
11959 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
11960 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
11961 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
11962 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
11963 * system.h (TRULY_NOOP_TRUNCATION): Poison.
11965 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
11968 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
11969 (*cmp_ior): Likewise.
11970 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
11971 (*ior_scc_scc_cmp): Likewise.
11972 (*and_scc_scc): Likewise.
11973 (*and_scc_scc_cmp): Likewise.
11975 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11976 Alan Hayard <alan.hayward@arm.com>
11977 David Sherwood <david.sherwood@arm.com>
11979 * target.def (can_change_mode_class): New hook.
11980 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
11981 (hard_regno_nregs): Likewise.
11982 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
11983 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
11984 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
11985 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
11986 (LOAD_EXTEND_OP): Update accordingly.
11987 * doc/tm.texi: Regenerate.
11988 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
11989 CANNOT_CHANGE_MODE_CLASS.
11990 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
11991 (REG_CAN_CHANGE_MODE_P): ...this new macro.
11992 * combine.c (simplify_set): Update accordingly.
11993 * emit-rtl.c (validate_subreg): Likewise.
11994 * recog.c (general_operand): Likewise.
11995 * regcprop.c (mode_change_ok): Likewise.
11996 * reload1.c (choose_reload_regs): Likewise.
11997 (inherit_piecemeal_p): Likewise.
11998 * rtlanal.c (simplify_subreg_regno): Likewise.
11999 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
12000 instead of CANNOT_CHANGE_MODE_CLASS.
12001 (reload_cse_simplify_operands): Likewise.
12002 * reload.c (push_reload): Use targetm.can_change_mode_class
12003 instead of CANNOT_CHANGE_MODE_CLASS.
12004 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
12005 REG_CANNOT_CHANGE_MODE_P.
12006 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12007 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
12008 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12009 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12010 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12011 (arm_can_change_mode_class): New function.
12012 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
12013 than CANNOT_CHANGE_MODE_CLASS in comments.
12014 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12015 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
12016 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
12017 (ix86_can_change_mode_class): ...this new function, inverting the
12018 sense of the return value.
12019 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12020 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12021 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12022 (ia64_can_change_mode_class): New function.
12023 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12024 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
12025 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
12026 (m32c_can_change_mode_class): ...this new function, inverting the
12027 sense of the return value.
12028 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12029 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12030 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
12031 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
12032 (mips_can_change_mode_class): ...this new function, inverting the
12033 sense of the return value.
12034 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12035 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12036 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12037 (msp430_can_change_mode_class): New function.
12038 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12039 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
12040 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12041 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12042 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12043 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
12044 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12045 (pa_cannot_change_mode_class): Replace with...
12046 (pa_can_change_mode_class): ...this new function, inverting the
12047 sense of the return value.
12048 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
12049 than CANNOT_CHANGE_MODE_CLASS in comments.
12050 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12051 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
12052 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12053 (pdp11_cannot_change_mode_class): Replace with...
12054 (pdp11_can_change_mode_class): ...this new function, inverting the
12055 sense of the return value.
12056 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12057 * config/powerpcspe/powerpcspe-protos.h
12058 (rs6000_cannot_change_mode_class_ptr): Delete.
12059 * config/powerpcspe/powerpcspe.c
12060 (rs6000_cannot_change_mode_class_ptr): Delete.
12061 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12062 (rs6000_option_override_internal): Assign to
12063 targetm.can_change_mode_class instead of
12064 rs6000_cannot_change_mode_class_ptr.
12065 (rs6000_cannot_change_mode_class): Replace with...
12066 (rs6000_can_change_mode_class): ...this new function, inverting the
12067 sense of the return value.
12068 (rs6000_debug_cannot_change_mode_class): Replace with...
12069 (rs6000_debug_can_change_mode_class): ...this new function.
12070 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12071 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
12072 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12073 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12074 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
12076 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
12077 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12078 (rs6000_option_override_internal): Assign to
12079 targetm.can_change_mode_class instead of
12080 rs6000_cannot_change_mode_class_ptr.
12081 (rs6000_cannot_change_mode_class): Replace with...
12082 (rs6000_can_change_mode_class): ...this new function, inverting the
12083 sense of the return value.
12084 (rs6000_debug_cannot_change_mode_class): Replace with...
12085 (rs6000_debug_can_change_mode_class): ...this new function.
12086 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12087 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
12088 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
12089 (s390_can_change_mode_class): ...this new function, inverting the
12090 sense of the return value.
12091 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12092 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12093 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
12094 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12095 (sh_cannot_change_mode_class): Replace with...
12096 (sh_can_change_mode_class): ...this new function, inverting the
12097 sense of the return value.
12098 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12099 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12100 (sparc_can_change_mode_class): New function.
12101 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12102 * config/spu/spu.c (spu_can_change_mode_class): New function.
12103 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12104 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12105 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12106 (visium_can_change_mode_class): New function.
12107 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
12109 2017-09-15 Richard Biener <rguenther@suse.de>
12111 PR tree-optimization/82217
12112 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
12113 but not undefined case.
12115 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12118 * postreload.c (reload_cse_simplify_operands): Skip
12119 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
12121 2017-09-15 Richard Biener <rguenther@suse.de>
12123 PR tree-optimization/68823
12124 * graphite-scop-detection.c (build_alias_set): If we have a
12125 possible dependence check whether we can handle them by just
12126 looking at the DRs DR_ACCESS_FNs.
12127 (build_scops): If build_alias_set fails, fail the SCOP.
12129 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
12131 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
12132 to support float128 built-in functions that require the ISA 3.0
12134 (BU_FLOAT128_3_HW): Likewise.
12135 (SQRTF128): Add support for the IEEE 128-bit square root and fma
12136 built-in functions.
12137 (FMAF128): Likewise.
12139 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
12140 support for built-in functions that need the ISA 3.0 IEEE 128-bit
12141 floating point instructions.
12142 (rs6000_invalid_builtin): Likewise.
12143 (rs6000_builtin_mask_names): Likewise.
12144 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
12145 (RS6000_BTM_FLOAT128_HW): Likewise.
12146 (RS6000_BTM_COMMON): Likewise.
12147 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
12149 * doc/extend.texi (RS/6000 built-in functions): Document the
12150 IEEE 128-bit floating point square root and fused multiply-add
12151 built-in functions.
12153 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
12155 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
12156 reg (r2) isn't in the set of registers defined in the prologue.
12158 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12159 Alan Hayward <alan.hayward@arm.com>
12160 David Sherwood <david.sherwood@arm.com>
12162 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
12163 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
12164 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
12165 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
12166 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
12168 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
12169 max_vectorization_factor.
12170 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
12172 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12173 Alan Hayward <alan.hayward@arm.com>
12174 David Sherwood <david.sherwood@arm.com>
12176 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
12177 (vect_worthwhile_without_simd_p): Declare.
12178 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
12179 (vectorizable_reduction): Use it.
12180 * tree-vect-stmts.c (vectorizable_shift): Likewise.
12181 (vectorizable_operation): Likewise.
12183 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12184 Alan Hayward <alan.hayward@arm.com>
12185 David Sherwood <david.sherwood@arm.com>
12187 * tree-vectorizer.h (vect_get_num_copies): New function.
12188 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
12189 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12190 (vectorizable_induction): Likewise.
12191 (vectorizable_live_operation): Likewise.
12192 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12193 (vectorizable_bswap): Likewise.
12194 (vectorizable_call): Likewise.
12195 (vectorizable_conversion): Likewise.
12196 (vectorizable_assignment): Likewise.
12197 (vectorizable_shift): Likewise.
12198 (vectorizable_operation): Likewise.
12199 (vectorizable_store): Likewise.
12200 (vectorizable_load): Likewise.
12201 (vectorizable_condition): Likewise.
12202 (vectorizable_comparison): Likewise.
12203 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
12205 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12206 Alan Hayward <alan.hayward@arm.com>
12207 David Sherwood <david.sherwood@arm.com>
12209 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
12210 of vect_init_vector.
12212 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12213 Alan Hayward <alan.hayward@arm.com>
12214 David Sherwood <david.sherwood@arm.com>
12216 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
12217 an inline wrapper that provides a location.
12218 (gimple_build_vector): Likewise.
12219 * gimple-fold.c (gimple_build_vector_from_val): New function.
12220 (gimple_build_vector): Likewise.
12221 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
12222 functions to build the initial value. Always return a gimple value.
12223 (get_initial_defs_for_reduction): Likewise. Only compute
12225 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
12226 vect_init_vector on the results from get_initial_def(s)_for_reduction.
12227 (vectorizable_induction): Use gimple_build_vector rather than
12230 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12231 Alan Hayward <alan.hayward@arm.com>
12232 David Sherwood <david.sherwood@arm.com>
12234 * target.h (vec_perm_indices): New typedef.
12235 (auto_vec_perm_indices): Likewise.
12236 * optabs-query.h: Include target.h
12237 (can_vec_perm_p): Take a vec_perm_indices *.
12238 * optabs-query.c (can_vec_perm_p): Likewise.
12239 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
12240 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12241 * tree-vect-generic.c (lower_vec_perm): Likewise.
12242 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
12243 (vect_grouped_load_supported): Likewise.
12244 (vect_shift_permute_load_chain): Likewise.
12245 (vect_permute_store_chain): Use auto_vec_perm_indices.
12246 (vect_permute_load_chain): Likewise.
12247 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
12248 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
12249 Update uses of can_vec_perm_p.
12250 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
12251 mode with a number of elements. Take a vec_perm_indices *.
12252 (vect_create_epilog_for_reduction): Update accordingly.
12253 Use auto_vec_perm_indices.
12254 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
12255 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
12256 (vect_transform_slp_perm_load): Likewise.
12257 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
12258 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
12259 (vect_gen_perm_mask_checked): Likewise.
12260 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
12261 (vect_gen_perm_mask_checked): Likewise.
12262 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
12263 (vectorizable_store): Likewise.
12264 (vectorizable_load): Likewise.
12265 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
12266 (vectorizable_bswap): Likewise.
12268 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12269 Alan Hayward <alan.hayward@arm.com>
12270 David Sherwood <david.sherwood@arm.com>
12272 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
12273 * tree.c (build_vector): Likewise.
12274 (build_vector_from_ctor): Update accordingly.
12275 (build_vector_from_val): Likewise.
12276 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
12277 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12278 * tree-vect-generic.c (add_rshift): Likewise.
12279 (expand_vector_divmod): Likewise.
12280 (optimize_vector_constructor): Likewise.
12281 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
12282 (vect_transform_slp_perm_load): Likewise.
12283 (vect_schedule_slp_instance): Likewise.
12284 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
12285 (vectorizable_call): Likewise.
12286 (vect_gen_perm_mask_any): Likewise. Add elements in order.
12287 * expmed.c (make_tree): Likewise.
12288 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
12289 a vector passed to build_vector.
12290 (fold_convert_const): Likewise.
12291 (exact_inverse): Likewise.
12292 (fold_ternary_loc): Likewise.
12293 (fold_relational_const): Likewise.
12294 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
12295 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
12296 (const_unop): Likewise. Store the reduction accumulator in a
12297 variable rather than an array.
12298 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
12299 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
12300 the new vector, rather than constructing it after the input arrays.
12301 (native_interpret_vector): Use auto_vec<tree> when building
12302 a vector passed to build_vector. Add elements in order.
12303 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
12304 auto_vec<tree> when building a vector passed to build_vector.
12305 (vect_create_epilog_for_reduction): Likewise.
12306 (vectorizable_induction): Likewise.
12307 (get_initial_def_for_reduction): Likewise. Fix indentation of
12309 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
12311 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
12312 passed to build_vector.
12314 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12315 Alan Hayward <alan.hayward@arm.com>
12316 David Sherwood <david.sherwood@arm.com>
12318 * tree-core.h (tree_base::u): Add an "nelts" field.
12319 (tree_vector): Use VECTOR_CST_NELTS as the length.
12320 * tree.c (tree_size): Likewise.
12321 (make_vector): Initialize VECTOR_CST_NELTS.
12322 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
12323 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
12324 TYPE_VECTOR_SUBPARTS.
12325 * expr.c (const_vector_mask_from_tree): Consistently use "units"
12326 as the number of units, setting it from VECTOR_CST_NELTS.
12327 (const_vector_from_tree): Likewise.
12328 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
12329 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
12330 (fold_negate_expr_1): Likewise.
12331 (fold_convert_const): Likewise.
12332 (const_binop): Likewise. Differentiate the number of output and
12334 (const_unop): Likewise.
12335 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
12336 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
12337 in cases that did the opposite.
12339 2017-09-14 Richard Biener <rguenther@suse.de>
12341 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
12344 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
12346 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
12348 2017-09-14 Jakub Jelinek <jakub@redhat.com>
12351 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
12352 if and where to split a bb, except for splitting before debug insn
12353 sequences followed by non-label real insn. Delete debug insns
12354 in between basic blocks.
12356 * combine.c (make_compound_operation_int): Formatting fixes.
12358 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
12359 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12360 * config/netbsd.h (LINK_EH_SPEC): Likewise.
12361 * config/sol2.h (LINK_EH_SPEC): Likewise.
12362 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12363 * config/s390/linux.h (LINK_SPEC): Likewise.
12364 * config/freebsd.h (LINK_EH_SPEC): Likewise.
12365 * config/openbsd.h (LINK_EH_SPEC): Likewise.
12366 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12367 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
12368 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
12369 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12370 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
12371 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
12373 2017-09-13 Jakub Jelinek <jakub@redhat.com>
12375 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
12377 (ENDFILE_LINUX_SPEC): Likewise.
12378 (LINK_EH_SPEC): Likewise.
12379 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
12380 (LINK_OS_LINUX_SPEC32): Likewise.
12381 (LINK_OS_LINUX_SPEC64): Likewise.
12382 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
12383 (LINK_OS_LINUX_SPEC): Likewise.
12385 2017-09-13 Martin Liska <mliska@suse.cz>
12387 PR middle-end/82154
12388 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
12389 CASE_HIGH is NULL_TREE.
12391 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12392 Alan Hayward <alan.hayward@arm.com>
12393 David Sherwood <david.sherwood@arm.com>
12395 * target.def (secondary_memory_needed): New hook.
12396 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
12397 instead of SECONDARY_MEMORY_NEEDED.
12398 (secondary_memory_needed_mode): Likewise.
12399 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
12400 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
12401 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
12402 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
12403 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
12404 * doc/tm.texi: Regenerate.
12405 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
12406 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
12407 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12408 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
12409 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
12410 * config/i386/i386.c (inline_secondary_memory_needed): Put the
12411 mode argument first and change the reg_class arguments to reg_class_t.
12412 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
12413 Make static. Update the call to inline_secondary_memory_needed.
12414 (ix86_register_move_cost): Update the call to
12415 inline_secondary_memory_needed.
12416 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12417 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
12419 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
12420 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12422 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
12423 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
12424 * config/mips/mips.c (mips_secondary_memory_needed): Make static
12425 and match hook interface. Add comment from mips.h.
12426 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12427 * config/mmix/mmix.md (truncdfsf2): Refer to
12428 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12430 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
12431 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
12432 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12433 (pa_secondary_memory_needed): New function.
12434 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
12435 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
12436 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12437 (pdp11_secondary_memory_needed): Make static and match hook interface.
12438 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
12439 * config/powerpcspe/powerpcspe-protos.h
12440 (rs6000_secondary_memory_needed_ptr): Delete.
12441 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
12443 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12444 (rs6000_option_override_internal): Assign to
12445 targetm.secondary_memory_needed rather than
12446 rs6000_secondary_memory_needed_ptr.
12447 (rs6000_secondary_memory_needed): Match hook interface.
12448 (rs6000_debug_secondary_memory_needed): Likewise.
12449 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
12450 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
12451 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
12452 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12453 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
12454 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
12456 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
12457 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12458 (rs6000_option_override_internal): Assign to
12459 targetm.secondary_memory_needed rather than
12460 rs6000_secondary_memory_needed_ptr.
12461 (rs6000_secondary_memory_needed): Match hook interface.
12462 (rs6000_debug_secondary_memory_needed): Likewise.
12463 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
12464 * config/s390/s390.c (s390_secondary_memory_needed): New function.
12465 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12466 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
12467 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12468 (sparc_secondary_memory_needed): New function.
12469 * lra-constraints.c (check_and_process_move): Refer to
12470 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12472 (curr_insn_transform): Likewise.
12473 (process_alt_operands): Use targetm.secondary_memory_needed
12474 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12475 (check_secondary_memory_needed_p): Likewise.
12476 (choose_split_class): Likewise.
12477 * reload.c: Unconditionally include code that was previously
12478 conditional on SECONDARY_MEMORY_NEEDED.
12479 (push_secondary_reload): Use targetm.secondary_memory_needed
12480 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12481 (push_reload): Likewise.
12482 * reload1.c: Unconditionally include code that was previously
12483 conditional on SECONDARY_MEMORY_NEEDED.
12484 (choose_reload_regs): Use targetm.secondary_memory_needed
12485 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12486 (gen_reload): Likewise.
12487 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
12489 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12490 Alan Hayward <alan.hayward@arm.com>
12491 David Sherwood <david.sherwood@arm.com>
12493 * target.def (secondary_memory_needed_mode): New hook:
12494 * targhooks.c (default_secondary_memory_needed_mode): Declare.
12495 * targhooks.h (default_secondary_memory_needed_mode): New function.
12496 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
12497 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
12498 * doc/tm.texi: Regenerate.
12499 * lra-constraints.c (check_and_process_move): Use
12500 targetm.secondary_memory_needed_mode instead of
12501 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
12502 (curr_insn_transform): Likewise.
12503 * reload.c (get_secondary_mem): Likewise.
12504 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12505 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
12507 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12508 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12509 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
12510 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12511 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
12513 * config/powerpcspe/powerpcspe-protos.h
12514 (rs6000_secondary_memory_needed_mode): Delete.
12515 * config/powerpcspe/powerpcspe.c
12516 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12517 (rs6000_secondary_memory_needed_mode): Make static.
12518 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12519 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
12521 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12523 (rs6000_secondary_memory_needed_mode): Make static.
12524 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12525 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
12526 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12527 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12528 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12530 (sparc_secondary_memory_needed_mode): New function.
12531 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
12533 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
12535 * config/aarch64/constraints.md (Umq): New constraint.
12536 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
12538 (mov<mode>): Update condition.
12540 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12542 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
12543 when bitposition is the same.
12545 2017-09-13 Richard Biener <rguenther@suse.de>
12547 * dwarf2out.c (output_die_symbol): Remove.
12548 (output_die): Do not output a DIEs symbol.
12550 2017-09-13 Richard Biener <rguenther@suse.de>
12552 PR middle-end/82128
12553 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
12554 default-def to avoid breaking iterator update with the weird
12555 interaction with cgraph_update_edges_for_call_stmt_node.
12557 2017-09-13 Richard Biener <rguenther@suse.de>
12559 * tree-cfg.c (verify_gimple_assign_binary): Add verification
12560 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
12561 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
12562 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
12564 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
12566 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
12567 Disable pc relative literal load irrespective of
12568 TARGET_FIX_ERR_A53_84341 for default.
12570 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
12572 * config/sparc/sparc.c (output_return): Output the source location of
12573 the insn in the delay slot, if any.
12574 (output_sibcall): Likewise.
12576 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
12579 * common.opt (-static-pie): New alias.
12580 (shared): Negate static-pie.
12581 (-no-pie): Update help text.
12583 (static-pie): New option.
12584 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
12585 -static-pie support.
12586 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
12587 (LINK_EH_SPEC): Likewise.
12588 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12589 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
12590 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
12591 * gcc.c (LINK_COMMAND_SPEC): Likewise.
12592 (init_gcc_specs): Likewise.
12593 (init_spec): Likewise.
12594 (display_help): Update help message for -pie.
12595 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
12598 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
12600 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
12601 (movdi_aarch64): Likewise.
12602 (movti_aarch64): Likewise.
12604 2017-09-12 Simon Wright <simon@pushface.org>
12607 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
12608 calculation of the minor version, always output as 0.
12610 2017-09-12 Jakub Jelinek <jakub@redhat.com>
12613 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
12614 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
12615 on it early, rather than manual conversion late. For
12616 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
12617 instead of performing manual conversion.
12619 2017-09-12 Carl Love <cel@us.ibm.com>
12621 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
12622 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
12624 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12625 VMULOSW): Add definitions.
12626 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12627 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12628 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12629 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12630 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12632 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
12634 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
12636 (movti_aarch64): Likewise.
12637 (movdf_aarch64): Likewise.
12638 (movtf_aarch64): Likewise.
12639 (load_pairdi): Likewise.
12640 (store_pairdi): Likewise.
12641 (load_pairdf): Likewise.
12642 (store_pairdf): Likewise.
12643 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
12644 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
12645 (ldr_got_small_<mode>): Likewise.
12646 (ldr_got_small_28k_<mode>): Likewise.
12647 (ldr_got_tiny): Likewise.
12648 * config/aarch64/iterators.md (ldst_sz): New.
12649 (ldpstp_sz): Likewise.
12650 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
12652 (thunderx_load): Split load_8 to load_16.
12653 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
12655 (thunderx2t99_storepair_basic): Split store_8 to store_16.
12656 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
12657 (xgene1_store_pair): Split store_8 to store_16.
12658 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
12659 (falkor_st_0_st_sd): Split store_8 to store_16.
12661 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
12663 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
12664 and store1/2/3/4 to store_4/8/12/16.
12665 * config/aarch64/aarch64.md: Update for rename.
12666 * config/arm/arm.md: Likewise.: Likewise.
12667 * config/arm/arm.c: Likewise.
12668 * config/arm/thumb1.md: Likewise.
12669 * config/arm/thumb2.md: Likewise.
12670 * config/arm/vfp.md: Likewise.
12671 * config/arm/arm-generic.md: Likewise.
12672 * config/arm/arm1020e.md: Likewise.
12673 * config/arm/arm1026ejs.md: Likewise.
12674 * config/arm/arm1136jfs.md: Likewise.
12675 * config/arm/arm926ejs.md: Likewise.
12676 * config/arm/cortex-a15.md: Likewise.
12677 * config/arm/cortex-a17.md: Likewise.
12678 * config/arm/cortex-a5.md: Likewise.
12679 * config/arm/cortex-a53.md: Likewise.
12680 * config/arm/cortex-a57.md: Likewise.
12681 * config/arm/cortex-a7.md: Likewise.
12682 * config/arm/cortex-a8.md: Likewise.
12683 * config/arm/cortex-a9.md: Likewise.
12684 * config/arm/cortex-m4.md: Likewise.
12685 * config/arm/cortex-m7.md: Likewise.
12686 * config/arm/cortex-r4.md: Likewise.
12687 * config/arm/exynos-m1.md: Likewise.
12688 * config/arm/fa526.md: Likewise.
12689 * config/arm/fa606te.md: Likewise.
12690 * config/arm/fa626te.md: Likewise.
12691 * config/arm/fa726te.md: Likewise.
12692 * config/arm/fmp626.md: Likewise.
12693 * config/arm/iwmmxt.md: Likewise.
12694 * config/arm/ldmstm.md: Likewise.
12695 * config/arm/marvell-pj4.md: Likewise.
12696 * config/arm/xgene1.md: Likewise.
12697 * config/aarch64/thunderx.md: Likewise.
12698 * config/aarch64/thunderx2t99.md: Likewise.
12699 * config/aarch64/falkor.md: Likewise.
12701 2017-09-12 Martin Liska <mliska@suse.cz>
12703 * attribs.c (private_lookup_attribute): New function.
12704 * attribs.h (private_lookup_attribute): Declared here.
12705 (lookup_attribute): Called from this place.
12707 2017-09-12 Richard Biener <rguenther@suse.de>
12709 PR tree-optimization/82157
12710 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
12711 stmts with side-effects.
12713 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12714 Alan Hayward <alan.hayward@arm.com>
12715 David Sherwood <david.sherwood@arm.com>
12717 * target.def (hard_regno_nregs): New hook.
12718 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
12719 * targhooks.h (default_hard_regno_nregs): Declare.
12720 * targhooks.c (default_hard_regno_nregs): New function.
12721 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
12722 (TARGET_HARD_REGNO_NREGS): ...this hook.
12723 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
12724 (CLASS_MAX_NREGS): Likewise.
12725 * doc/tm.texi: Regenerate.
12726 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
12727 instead of HARD_REGNO_NREGS.
12728 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
12729 HARD_REGNO_NREGS in the comment.
12730 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
12731 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
12732 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
12733 Return an unsigned int.
12734 (TARGET_HARD_REGNO_NREGS): Redefine.
12735 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
12736 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
12737 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
12738 (arc_hard_regno_nregs): New function.
12739 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
12740 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
12741 (arm_hard_regno_nregs): New function.
12742 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
12743 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
12744 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
12745 (TARGET_HARD_REGNO_NREGS): Redefine.
12746 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
12747 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
12748 (HARD_REGNO_NREGS): Delete.
12749 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
12750 (cr16_hard_regno_nregs): New function.
12751 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
12752 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
12753 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
12754 (cris_hard_regno_nregs): New function.
12755 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
12756 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
12757 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
12758 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
12759 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
12760 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
12761 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
12762 (frv_hard_regno_nregs): Make static. Take and return an
12764 (frv_class_max_nregs): Remove outdated copy of documentation.
12765 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
12766 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
12767 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
12768 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
12769 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
12770 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
12771 (TARGET_HARD_REGNO_NREGS): Redefine.
12772 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
12773 (CLASS_MAX_NREGS): Update comment.
12774 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
12775 (ia64_hard_regno_nregs): New function.
12776 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
12777 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
12778 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
12779 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
12780 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
12782 (m32c_hard_regno_nregs): Likewise. Make static.
12783 (TARGET_HARD_REGNO_NREGS): Redefine.
12784 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
12785 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
12786 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
12787 (m68k_hard_regno_nregs): New function.
12788 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
12789 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
12790 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
12791 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
12792 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
12793 Take and return an unsigned int.
12794 (TARGET_HARD_REGNO_NREGS): Redefine.
12795 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
12796 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
12797 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
12798 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
12799 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
12800 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
12801 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
12802 (msp430_hard_regno_nregs): Make static. Take and return an
12804 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
12805 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
12806 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
12807 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
12808 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
12809 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
12810 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
12811 (TARGET_HARD_REGNO_NREGS): Redefine.
12812 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
12813 (PA_HARD_REGNO_NREGS): ...this.
12814 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
12815 (PA_HARD_REGNO_NREGS): ...this.
12816 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
12817 (pa_hard_regno_nregs): New function.
12818 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
12819 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
12820 (pdp11_hard_regno_nregs): New function.
12821 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
12822 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
12823 (rs6000_hard_regno_nregs_hook): New function.
12824 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
12825 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
12826 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
12827 Take and return an unsigned int. Move earlier in file.
12828 (TARGET_HARD_REGNO_NREGS): Redefine.
12829 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
12830 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
12831 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
12832 (rl78_hard_regno_nregs): Make static. Take and return an
12834 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
12835 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
12836 (rs6000_hard_regno_nregs_hook): New function.
12837 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
12838 * config/rx/rx.c (rx_hard_regno_nregs): New function.
12839 (TARGET_HARD_REGNO_NREGS): Redefine.
12840 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
12841 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
12842 instead of HARD_REGNO_NREGS.
12843 (s390_hard_regno_nregs): New function.
12844 (s390_hard_regno_mode_ok): Add comment from s390.h.
12845 (TARGET_HARD_REGNO_NREGS): Redefine.
12846 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
12847 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
12848 (sh_hard_regno_nregs): New function.
12849 (sh_pass_in_reg_p): Use it.
12850 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
12851 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
12852 (sparc_hard_regno_nregs): New function.
12853 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
12854 * config/spu/spu.c (spu_hard_regno_nregs): New function.
12855 (spu_function_arg_advance): Use it, supplying a valid register number.
12856 (TARGET_HARD_REGNO_NREGS): Redefine.
12857 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
12858 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
12859 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
12860 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
12861 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
12862 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
12863 (CLASS_MAX_NREGS): Remove copy of old documentation.
12864 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
12865 (visium_hard_regno_nregs): New function.
12866 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
12867 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
12868 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
12869 xtensa_hard_regno_nregs): New function.
12870 * system.h (HARD_REGNO_NREGS): Poison.
12872 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12874 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
12875 hard_regno_nregs instead of HARD_REGNO_NREGS.
12876 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
12877 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
12878 (c6x_expand_epilogue): Likewise.
12879 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
12880 (frv_read_iacc_argument): Likewise.
12881 * config/sh/sh.c: Include regs.h.
12882 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12883 (regs_used): Likewise.
12884 (output_stack_adjust): Likewise.
12885 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
12886 * expmed.c: Include regs.h.
12887 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12888 * ree.c: Include regs.h.
12889 (combine_reaching_defs): Use hard_regno_nregs instead of
12891 (add_removable_extension): Likewise.
12893 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12895 * regs.h (hard_regno_nregs): Turn into a function.
12896 (end_hard_regno): Update accordingly.
12897 * caller-save.c (setup_save_areas): Likewise.
12898 (save_call_clobbered_regs): Likewise.
12899 (replace_reg_with_saved_mem): Likewise.
12900 (insert_restore): Likewise.
12901 (insert_save): Likewise.
12902 * combine.c (can_change_dest_mode): Likewise.
12903 (move_deaths): Likewise.
12904 (distribute_notes): Likewise.
12905 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
12906 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
12907 (rs6000_split_multireg_move): Likewise.
12908 (rs6000_register_move_cost): Likewise.
12909 (rs6000_memory_move_cost): Likewise.
12910 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
12911 (rs6000_split_multireg_move): Likewise.
12912 (rs6000_register_move_cost): Likewise.
12913 (rs6000_memory_move_cost): Likewise.
12914 * cselib.c (cselib_reset_table): Likewise.
12915 (cselib_lookup_1): Likewise.
12916 * emit-rtl.c (set_mode_and_regno): Likewise.
12917 * function.c (aggregate_value_p): Likewise.
12918 * ira-color.c (setup_profitable_hard_regs): Likewise.
12919 (check_hard_reg_p): Likewise.
12920 (calculate_saved_nregs): Likewise.
12921 (assign_hard_reg): Likewise.
12922 (improve_allocation): Likewise.
12923 (calculate_spill_cost): Likewise.
12924 * ira-emit.c (modify_move_list): Likewise.
12925 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
12926 (ira_hard_reg_in_set_p): Likewise.
12927 * ira.c (setup_reg_mode_hard_regset): Likewise.
12928 (clarify_prohibited_class_mode_regs): Likewise.
12929 (check_allocation): Likewise.
12930 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12931 (lra_setup_reg_renumber): Likewise.
12932 (setup_try_hard_regno_pseudos): Likewise.
12933 (spill_for): Likewise.
12934 (assign_hard_regno): Likewise.
12935 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
12936 * lra-constraints.c (in_class_p): Likewise.
12937 (lra_constraint_offset): Likewise.
12938 (simplify_operand_subreg): Likewise.
12939 (lra_constraints): Likewise.
12940 (split_reg): Likewise.
12941 (split_if_necessary): Likewise.
12942 (invariant_p): Likewise.
12943 (inherit_in_ebb): Likewise.
12944 * lra-lives.c (process_bb_lives): Likewise.
12945 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
12946 (get_hard_regs): Likewise.
12947 (do_remat): Likewise.
12948 * lra-spills.c (assign_spill_hard_regs): Likewise.
12949 * mode-switching.c (create_pre_exit): Likewise.
12950 * postreload.c (reload_combine_recognize_pattern): Likewise.
12951 * recog.c (peep2_find_free_register): Likewise.
12952 * regcprop.c (kill_value_regno): Likewise.
12953 (set_value_regno): Likewise.
12954 (copy_value): Likewise.
12955 (maybe_mode_change): Likewise.
12956 (find_oldest_value_reg): Likewise.
12957 (copyprop_hardreg_forward_1): Likewise.
12958 * regrename.c (check_new_reg_p): Likewise.
12959 (regrename_do_replace): Likewise.
12960 * reload.c (push_reload): Likewise.
12961 (combine_reloads): Likewise.
12962 (find_dummy_reload): Likewise.
12963 (operands_match_p): Likewise.
12964 (find_reloads): Likewise.
12965 (find_equiv_reg): Likewise.
12966 (reload_adjust_reg_for_mode): Likewise.
12967 * reload1.c (count_pseudo): Likewise.
12968 (count_spilled_pseudo): Likewise.
12969 (find_reg): Likewise.
12970 (clear_reload_reg_in_use): Likewise.
12971 (free_for_value_p): Likewise.
12972 (allocate_reload_reg): Likewise.
12973 (choose_reload_regs): Likewise.
12974 (reload_adjust_reg_for_temp): Likewise.
12975 (emit_reload_insns): Likewise.
12976 (delete_output_reload): Likewise.
12977 * rtlanal.c (subreg_get_info): Likewise.
12978 * sched-deps.c (sched_analyze_reg): Likewise.
12979 * sel-sched.c (init_regs_for_mode): Likewise.
12980 (mark_unavailable_hard_regs): Likewise.
12981 (choose_best_reg_1): Likewise.
12982 (verify_target_availability): Likewise.
12983 * valtrack.c (dead_debug_insert_temp): Likewise.
12984 * var-tracking.c (track_loc_p): Likewise.
12985 (emit_note_insn_var_location): Likewise.
12986 * varasm.c (make_decl_rtl): Likewise.
12987 * reginfo.c (choose_hard_reg_mode): Likewise.
12988 (init_reg_modes_target): Refer directly to
12989 this_target_regs->x_hard_regno_nregs.
12991 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12993 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
12994 instead of hard_regno_nregs.
12996 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12998 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
12999 end_hard_regno instead of hard_regno_nregs.
13000 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
13001 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
13002 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
13003 * ira-color.c (improve_allocation): Likewise.
13004 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13005 * lra-lives.c (mark_regno_live): Likewise.
13006 (mark_regno_dead): Likewise.
13007 * lra-remat.c (operand_to_remat): Likewise.
13008 * lra.c (collect_non_operand_hard_regs): Likewise.
13009 * postreload.c (reload_combine_note_store): Likewise.
13010 (move2add_valid_value_p): Likewise.
13011 * reload.c (regno_clobbered_p): Likewise.
13013 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13015 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
13017 * config/v850/v850.c (v850_reorg): Likewise.
13018 * reload.c (refers_to_regno_for_reload_p): Likewise.
13019 (find_equiv_reg): Likewise.
13020 * reload1.c (reload_reg_reaches_end_p): Likewise.
13022 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13024 * caller-save.c (add_used_regs): Use REG_NREGS instead of
13026 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
13027 * config/arm/arm.c (output_move_neon): Likewise.
13028 (arm_attr_length_move_neon): Likewise.
13029 (neon_split_vcombine): Likewise.
13030 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
13031 (c6x_mark_reg_written): Likewise.
13032 (c6x_dwarf_register_span): Likewise.
13033 * config/i386/i386.c (ix86_save_reg): Likewise.
13034 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
13035 (rws_access_reg): Likewise.
13036 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
13037 * mode-switching.c (create_pre_exit): Likewise.
13038 * ree.c (combine_reaching_defs): Likewise.
13039 (add_removable_extension): Likewise.
13040 * regcprop.c (find_oldest_value_reg): Likewise.
13041 (copyprop_hardreg_forward_1): Likewise.
13042 * reload.c (reload_inner_reg_of_subreg): Likewise.
13043 (push_reload): Likewise.
13044 (combine_reloads): Likewise.
13045 (find_dummy_reload): Likewise.
13046 (reload_adjust_reg_for_mode): Likewise.
13047 * reload1.c (find_reload_regs): Likewise.
13048 (forget_old_reloads_1): Likewise.
13049 (reload_reg_free_for_value_p): Likewise.
13050 (reload_adjust_reg_for_temp): Likewise.
13051 (emit_reload_insns): Likewise.
13052 (delete_output_reload): Likewise.
13053 * sel-sched.c (choose_best_reg_1): Likewise.
13054 (choose_best_pseudo_reg): Likewise.
13056 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13057 Alan Hayward <alan.hayward@arm.com>
13058 David Sherwood <david.sherwood@arm.com>
13060 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
13061 * target.def (slow_unaligned_access): New hook.
13062 * targhooks.h (default_slow_unaligned_access): Declare.
13063 * targhooks.c (default_slow_unaligned_access): New function.
13064 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
13065 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
13066 * doc/tm.texi: Regenerate.
13067 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
13068 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
13069 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
13071 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
13072 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
13074 (rs6000_slow_unaligned_access): New function.
13075 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
13076 (expand_block_compare): Likewise.
13077 (expand_strn_compare): Likewise.
13078 (rs6000_rtx_costs): Likewise.
13079 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
13080 (riscv_slow_unaligned_access): Likewise.
13081 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
13082 (riscv_slow_unaligned_access_p): ...this and make static.
13083 (riscv_option_override): Update accordingly.
13084 (riscv_slow_unaligned_access): New function.
13085 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
13086 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
13087 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
13088 (rs6000_slow_unaligned_access): New function.
13089 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
13090 (rs6000_rtx_costs): Likewise.
13091 * config/rs6000/rs6000-string.c (expand_block_compare)
13092 (expand_strn_compare): Use targetm.slow_unaligned_access instead
13093 of SLOW_UNALIGNED_ACCESS.
13094 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
13095 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
13096 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
13097 of SLOW_UNALIGNED_ACCESS.
13098 * expmed.c (simple_mem_bitfield_p): Likewise.
13099 * expr.c (alignment_for_piecewise_move): Likewise.
13100 (emit_group_load_1): Likewise.
13101 (emit_group_store): Likewise.
13102 (copy_blkmode_from_reg): Likewise.
13103 (emit_push_insn): Likewise.
13104 (expand_assignment): Likewise.
13105 (store_field): Likewise.
13106 (expand_expr_real_1): Likewise.
13107 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
13108 * lra-constraints.c (simplify_operand_subreg): Likewise.
13109 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13110 * gimple-ssa-store-merging.c: Likewise in block comment at start
13112 * tree-ssa-strlen.c: Include target.h.
13113 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
13114 of SLOW_UNALIGNED_ACCESS.
13115 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
13117 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13119 PR rtl-optimization/82185
13120 * expmed.c (emit_store_flag_int): Only test tem if it has been
13123 2017-09-12 Richard Biener <rguenther@suse.de>
13125 PR middle-end/82149
13126 * match.pd ((FTYPE) N CMP CST): Fix typo.
13128 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
13130 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
13132 (mips_near_type_p): Add 'short_call' attribute as a synonym
13134 * doc/extend.texi (short_call): Document new function attribute.
13136 2017-09-12 Jakub Jelinek <jakub@redhat.com>
13139 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
13140 assertion check that in the condition.
13141 (get_atomic_generic_size): Likewise. Before testing if parameter
13142 has pointer type, if it has array type, call for C++
13143 default_conversion to perform array-to-pointer conversion.
13145 2017-09-12 Richard Biener <rguenther@suse.de>
13147 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
13148 for operations we cannot scalarize.
13150 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
13152 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
13153 vectors heap vectors. Clean up comments.
13154 Make visited_bbs a reference.
13155 (profitable_jump_thread_path): Make GC
13156 vectors heap vectors. Clean up comments.
13158 (convert_and_register_jump_thread_path): Make GC vectors heap
13160 (check_subpath_and_update_thread_path): Same. Clean up comments.
13161 Make visited_bbs a reference.
13162 (handle_phi): Abstract common code to to
13163 register_jump_thread_path_if_profitable.
13164 Rename VAR_BB to DEF_BB.
13166 Make GC vectors heap vectors.
13167 Make visited_bbs a reference.
13168 (handle_assignment): Same.
13169 (register_jump_thread_path_if_profitable): New.
13170 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
13172 Make GC vectors heap vectors. Clean up comments.
13173 Make visited_bbs a reference.
13174 (find_jump_threads_backwards): Make visited_bbs live in the stack.
13175 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
13178 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
13181 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
13182 words of E_DImode object are reachable by xtensa_uimm8x4 access.
13184 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
13186 Revert r251800 and r251799.
13188 2017-09-11 Martin Jambor <mjambor@suse.cz>
13191 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
13192 arguments in advance.
13193 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
13194 use it to find predecessor edges.
13195 (naive_outof_ssa): Collect vector of predecessors.
13197 2017-09-08 Jason Merrill <jason@redhat.com>
13199 PR c++/70029 - ICE with ref-qualifier and -flto
13200 * langhooks.h (struct lang_hooks_for_types): Add
13201 copy_lang_qualifiers.
13202 * attribs.c (build_type_attribute_qual_variant): Use it.
13203 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
13205 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
13206 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
13208 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
13211 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
13212 (*mulsi3_sp64): New instruction.
13213 (mulsi3): New expander.
13215 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
13217 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
13219 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13221 * sancov.c: Include memmodel.h.
13223 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
13226 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
13229 2017-09-07 Carl Love <cel@us.ibm.com>
13231 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
13232 the sldi instruction.
13234 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
13236 * sancov.c: Include tm_p.h.
13238 2017-09-07 Jakub Jelinek <jakub@redhat.com>
13241 * output.h (switch_to_other_text_partition): New declaration.
13242 * varasm.c (switch_to_other_text_partition): New function.
13243 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
13244 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
13245 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
13246 to the other text partition before emitting LCL label and switch back
13247 after emitting the word after it.
13249 2017-09-07 Richard Biener <rguenther@suse.de>
13251 * passes.def (pass_split_crit_edges): Remove instance before PRE.
13252 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
13253 critical edges here, after loop init.
13254 (pass_data_pre): Remove PROP_no_crit_edges flags.
13255 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
13256 for valueization of call args to avoid leaking VN_TOP.
13257 (visit_use): Assert we do not visit default defs.
13258 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
13259 Use error_mark_node to more easily detect leaking VN_TOP.
13260 All default-defs are varying, not VN_TOP. Mark them visited.
13261 (run_scc_vn): Make code match comment.
13263 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13265 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
13266 OPTION_MASK_FLOAT128_KEYWORD.
13267 (POWERPC_MASKS): Likewise.
13268 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
13269 support for the -mfloat128-type option, and make -mfloat128
13270 default on PowerPC Linux systems. Define or undefine
13271 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
13272 Define __float128 to be __ieee128 if IEEE 128-bit support is
13273 enabled, or undefine it.
13274 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
13275 Delete defining __FLOAT128_TYPE__.
13276 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
13277 -mfloat128-type option and make -mfloat128 default on PowerPC
13279 (TARGET_FLOAT128_TYPE): Likewise.
13280 (-mfloat128-type): Likewise.
13281 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13282 Delete the -mfloat128-type option and make -mfloat128 default on
13283 PowerPC Linux systems. Always use __ieee128 to be the keyword for
13284 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
13285 128-bit floating point is enabled. Change tests from using
13286 -mfloat128-type to -mfloat128.
13287 (rs6000_mangle_type): Use the correct mangling for the __float128
13288 type even if normal long double is restricted to 64-bits.
13289 (floatn_mode): Enable the _Float128 type by default on VSX Linux
13291 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
13292 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
13293 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
13295 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
13296 documentation for -mfloat128.
13298 2017-09-06 Olivier Hainque <hainque@adacore.com>
13300 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
13302 2017-09-06 Wish Wu <wishwu007@gmail.com>
13303 Jakub Jelinek <jakub@redhat.com>
13305 * asan.c (initialize_sanitizer_builtins): Add
13306 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
13307 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
13308 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
13309 BT_FN_VOID_UINT64_PTR variables.
13310 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
13311 (BT_FN_VOID_UINT16_UINT16): Likewise.
13312 (BT_FN_VOID_UINT32_UINT32): Likewise.
13313 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
13314 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
13315 (BT_FN_VOID_UINT64_PTR): Likewise.
13316 * common.opt (flag_sanitize_coverage): New variable.
13317 (fsanitize-coverage=trace-pc): Remove.
13318 (fsanitize-coverage=): Add.
13319 * flag-types.h (enum sanitize_coverage_code): New enum.
13320 * fold-const.c (fold_range_test): Disable non-short-circuit
13321 optimization if flag_sanitize_coverage.
13322 (fold_truth_andor): Likewise.
13323 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
13324 * opts.c (COVERAGE_SANITIZER_OPT): Define.
13325 (coverage_sanitizer_opts): New array.
13326 (get_closest_sanitizer_option): Add OPTS argument, handle also
13327 OPT_fsanitize_coverage_.
13328 (parse_sanitizer_options): Adjusted to also handle
13329 OPT_fsanitize_coverage_.
13330 (common_handle_option): Add OPT_fsanitize_coverage_.
13331 * sancov.c (instrument_comparison, instrument_switch): New function.
13332 (sancov_pass): Add trace-cmp support.
13333 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
13334 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
13335 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
13336 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
13337 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
13338 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
13339 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
13340 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
13341 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
13342 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
13344 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13346 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13347 error. Only quit immediately if parsing is complete.
13348 (BEGIN): Initialize fatal_err and parse_done.
13349 (begin fpu, end fpu): Check number of arguments.
13350 (begin arch, end arch): Likewise.
13351 (begin cpu, end cpu): Likewise.
13352 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13353 (optalias): Likewise.
13355 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13357 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13358 * config/arm/arm-isa.h: Delete. Move definitions to ...
13359 * arm-cpus.in: ... here. Use new feature and fgroup values.
13360 * config/arm/arm.c (arm_option_override): Use lower case for feature
13362 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13363 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13364 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13366 (print_isa_bits_for): New function.
13367 (gen_isa): New function.
13368 (gen_comm_data): Use print_isa_bits_for.
13369 (define feature): New keyword.
13370 (define fgroup): New keyword.
13371 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
13372 (arm-isa.h): Add rule to generate file.
13373 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13374 case for feature bit names.
13376 2017-09-06 Richard Biener <rguenther@suse.de>
13378 * tree-ssa-pre.c (NECESSARY): Remove.
13379 (create_expression_by_pieces): Do not touch pass-local flags.
13380 (insert_into_preds_of_block): Likewise.
13381 (do_pre_regular_insertion): Likewise.
13382 (eliminate_insert): Likewise.
13383 (eliminate_dom_walker::before_dom_children): Likewise.
13384 (fini_eliminate): Do not look at inserted_exprs.
13385 (mark_operand_necessary): Remove.
13386 (remove_dead_inserted_code): Replace with simple work-list
13387 algorithm based on inserted_exprs and SSA uses.
13388 (pass_pre::execute): Re-order fini_eliminate and
13389 remove_dead_inserted_code.
13391 2017-09-06 Olivier Hainque <hainque@adacore.com>
13393 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
13394 for VxWorks 7. Adjust surrounding comments.
13396 2017-09-06 Richard Biener <rguenther@suse.de>
13398 * gimple-ssa-strength-reduction.c
13399 (find_candidates_dom_walker::before_dom_children): Also allow
13402 2017-09-06 Richard Biener <rguenther@suse.de>
13404 PR tree-optimization/82108
13405 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
13406 for gap in the non-permutation SLP case.
13408 2017-09-06 Martin Jambor <mjambor@suse.cz>
13410 PR tree-optimization/82078
13411 * tree-sra.c (sort_and_splice_var_accesses): Move call to
13412 add_access_to_work_queue...
13413 (build_accesses_from_assign): ...here.
13414 (propagate_all_subaccesses): Make sure racc is the group
13415 representative, if there is one.
13417 2017-09-06 Jakub Jelinek <jakub@redhat.com>
13419 PR middle-end/82095
13420 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
13423 2017-09-06 Richard Biener <rguenther@suse.de>
13425 * gimple-ssa-strength-reduction.c
13426 (find_candidates_dom_walker::before_doom_children): Use a
13427 type and not a mode check.
13429 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
13432 * config/arm/predicates.md (arm_general_adddi_operand): Create new
13434 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
13436 2017-09-05 Jeff Law <law@redhat.com>
13438 PR tree-optimization/64910
13439 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
13440 cases where we have 3 or more operands.
13442 2017-09-05 Jakub Jelinek <jakub@redhat.com>
13444 PR middle-end/81768
13445 * omp-low.c (lower_omp_for): Recompute tree invariant if
13446 gimple_omp_for_initial/final is ADDR_EXPR.
13448 PR middle-end/81768
13449 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
13450 into gimple val before gimplification fo the COND_EXPR.
13452 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
13454 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
13455 REGION_COPY argument.
13456 (thread_through_all_blocks): Remove unused argument to
13457 duplicate_thread_path.
13459 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13460 Alan Hayward <alan.hayward@arm.com>
13461 David Sherwood <david.sherwood@arm.com>
13463 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
13464 Take a scalar_mode rather than a machine_mode.
13465 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13466 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
13467 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13468 (aarch64_gen_adjusted_ldpstp): Likewise.
13469 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
13471 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13472 Alan Hayward <alan.hayward@arm.com>
13473 David Sherwood <david.sherwood@arm.com>
13475 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
13476 Take a scalar_int_mode instead of a machine_mode.
13477 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13478 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13479 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13480 (aarch64_simd_attr_length_rglist): Delete.
13481 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
13482 a scalar_int_mode instead of a machine_mode.
13483 (aarch64_add_offset): Likewise.
13484 (aarch64_internal_mov_immediate): Likewise
13485 (aarch64_add_constant_internal): Likewise.
13486 (aarch64_add_constant): Likewise.
13487 (aarch64_movw_imm): Likewise.
13488 (aarch64_rtx_arith_op_extract_p): Likewise.
13489 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13490 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13491 Remove assert that the mode isn't a vector.
13492 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13493 (aarch64_expand_mov_immediate): Update calls after above changes.
13494 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
13495 (aarch64_and_bitmask_imm): Check for scalar integer modes.
13496 (aarch64_move_imm): Likewise.
13497 (aarch64_can_const_movi_rtx_p): Likewise.
13498 (aarch64_strip_extend): Likewise.
13499 (aarch64_extr_rtx_p): Likewise.
13500 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
13501 a CONST_INT when the mode parameter is VOIDmode.
13502 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
13504 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13506 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
13507 * stor-layout.c (bitwise_mode_for_mode): Likewise.
13508 (bitwise_type_for_mode): Update accordingly.
13510 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13512 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
13513 * stor-layout.c (mode_for_size_tree): Likewise.
13514 (mode_for_array): Update accordingly.
13515 (layout_decl): Likewise.
13516 (compute_record_mode): Likewise. Only set the mode once.
13518 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13520 * target.def (get_mask_mode): Change return type to opt_mode.
13522 * doc/tm.texi: Regenerate.
13523 * targhooks.h (default_get_mask_mode): Return an opt_mode.
13524 * targhooks.c (default_get_mask_mode): Likewise.
13525 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
13526 * optabs-query.c (can_vec_mask_load_store_p): Update use of
13527 targetm.get_mask_mode.
13528 * tree.c (build_truth_vector_type): Likewise.
13530 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13532 * machmode.h (mode_for_vector): Return an opt_mode.
13533 * stor-layout.c (mode_for_vector): Likewise.
13534 (mode_for_int_vector): Update accordingly.
13535 (layout_type): Likewise.
13536 * config/i386/i386.c (emit_memmov): Likewise.
13537 (ix86_expand_set_or_movmem): Likewise.
13538 (ix86_expand_vector_init): Likewise.
13539 (ix86_get_mask_mode): Likewise.
13540 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
13542 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
13543 * expmed.c (extract_bit_field_1): Likewise.
13544 * expr.c (expand_expr_real_2): Likewise.
13545 * optabs-query.c (can_vec_perm_p): Likewise.
13546 (can_vec_mask_load_store_p): Likewise.
13547 * optabs.c (expand_vec_perm): Likewise.
13548 * targhooks.c (default_get_mask_mode): Likewise.
13549 * tree-vect-stmts.c (vectorizable_store): Likewise.
13550 (vectorizable_load): Likewise.
13551 (get_vectype_for_scalar_type_and_size): Likewise.
13553 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13555 * machmode.h (mode_for_int_vector): New function.
13556 * stor-layout.c (mode_for_int_vector): Likewise.
13557 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
13558 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
13559 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
13560 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
13561 (s390_expand_vcond): Likewise.
13563 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13565 * machmode.h (opt_machine_mode): New type.
13566 (opt_mode<T>): Allow construction from anything that can be
13568 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
13569 (mode_for_size): Return an opt_machine_mode.
13570 * stor-layout.c (mode_for_size): Likewise.
13571 (mode_for_size_tree): Update call accordingly.
13572 (bitwise_mode_for_mode): Likewise.
13573 (make_fract_type): Likewise.
13574 (make_accum_type): Likewise.
13575 * caller-save.c (replace_reg_with_saved_mem): Update call
13577 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13578 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13579 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13580 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13581 * expmed.c (extract_bit_field_1): Likewise.
13582 * reload.c (get_secondary_mem): Likewise.
13583 * varasm.c (assemble_integer): Likewise.
13584 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
13587 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13589 * machmode.h (decimal_float_mode_for_size): New function.
13590 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
13591 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
13592 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
13593 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
13594 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
13595 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
13597 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13599 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
13600 (get_builtin_sync_mode): Likewise.
13601 (expand_ifn_atomic_compare_exchange): Likewise.
13602 (expand_builtin_atomic_clear): Likewise.
13603 (expand_builtin_atomic_test_and_set): Likewise.
13604 (fold_builtin_atomic_always_lock_free): Likewise.
13605 * calls.c (compute_argument_addresses): Likewise.
13606 (emit_library_call_value_1): Likewise.
13607 (store_one_arg): Likewise.
13608 * combine.c (combine_instructions): Likewise.
13609 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
13610 * config/arm/arm.c (arm_function_value): Likewise.
13611 (aapcs_allocate_return_reg): Likewise.
13612 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
13613 * config/i386/i386.c (construct_container): Likewise.
13614 (ix86_gimplify_va_arg): Likewise.
13615 (ix86_expand_sse_cmp): Likewise.
13616 (emit_memmov): Likewise.
13617 (emit_memset): Likewise.
13618 (expand_small_movmem_or_setmem): Likewise.
13619 (ix86_expand_pextr): Likewise.
13620 (ix86_expand_pinsr): Likewise.
13621 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
13622 * config/microblaze/microblaze.c (microblaze_block_move_straight):
13624 * config/mips/mips.c (mips_function_value_1) Likewise.
13625 (mips_block_move_straight): Likewise.
13626 (mips_expand_ins_as_unaligned_store): Likewise.
13627 * config/powerpcspe/powerpcspe.c
13628 (rs6000_darwin64_record_arg_advance_flush): Likewise.
13629 (rs6000_darwin64_record_arg_flush): Likewise.
13630 * config/rs6000/rs6000.c
13631 (rs6000_darwin64_record_arg_advance_flush): Likewise.
13632 (rs6000_darwin64_record_arg_flush): Likewise.
13633 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
13634 (sparc_function_value_1): Likewise.
13635 * config/spu/spu.c (adjust_operand): Likewise.
13636 (spu_emit_branch_or_set): Likewise.
13637 (arith_immediate_p): Likewise.
13638 * emit-rtl.c (gen_lowpart_common): Likewise.
13639 * expr.c (expand_expr_real_1): Likewise.
13640 * function.c (assign_parm_setup_block): Likewise.
13641 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
13642 * reload1.c (alter_reg): Likewise.
13643 * stor-layout.c (mode_for_vector): Likewise.
13644 (layout_type): Likewise.
13646 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13648 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
13649 (spu_convert_move): Likewise.
13650 * lower-subreg.c (resolve_simple_move): Likewise.
13652 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13655 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
13656 define_insn to a define_expand.
13657 (altivec_vsum2sws_direct): New define_insn.
13658 (altivec_vsumsws): Convert from a define_insn to a define_expand.
13660 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
13662 * config/arm/arm.c (arm_option_params_internal): Improve setting of
13665 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
13670 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
13671 realign stack if stack alignment needed is less than incoming
13674 2017-09-05 Marek Polacek <polacek@redhat.com>
13677 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
13680 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
13682 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
13684 2017-09-05 Richard Biener <rguenther@suse.de>
13686 PR tree-optimization/82084
13687 * fold-const.c (can_native_encode_string_p): Handle wide characters.
13689 2017-09-05 Richard Biener <rguenther@suse.de>
13691 PR tree-optimization/82102
13692 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
13694 2017-09-05 Martin Liska <mliska@suse.cz>
13696 PR tree-optimization/82032
13697 * tree-cfg.c (generate_range_test): New function.
13698 * tree-cfg.h (generate_range_test): Declared here.
13699 * tree-cfgcleanup.c (convert_single_case_switch): New function.
13700 (cleanup_control_expr_graph): Use it.
13701 * tree-switch-conversion.c (try_switch_expansion): Remove
13703 (emit_case_nodes): Use generate_range_test.
13705 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
13708 * config/i386/i386.md (*<btsc><mode>_mask): Add
13709 TARGET_USE_BT to insn constraint.
13710 (*btr<mode>_mask): Ditto.
13712 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
13714 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
13715 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
13717 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
13720 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
13721 TARGET_NEON and TARGET_IWMMXT.
13722 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
13723 TARGET_NEON and TARGET_IWMMXT.
13724 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
13726 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
13728 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
13729 (ix86_rewrite_tls_address): Ditto.
13730 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
13731 (ix86_rewrite_tls_address_1): Ditto.
13732 (ix86_rewrite_tls_address): Ditto.
13733 * config/i386/predicates.md (tls_address_pattern): New predicate.
13734 * config/i386/i386.md (TLS address splitter): New splitter.
13736 2017-09-04 Richard Biener <rguenther@suse.de>
13738 PR tree-optimization/82084
13739 * fold-const.h (can_native_encode_string_p): Declare.
13740 * fold-const.c (can_native_encode_string_p): Factor out from ...
13741 (native_encode_string): ... here.
13742 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
13743 vectorizing stores from constants we later cannot handle.
13745 2017-09-04 Marek Polacek <polacek@redhat.com>
13748 * doc/invoke.texi: Update -Wtautological-compare documentation.
13750 2017-09-04 Jeff Law <law@redhat.com>
13752 PR tree-optimization/64910
13753 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
13754 swap the first and last operand if the last is a constant.
13756 2017-09-04 Marek Polacek <polacek@redhat.com>
13759 * convert.c (do_narrow): When sanitizing signed integer overflows,
13760 bail out for signed types.
13761 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
13763 2017-09-04 Richard Biener <rguenther@suse.de>
13765 PR tree-optimization/82060
13766 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13767 Move devirtualization after stmt folding and before EH/AB/noreturn
13768 cleanup to get the stmt refs canonicalized. Use a bool instead
13769 of gimple_modified_p since that doesn't work for NOPs. Schedule
13770 NOPs generated by folding for removal.
13772 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13773 Alan Hayward <alan.hayward@arm.com>
13774 David Sherwood <david.sherwood@arm.com>
13776 * coretypes.h (pad_direction): New enum.
13777 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
13778 (FUNCTION_ARG_PADDING): Likewise.
13779 * target.def (function_arg_padding): New hook.
13780 * targhooks.h (default_function_arg_padding): Declare.
13781 * targhooks.c (default_function_arg_padding): New function.
13782 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
13783 (TARGET_FUNCTION_ARG_PADDING): ...this.
13784 * doc/tm.texi: Regenerate.
13785 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
13786 instead of direction.
13787 (compute_argument_addresses): Likewise.
13788 (load_register_parameters): Likewise.
13789 (emit_library_call_value_1): Likewise.
13790 (store_one_arg): Use targetm.calls.function_arg_padding instead
13791 of FUNCTION_ARG_PADDING.
13792 (must_pass_in_stack_var_size_or_pad): Likewise.
13793 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
13794 (emit_group_store): Likewise.
13795 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
13796 instead of FUNCTION_ARG_PADDING.
13797 (emit_push_insn): Likewise, and propagate enum change throughout
13799 * function.h (direction): Delete.
13800 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
13802 * function.c (assign_parm_find_stack_rtl): Likewise.
13803 (assign_parm_setup_block_p): Likewise.
13804 (assign_parm_setup_block): Likewise.
13805 (gimplify_parameters): Likewise.
13806 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
13807 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
13809 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
13810 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13811 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
13812 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
13813 (aarch64_function_arg_padding): ...this new function.
13814 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
13815 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13816 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
13817 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13818 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
13819 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13820 (arm_pad_arg_upward): Replace with...
13821 (arm_function_arg_padding): ...this new function.
13822 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
13824 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
13825 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
13826 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13827 (ia64_hpux_function_arg_padding): Replace with...
13828 (ia64_function_arg_padding): ...this new function. Use pad_direction
13829 instead of direction. Check for TARGET_HPUX.
13830 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
13831 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13832 (iq2000_function_arg_padding): New function.
13833 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
13834 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
13835 (mips_function_arg_padding): ...this new function.
13836 (mips_pad_reg_upward): Update accordingly.
13837 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13838 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
13839 targetm.calls.function_arg_padding.
13840 (FUNCTION_ARG_PADDING): Delete.
13841 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13842 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
13843 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13844 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
13845 (nios2_block_reg_padding): Return pad_direction instead of direction.
13846 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
13847 instead of direction.
13848 (nios2_function_arg_padding): Likewise. Make static.
13849 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13850 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
13851 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
13852 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
13853 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13854 (pa_function_arg_padding): Make static. Return pad_direction instead
13856 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
13857 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13858 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
13859 instead of direction. Use targetm.calls.function_arg_padding.
13860 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
13861 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13862 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
13863 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
13864 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
13866 (function_arg_padding): Rename to...
13867 (rs6000_function_arg_padding): ...this. Make static. Return
13868 pad_direction instead of direction.
13869 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13870 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
13871 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13872 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
13873 instead of direction. Use targetm.calls.function_arg_padding.
13874 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
13875 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13876 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
13877 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
13878 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13879 (function_arg_padding): Rename to...
13880 (rs6000_function_arg_padding): ...this. Make static. Return
13881 pad_direction instead of direction.
13882 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13883 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
13884 * config/s390/s390.c (s390_function_arg_padding): New function.
13885 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13886 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
13887 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
13888 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13889 (function_arg_padding): Rename to...
13890 (sparc_function_arg_padding): ...this. Make static. Return
13891 pad_direction instead of direction.
13892 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
13893 * config/spu/spu.c (spu_function_arg_padding): New function.
13894 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13895 * system.h (FUNCTION_ARG_PADDING): Poison.
13897 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13898 Alan Hayward <alan.hayward@arm.com>
13899 David Sherwood <david.sherwood@arm.com>
13901 * target.def (modes_tieable_p): New hook.
13902 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
13903 (TARGET_MODES_TIEABLE_P): ...this.
13904 * doc/tm.texi.in: Regenerate.
13905 * hooks.h (hook_bool_mode_mode_true): Declare.
13906 * hooks.c (hook_bool_mode_mode_true): New function.
13907 * combine.c (subst): Use targetm.modes_tieable_p instead of
13909 * dse.c (find_shift_sequence): Likewise.
13910 * expmed.c (extract_low_bits): Likewise.
13911 * lower-subreg.c: Include target.h.
13912 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
13914 * rtlanal.c (rtx_cost): Likewise.
13915 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
13916 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
13917 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
13918 (TARGET_MODES_TIEABLE_P): Redefine.
13919 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
13920 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
13921 (TARGET_MODES_TIEABLE_P): Redefine.
13922 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
13923 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
13924 (arc_modes_tieable_p): New function.
13925 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
13926 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
13927 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
13928 (arm_modes_tieable_p): Make static.
13929 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
13930 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
13931 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
13932 (TARGET_MODES_TIEABLE_P): Redefine.
13933 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
13934 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
13935 (TARGET_MODES_TIEABLE_P): Redefine.
13936 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
13937 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
13938 (cr16_modes_tieable_p): New function.
13939 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
13940 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
13941 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
13942 (TRULY_NOOP_TRUNCATION): Update comment.
13943 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
13944 (TRULY_NOOP_TRUNCATION): Update comment.
13945 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
13946 (frv_modes_tieable_p): New function.
13947 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
13948 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
13949 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
13950 (TARGET_MODES_TIEABLE_P): Redefine.
13951 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
13952 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
13953 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
13954 (TARGET_MODES_TIEABLE_P): Redefine.
13955 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
13956 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
13957 (ia64_modes_tieable_p): New function.
13958 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
13959 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
13960 (iq2000_modes_tieable_p): New function.
13961 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
13962 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
13963 (lm32_modes_tieable_p): New function.
13964 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
13965 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
13966 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
13967 (TARGET_MODES_TIEABLE_P): Redefine.
13968 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
13969 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
13970 (m32r_modes_tieable_p): New function.
13971 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
13972 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
13973 (m68k_modes_tieable_p): New function.
13974 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
13975 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
13976 (mcore_modes_tieable_p): New function.
13977 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
13978 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
13980 (TARGET_MODES_TIEABLE_P): Redefine.
13981 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
13982 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
13983 * config/mips/mips.c (mips_modes_tieable_p): Make static.
13984 (TARGET_MODES_TIEABLE_P): Redefine.
13985 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
13986 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
13987 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
13988 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
13989 (mn10300_modes_tieable_p): ...this and make static.
13990 (TARGET_MODES_TIEABLE_P): Redefine.
13991 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
13992 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
13993 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
13994 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
13995 (msp430_modes_tieable_p): Make static.
13996 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
13997 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
13998 (TARGET_MODES_TIEABLE_P): Redefine.
13999 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
14000 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
14001 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
14002 (TARGET_MODES_TIEABLE_P): Redefine.
14003 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
14004 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
14005 * config/pa/pa.c (pa_modes_tieable_p): Make static.
14006 (TARGET_MODES_TIEABLE_P): Redefine.
14007 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
14008 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
14009 (pdp11_modes_tieable_p): New function.
14010 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
14011 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
14012 (rs6000_modes_tieable_p): New function.
14013 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
14014 * config/powerpcspe/powerpcspe.md: Update comment.
14015 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
14016 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
14017 (TARGET_MODES_TIEABLE_P): Redefine.
14018 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
14019 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
14020 (rl78_modes_tieable_p): New function.
14021 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
14022 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
14023 (rs6000_modes_tieable_p): New function.
14024 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
14025 * config/rs6000/rs6000.md: Update comment.
14026 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
14027 * config/rx/rx.c (rx_modes_tieable_p): New function.
14028 (TARGET_MODES_TIEABLE_P): Redefine.
14029 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
14030 * config/s390/s390.c (s390_modes_tieable_p): New function.
14031 (TARGET_MODES_TIEABLE_P): Redefine.
14032 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
14033 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
14034 (sh_modes_tieable_p): New function.
14035 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
14036 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
14037 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
14038 (sparc_modes_tieable_p): Make static.
14039 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
14040 * config/spu/spu.c (spu_modes_tieable_p): New function.
14041 (TARGET_MODES_TIEABLE_P): Redefine.
14042 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
14043 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
14044 (TARGET_MODES_TIEABLE_P): Redefine.
14045 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
14046 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
14047 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
14048 * config/v850/v850.c (v850_modes_tieable_p): New function.
14049 (TARGET_MODES_TIEABLE_P): Redefine.
14050 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
14051 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
14052 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
14053 (visium_modes_tieable_p): New function.
14054 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
14055 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
14056 (xtensa_modes_tieable_p): New function.
14057 * system.h (MODES_TIEABLE_P): Poison.
14059 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14060 Alan Hayward <alan.hayward@arm.com>
14061 David Sherwood <david.sherwood@arm.com>
14063 * target.def (hard_regno_mode_ok): New hook.
14064 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
14065 (TARGET_HARD_REGNO_MODE_OK): ...this.
14066 * doc/tm.texi.in: Regenerate.
14067 * hooks.h (hook_bool_uint_mode_true): Declare.
14068 * hooks.c (hook_bool_uint_mode_true): New function.
14069 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
14070 HARD_REGNO_MODE_OK.
14071 * genpreds.c (write_insn_preds_c): Add an include of target.h.
14072 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
14073 instead of HARD_REGNO_MODE_OK.
14074 * caller-save.c: Include target.h.
14075 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
14076 HARD_REGNO_MODE_OK.
14077 * combine.c (can_combine_p): Likewise.
14078 (combinable_i3pat): Likewise.
14079 (can_change_dest_mode): Likewise.
14080 * expr.c (init_expr_target): Likewise.
14081 (convert_move): Likewise.
14082 (convert_modes): Likewise.
14083 * ira.c (setup_prohibited_class_mode_regs): Likewise.
14084 (setup_prohibited_mode_move_regs): Likewise.
14085 * ira.h (target_ira): Likewise.
14086 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14087 * lra-constraints.c (process_alt_operands): Likewise.
14088 (split_reg): Likewise.
14089 * recog.c (peep2_find_free_register): Likewise.
14090 * ree.c (combine_reaching_defs): Likewise.
14091 * regcprop.c (maybe_mode_change): Likewise.
14092 * reginfo.c (init_reg_sets_1): Likewise.
14093 (choose_hard_reg_mode): Likewise.
14094 (simplifiable_subregs): Likewise.
14095 * regrename.c (check_new_reg_p): Likewise.
14096 * reload.c (find_valid_class): Likewise.
14097 (find_valid_class_1): Likewise.
14098 (reload_inner_reg_of_subreg): Likewise.
14099 (push_reload): Likewise.
14100 (combine_reloads): Likewise.
14101 (find_dummy_reload): Likewise.
14102 (find_reloads): Likewise.
14103 * reload1.c (find_reg): Likewise.
14104 (set_reload_reg): Likewise.
14105 (allocate_reload_reg): Likewise.
14106 (choose_reload_regs): Likewise.
14107 (reload_adjust_reg_for_temp): Likewise.
14108 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
14109 (simplify_subreg_regno): Likewise.
14110 * sel-sched.c (init_regs_for_mode): Likewise.
14111 * varasm.c (make_decl_rtl): Likewise.
14112 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
14113 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
14114 HARD_REGNO_MODE_OK.
14115 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
14116 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
14117 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14118 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
14119 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
14120 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14121 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
14122 (arc_mode_class): Delete.
14123 (HARD_REGNO_MODE_OK): Delete.
14124 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14125 (arc_hard_regno_mode_ok): Rename old array to...
14126 (arc_hard_regno_mode_ok_modes): ...this.
14127 (arc_conditional_register_usage): Update accordingly.
14128 (arc_mode_class): Make static.
14129 (arc_hard_regno_mode_ok): New function.
14130 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
14131 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
14132 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14133 (arm_hard_regno_mode_ok): Make static.
14134 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
14135 HARD_REGNO_MODE_OK.
14136 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
14137 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
14138 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
14140 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14141 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
14142 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
14143 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
14144 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
14145 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14146 * config/bfin/predicates.md (valid_reg_operand): Use
14147 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
14148 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
14149 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
14150 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14151 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
14152 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
14153 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14154 (cr16_hard_regno_mode_ok): Make static and return a bool.
14155 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
14156 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14157 (cris_hard_regno_mode_ok): New function.
14158 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
14159 (epiphany_mode_class): Delete.
14160 (HARD_REGNO_MODE_OK): Delete.
14161 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
14162 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14163 (hard_regno_mode_ok): Rename to...
14164 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
14165 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
14166 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
14167 HARD_REGNO_MODE_OK.
14168 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
14169 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
14170 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14171 (frv_hard_regno_mode_ok): Make static and return a bool.
14172 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
14173 HARD_REGNO_MODE_OK.
14174 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
14175 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
14176 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
14177 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
14179 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14180 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
14181 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
14182 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
14184 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14185 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
14186 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14187 (ia64_hard_regno_mode_ok): New function.
14188 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
14189 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14190 (iq2000_hard_regno_mode_ok): New function.
14191 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
14192 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14193 (lm32_hard_regno_mode_ok): New function.
14194 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
14195 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
14196 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
14197 instead of HARD_REGNO_MODE_OK.
14198 (m32c_hard_regno_ok): Rename to...
14199 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
14200 (m32c_cannot_change_mode_class): Update accordingly.
14201 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14202 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
14203 (m32r_mode_class): Delete.
14204 (HARD_REGNO_MODE_OK): Delete.
14205 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14206 (m32r_hard_regno_mode_ok): Rename to...
14207 (m32r_hard_regno_modes): ...this.
14208 (m32r_mode_class): Make static.
14209 (m32r_hard_regno_mode_ok): New function.
14210 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
14211 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
14212 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14213 (m68k_hard_regno_mode_ok): Make static.
14214 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
14215 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14216 (mcore_hard_regno_mode_ok): New function.
14217 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
14218 (HARD_REGNO_MODE_OK): Delete.
14219 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
14221 (microblaze_hard_regno_mode_ok_p): ...this and make static.
14222 (microblaze_hard_regno_mode_ok): New function.
14223 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14224 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
14225 (mips_hard_regno_mode_ok): Delete.
14226 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
14227 (mips_hard_regno_mode_ok_p): ...this and make static.
14228 (mips_hard_regno_mode_ok_p): Rename to...
14229 (mips_hard_regno_mode_ok_uncached): ...this.
14230 (mips_hard_regno_mode_ok): New function.
14231 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
14232 of HARD_REGNO_MODE_OK.
14233 (mips_option_override): Update after above name changes.
14234 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14235 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
14236 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
14237 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
14238 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
14239 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14240 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
14241 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
14242 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
14243 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14244 (msp430_hard_regno_mode_ok): Make static and return a bool.
14245 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
14246 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
14247 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
14249 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14250 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
14251 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
14252 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
14253 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
14254 (PA_HARD_REGNO_MODE_OK): ...this
14255 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
14256 (PA_HARD_REGNO_MODE_OK): ...this.
14257 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14258 (pa_hard_regno_mode_ok): New function.
14259 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
14260 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14261 (pdp11_hard_regno_mode_ok): New function.
14262 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
14263 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
14265 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
14267 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14268 (rs6000_hard_regno_mode_ok): Rename to...
14269 (rs6000_hard_regno_mode_ok_uncached): ...this.
14270 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14271 (rs6000_hard_regno_mode_ok): New function.
14272 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
14273 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
14274 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
14275 (riscv_hard_regno_mode_ok): ...this and make static.
14276 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14277 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
14278 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
14279 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14280 (rl78_hard_regno_mode_ok): Make static and return bool.
14281 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
14282 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
14284 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
14285 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14286 (rs6000_hard_regno_mode_ok): Rename to...
14287 (rs6000_hard_regno_mode_ok_uncached): ...this.
14288 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14289 (rs6000_hard_regno_mode_ok): New function.
14290 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
14291 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
14292 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14293 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
14294 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
14295 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
14296 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14297 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
14298 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
14299 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14300 (sh_hard_regno_mode_ok): Make static.
14301 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
14302 instead of HARD_REGNO_MODE_OK.
14303 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
14304 (sparc_mode_class): Delete.
14305 (HARD_REGNO_MODE_OK): Delete.
14306 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14307 (hard_regno_mode_classes): Make static.
14308 (sparc_mode_class): Likewise.
14309 (sparc_hard_regno_mode_ok): New function.
14310 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
14311 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
14312 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
14314 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14315 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
14316 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
14317 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
14318 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
14319 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14320 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
14321 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
14322 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14323 (visium_hard_regno_mode_ok): New function.
14324 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
14325 instead of HARD_REGNO_MODE_OK.
14326 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
14327 (HARD_REGNO_MODE_OK): Delete.
14328 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
14329 (xtensa_hard_regno_mode_ok_p): ...this and make static.
14330 (xtensa_option_override): Update accordingly.
14331 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14332 (xtensa_hard_regno_mode_ok): New function.
14333 * system.h (HARD_REGNO_MODE_OK): Poison.
14335 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14336 Alan Hayward <alan.hayward@arm.com>
14337 David Sherwood <david.sherwood@arm.com>
14339 * target.def (hard_regno_call_part_clobbered): New hook.
14340 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
14341 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
14342 * doc/tm.texi: Regenerate.
14343 * hooks.h (hook_bool_uint_mode_false): Declare.
14344 * hooks.c (hook_bool_uint_mode_false): New function.
14345 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14346 * cselib.c (cselib_process_insn): Use
14347 targetm.hard_regno_call_part_clobbered instead of
14348 HARD_REGNO_CALL_PART_CLOBBERED.
14349 * ira-conflicts.c (ira_build_conflicts): Likewise.
14350 * ira-costs.c (ira_tune_allocno_costs): Likewise.
14351 * lra-constraints.c (need_for_call_save_p): Likewise.
14352 * lra-lives.c: Include target.h.
14353 (check_pseudos_live_through_calls): Use
14354 targetm.hard_regno_call_part_clobbered instead of
14355 HARD_REGNO_CALL_PART_CLOBBERED.
14356 * regcprop.c: Include target.h.
14357 (copyprop_hardreg_forward_1): Use
14358 targetm.hard_regno_call_part_clobbered instead of
14359 HARD_REGNO_CALL_PART_CLOBBERED.
14360 * reginfo.c (choose_hard_reg_mode): Likewise.
14361 * regrename.c (check_new_reg_p): Likewise.
14362 * reload.c (find_equiv_reg): Likewise.
14363 * reload1.c (emit_reload_insns): Likewise.
14364 * sched-deps.c (deps_analyze_insn): Likewise.
14365 * sel-sched.c (init_regs_for_mode): Likewise.
14366 (mark_unavailable_hard_regs): Likewise.
14367 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
14368 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14369 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
14371 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14372 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14373 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
14375 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
14377 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14378 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14379 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
14381 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14382 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14383 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
14385 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14386 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
14388 * config/powerpcspe/powerpcspe.c
14389 (rs6000_hard_regno_call_part_clobbered): New function.
14390 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14391 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14392 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
14394 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14395 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14396 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
14398 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14399 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14400 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
14402 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14403 Alan Hayward <alan.hayward@arm.com>
14404 David Sherwood <david.sherwood@arm.com>
14406 * rtl.h (subreg_memory_offset): Declare.
14407 * emit-rtl.c (subreg_memory_offset): New function.
14408 * expmed.c (store_bit_field_1): Use it.
14409 * expr.c (undefined_operand_subword_p): Likewise.
14410 * simplify-rtx.c (simplify_subreg): Likewise.
14412 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
14414 PR rtl-optimization/57448
14417 * optabs.c (expand_atomic_load): Place compiler memory barriers if
14418 using atomic_load pattern.
14419 (expand_atomic_store): Likewise.
14421 2017-09-04 Jakub Jelinek <jakub@redhat.com>
14424 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
14425 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
14426 handling. Use replace_call_with_value with NULL instead of
14427 gsi_replace, unlink_stmt_vdef and release_defs.
14429 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
14432 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
14434 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14437 * rtl.h (emit_library_call_value_1): Declare.
14438 (emit_library_call): Replace declaration with a series of overloads.
14439 Remove the parameter count argument.
14440 (emit_library_call_value): Likewise.
14441 * calls.c (emit_library_call_value_1): Make global. Replace varargs
14442 with an "rtx_mode_t *".
14443 (emit_library_call_value): Delete.
14444 (emit_library_call): Likewise.
14445 * asan.c (asan_emit_stack_protection): Update calls accordingly.
14446 (asan_emit_allocas_unpoison): Likewise.
14447 * builtins.c (expand_builtin_powi): Likewise.
14448 (expand_asan_emit_allocas_unpoison): Likewise.
14449 * cfgexpand.c (expand_main_function): Likewise.
14450 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
14451 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
14452 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
14453 * config/arm/arm.c (arm_trampoline_init): Likewise.
14454 (arm_call_tls_get_addr): Likewise.
14455 (arm_expand_divmod_libfunc): Likewise.
14456 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
14457 (smulsi3_highpart): Likewise.
14458 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
14459 (c6x_expand_compare): Likewise.
14460 (c6x_expand_movmem): Likewise.
14461 * config/frv/frv.c (frv_trampoline_init): Likewise.
14462 * config/i386/i386.c (ix86_trampoline_init): Likewise.
14463 (ix86_expand_divmod_libfunc): Likewise.
14464 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
14465 (ia64_expand_compare): Likewise.
14466 (ia64_profile_hook): Likewise.
14467 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
14468 (nonlocal_goto): Likewise.
14469 (restore_stack_nonlocal): Likewise.
14470 * config/m32r/m32r.c (block_move_call): Likewise.
14471 (m32r_trampoline_init): Likewise.
14472 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
14473 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
14474 (m68k_call_m68k_read_tp): Likewise.
14475 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
14476 (microblaze_expand_divide): Likewise.
14477 * config/mips/mips.h (mips_args): Likewise.
14478 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
14479 (MIPS_ICACHE_SYNC): Likewise.
14480 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
14481 (nios2_trampoline_init): Likewise.
14482 * config/pa/pa.c (hppa_tls_call): Likewise.
14483 (pa_trampoline_init): Likewise.
14484 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
14485 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
14486 (expand_strn_compare): Likewise.
14487 (rs6000_generate_compare): Likewise.
14488 (rs6000_expand_float128_convert): Likewise.
14489 (output_profile_hook): Likewise.
14490 (rs6000_trampoline_init): Likewise.
14491 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
14492 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
14493 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
14494 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
14495 (rs6000_generate_compare): Likewise.
14496 (rs6000_expand_float128_convert): Likewise.
14497 (output_profile_hook): Likewise.
14498 (rs6000_trampoline_init): Likewise.
14499 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
14500 * config/sh/sh.c (sh_trampoline_init): Likewise.
14501 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
14502 (sparc_emit_float_lib_cmp): Likewise.
14503 (sparc32_initialize_trampoline): Likewise.
14504 (sparc64_initialize_trampoline): Likewise.
14505 (sparc_profile_hook): Likewise.
14506 * config/spu/spu.c (ea_load_store): Likewise.
14507 * config/spu/spu.md (floatunssidf2): Likewise.
14508 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
14509 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
14510 * config/visium/visium.c (expand_block_move_4): Likewise.
14511 (expand_block_move_2): Likewise.
14512 (expand_block_move_1): Likewise.
14513 (expand_block_set_4): Likewise.
14514 (expand_block_set_2): Likewise.
14515 (expand_block_set_1): Likewise.
14516 (visium_trampoline_init): Likewise.
14517 (visium_profile_hook): Likewise.
14518 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
14519 (xtensa_setup_frame_addresses): Likewise.
14520 (xtensa_trampoline_init): Likewise.
14521 * except.c (sjlj_emit_function_enter): Likewise.
14522 (sjlj_emit_function_exit): Likewise.
14523 * explow.c (allocate_dynamic_stack_space): Likewise.
14524 (probe_stack_range): Likewise.
14525 * expr.c (convert_mode_scalar): Likewise.
14526 * optabs.c (expand_binop): Likewise.
14527 (expand_twoval_binop_libfunc): Likewise.
14528 (expand_unop): Likewise.
14529 (prepare_cmp_insn): Likewise.
14530 (prepare_float_lib_cmp): Likewise.
14531 (expand_float): Likewise.
14532 (expand_fix): Likewise.
14533 (expand_fixed_convert): Likewise.
14534 (maybe_emit_sync_lock_test_and_set): Likewise.
14535 (expand_atomic_compare_and_swap): Likewise.
14536 (expand_mem_thread_fence): Likewise.
14537 (expand_atomic_fetch_op): Likewise.
14539 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
14541 * doc/generic.texi (OpenACC): Adjust URL.
14542 * doc/invoke.texi (C Dialect Options): Ditto.
14544 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
14546 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
14547 predicate for operand 1. Add (m,<S>) constraint.
14548 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
14549 Prevent memory operand 1 with register operand 2.
14551 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
14553 PR rtl-optimization/82024
14554 * combine.c (try_combine): If the combination result is a PARALLEL,
14555 and we only need to retain the SET in there that would be placed
14556 at I2, check that we can place that at I3 instead, before doing so.
14558 2017-09-01 Jakub Jelinek <jakub@redhat.com>
14561 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
14563 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
14564 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
14567 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
14568 Jeff Law <law@redhat.com>
14570 * varasm.c (bss_initializer_p): Do not put constants into .bss
14571 (categorize_decl_for_section): Handle bss_initializer_p returning
14572 false when DECL_INITIAL is NULL.
14574 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14577 * config/s390/s390.c (s390_can_inline_p): New function.
14579 2017-09-01 Jeff Law <law@redhat.com>
14581 PR tree-optimization/82052
14582 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
14583 Always initialize the returned slot after a hash table miss
14584 when INSERT is true.
14586 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
14588 * config/s390/s390.md (mem_signal_fence): Remove.
14589 * doc/md.texi (mem_signal_fence): Remove.
14590 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
14592 * target-insns.def (mem_signal_fence): Remove.
14594 2017-09-01 Jakub Jelinek <jakub@redhat.com>
14597 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
14600 * asan.c (create_odr_indicator): Strip name encoding from assembler
14601 name before appending it after __odr_asan_.
14603 2017-09-01 Martin Liska <mliska@suse.cz>
14605 PR tree-optimization/82059
14606 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
14607 frequency only when an edge is redirected.
14609 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14611 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
14612 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
14613 (arc_conditional_register_usage): Remove ARC600 lp_count
14615 (arc_file_start): Emit Tag_ARC_CPU_variation.
14616 (arc_can_use_doloop_p): New conditions to use ZOLs.
14617 (hwloop_fail): New function.
14618 (hwloop_optimize): Likewise.
14619 (hwloop_pattern_reg): Likewise.
14620 (arc_doloop_hooks): New struct, to be used with reorg_loops.
14621 (arc_reorg_loops): New function, calls reorg_loops.
14622 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
14623 (arc600_corereg_hazard): Remove ZOL checking, case handled by
14625 (arc_loop_hazard): Remove function, functionality moved into
14627 (arc_hazard): Remove arc_loop_hazard call.
14628 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
14629 into hwloop_optimize.
14630 (arc_label_align): Remove ZOL handling.
14631 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
14632 * config/arc/arc.md (doloop_begin): Remove pattern.
14633 (doloop_begin_i): Likewise.
14634 (doloop_end_i): Likewise.
14635 (doloop_fallback): Likewise.
14636 (doloop_fallback_m): Likewise.
14637 (doloop_end): Reimplement expand.
14638 (arc_lp): New pattern for LP instruction.
14639 (loop_end): New pattern.
14640 (loop_fail): Likewise.
14641 (decrement_and_branch_until_zero): Likewise.
14642 * config/arc/arc.opt (mlpc-width): New option.
14643 * doc/invoke.texi (mlpc-width): Document option.
14645 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14647 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
14648 (arc_ccfsm_advance): Fix checking for delay slots.
14649 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
14651 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14653 * config/arc/arc.md (movqi_insn): Add stores to save constant long
14655 (movhi_insn): Update store instruction constraint which are saving
14656 6-bit short immediates.
14657 (movsi_insn): Consider also short scaled load operations.
14658 (zero_extendhisi2_i): Use Usd constraint instead of T.
14659 (extendhisi2_i): Add q constraint.
14660 (arc_clzsi2): Add type and length attributes.
14661 (arc_ctzsi2): Likewise.
14662 * config/arc/constraints.md (Usc): Update constraint, the
14663 assembler can parse two relocations for a single instruction.
14665 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14667 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
14668 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
14670 2017-08-31 Olivier Hainque <hainque@adacore.com>
14672 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
14673 match as powerpc-wrs-vxworks*.
14675 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
14677 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
14678 register constraint for by-element operand.
14679 (aarch64_mls_elt_merge<mode>): Likewise.
14681 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14683 * config/arc/arc.c (arc_can_follow_jump): Check for short
14686 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14688 * config.gcc: Use g.opt for arc.
14689 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
14690 functionality moved to ...
14691 (legitimate_scaled_address_p): New function, ...here.
14692 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
14693 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
14694 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
14696 (arc_override_options): Handle G option.
14697 (arc_output_pic_addr_const): Correct function definition.
14698 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
14699 (arc_decl_anon_ns_mem_p): Delete.
14700 (arc_in_small_data_p): Overhaul this function to take into
14701 consideration the value given via G option.
14702 (arc_rewrite_small_data_1): Renamed and corrected old
14703 arc_rewrite_small_data function.
14704 (arc_rewrite_small_data): New function.
14705 (small_data_pattern): Don't use pic_offset_table_rtx.
14706 * config/arc/arc.h (CC1_SPEC): Recognize G option.
14707 * config/arc/simdext.md (movmisalignv2hi): Use
14708 prepare_move_operands function.
14710 (movmisalign*): Likewise.
14711 * doc/invoke.texi (ARC options): Document -G option.
14713 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14715 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
14717 * config/arc/arc.c (arc_print_operand): Output scalled address for
14718 sdata whenever is possible.
14719 (arc_in_small_data_p): Allow sdata for 64bit datum when double
14720 load/stores are available.
14721 (compact_sda_memory_operand): Check for the alignment required by
14722 code density instructions.
14723 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
14725 * config/arc/constraints.md (Usd): Update constraint.
14726 (Us0): New constraint.
14727 (Usc): Update constraint.
14729 2017-08-31 Richard Biener <rguenther@suse.de>
14731 PR middle-end/82054
14732 * dwarf2out.c (dwarf2out_early_global_decl): Process each
14733 function only once.
14735 2017-08-31 Tamar Christina <tamar.christina@arm.com>
14737 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
14738 Resize type_signature.
14740 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14741 Alan Hayward <alan.hayward@arm.com>
14742 David Sherwood <david.sherwood@arm.com>
14744 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
14745 subregs whose inner modes can be stored in GPRs.
14746 (aarch64_classify_index): Likewise.
14748 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14749 Alan Hayward <alan.hayward@arm.com>
14750 David Sherwood <david.sherwood@arm.com>
14752 * config/aarch64/iterators.md (V_cmp_result): Rename to...
14753 (V_INT_EQUIV): ...this.
14754 (v_cmp_result): Rename to...
14755 (v_int_equiv): ...this.
14756 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
14757 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
14758 (copysign<mode>3): Likewise.
14759 (aarch64_simd_bsl<mode>_internal): Likewise.
14760 (aarch64_simd_bsl<mode>): Likewise.
14761 (vec_cmp<mode><mode>): Likewise.
14762 (vcond<mode><mode>): Likewise.
14763 (vcond<v_cmp_mixed><mode>): Likewise.
14764 (vcondu<mode><v_cmp_mixed>): Likewise.
14765 (aarch64_cm<optab><mode>): Likewise.
14766 (aarch64_cmtst<mode>): Likewise.
14767 (aarch64_fac<optab><mode>): Likewise.
14768 (vec_perm_const<mode>): Likewise.
14769 (vcond_mask_<mode><v_cmp_result>): Rename to...
14770 (vcond_mask_<mode><v_int_equiv>): ...this.
14771 (vec_cmp<mode><v_cmp_result>): Rename to...
14772 (vec_cmp<mode><v_int_equiv>): ...this.
14774 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14775 Alan Hayward <alan.hayward@arm.com>
14776 David Sherwood <david.sherwood@arm.com>
14778 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
14780 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
14781 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
14782 (UNSPEC_LD4_DREG): New unspecs.
14783 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
14784 (aarch64_ld2<mode>_dreg_be): Replace with...
14785 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
14787 (aarch64_ld3<mode>_dreg_le)
14788 (aarch64_ld3<mode>_dreg_be): Replace with...
14789 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
14791 (aarch64_ld4<mode>_dreg_le)
14792 (aarch64_ld4<mode>_dreg_be): Replace with...
14793 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
14796 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14798 PR tree-optimization/81987
14799 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
14800 insert an initializer in a location not dominated by the stride
14803 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
14805 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
14806 on the entire header of the finally block in the fallthru case.
14808 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
14810 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
14812 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
14814 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
14815 rs6000_emit_move_from_cr and call renamed function.
14816 (rs6000_emit_prologue): Call renamed functions.
14817 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
14818 movesi_from_cr, remove volatile CRs.
14820 2017-08-30 Jon Beniston <jon@beniston.com>
14821 Richard Biener <rguenther@suse.de>
14823 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
14824 of VECTOR_MODE_P check.
14825 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
14826 element vector types.
14828 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14830 * df.h (df_read_modify_subreg_p): Remove in favor of...
14831 * rtl.h (read_modify_subreg_p): ...this new function. Take a
14832 const_rtx instead of an rtx.
14833 * cprop.c (local_cprop_find_used_regs): Update accordingly.
14834 * df-problems.c (df_word_lr_mark_ref): Likewise.
14835 * ira-lives.c (mark_pseudo_reg_live): Likewise.
14836 (mark_pseudo_reg_dead): Likewise.
14837 (mark_ref_dead): Likewise.
14838 * reginfo.c (init_subregs_of_mode): Likewise.
14839 * sched-deps.c (sched_analyze_1): Likewise.
14840 * df-scan.c (df_def_record_1): Likewise.
14841 (df_uses_record): Likewise.
14842 (df_read_modify_subreg_p): Remove in favor of...
14843 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
14844 const_rtx instead of an rtx.
14846 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14847 Alan Hayward <alan.hayward@arm.com>
14848 David Sherwood <david.sherwood@arm.com>
14850 * rtl.h (partial_subreg_p): New function.
14851 * caller-save.c (save_call_clobbered_regs): Use it.
14852 * calls.c (expand_call): Likewise.
14853 * combine.c (combinable_i3pat): Likewise.
14854 (simplify_set): Likewise.
14855 (make_extraction): Likewise.
14856 (make_compound_operation_int): Likewise.
14857 (gen_lowpart_or_truncate): Likewise.
14858 (force_to_mode): Likewise.
14859 (make_field_assignment): Likewise.
14860 (reg_truncated_to_mode): Likewise.
14861 (record_truncated_value): Likewise.
14862 (move_deaths): Likewise.
14863 * cse.c (record_jump_cond): Likewise.
14864 (cse_insn): Likewise.
14865 * cselib.c (cselib_lookup_1): Likewise.
14866 * expmed.c (extract_bit_field_using_extv): Likewise.
14867 * function.c (assign_parm_setup_reg): Likewise.
14868 * ifcvt.c (noce_convert_multiple_sets): Likewise.
14869 * ira-build.c (create_insn_allocnos): Likewise.
14870 * lra-coalesce.c (merge_pseudos): Likewise.
14871 * lra-constraints.c (match_reload): Likewise.
14872 (simplify_operand_subreg): Likewise.
14873 (curr_insn_transform): Likewise.
14874 * lra-lives.c (process_bb_lives): Likewise.
14875 * lra.c (new_insn_reg): Likewise.
14876 (lra_substitute_pseudo): Likewise.
14877 * regcprop.c (mode_change_ok): Likewise.
14878 (maybe_mode_change): Likewise.
14879 (copyprop_hardreg_forward_1): Likewise.
14880 * reload.c (push_reload): Likewise.
14881 (find_reloads): Likewise.
14882 (find_reloads_subreg_address): Likewise.
14883 * reload1.c (alter_reg): Likewise.
14884 (eliminate_regs_1): Likewise.
14885 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14887 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
14889 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
14890 back to if statements, including unpack.
14892 2017-08-30 Martin Liska <mliska@suse.cz>
14894 PR inline-asm/82001
14895 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
14897 (func_checker::compare_asm_inputs_outputs): ... this function.
14898 (func_checker::compare_gimple_asm): Use the function to compare
14899 also ASM constrains.
14900 * ipa-icf-gimple.h: Rename the function.
14902 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14903 Alan Hayward <alan.hayward@arm.com>
14904 David Sherwood <david.sherwood@arm.com>
14906 * coretypes.h (complex_mode): New type.
14907 * gdbhooks.py (build_pretty_printer): Handle it.
14908 * machmode.h (complex_mode): New class.
14909 (complex_mode::includes_p): New function.
14910 (is_complex_int_mode): Likewise.
14911 (is_complex_float_mode): Likewise.
14912 * genmodes.c (get_mode_class): Handle complex mode classes.
14913 * function.c (expand_function_end): Use is_complex_int_mode.
14915 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14916 Alan Hayward <alan.hayward@arm.com>
14917 David Sherwood <david.sherwood@arm.com>
14919 * coretypes.h (scalar_mode_pod): New typedef.
14920 * gdbhooks.py (build_pretty_printer): Handle it.
14921 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
14922 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
14923 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
14924 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
14925 as_a <scalar_mode>.
14927 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14928 Alan Hayward <alan.hayward@arm.com>
14929 David Sherwood <david.sherwood@arm.com>
14931 * machmode.h (mode_for_vector): Take a scalar_mode instead
14933 * stor-layout.c (mode_for_vector): Likewise.
14934 * explow.c (promote_mode): Use as_a <scalar_mode>.
14935 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
14937 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14938 Alan Hayward <alan.hayward@arm.com>
14939 David Sherwood <david.sherwood@arm.com>
14941 * target.def (preferred_simd_mode): Take a scalar_mode
14942 instead of a machine_mode.
14943 * targhooks.h (default_preferred_simd_mode): Likewise.
14944 * targhooks.c (default_preferred_simd_mode): Likewise.
14945 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
14946 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
14947 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
14948 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
14949 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
14950 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
14951 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
14952 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
14954 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
14955 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
14956 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
14957 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
14958 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
14959 * doc/tm.texi: Regenerate.
14960 * optabs-query.c (can_vec_mask_load_store_p): Return false for
14963 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14964 Alan Hayward <alan.hayward@arm.com>
14965 David Sherwood <david.sherwood@arm.com>
14967 * target.def (scalar_mode_supported_p): Take a scalar_mode
14968 instead of a machine_mode.
14969 * targhooks.h (default_scalar_mode_supported_p): Likewise.
14970 * targhooks.c (default_scalar_mode_supported_p): Likewise.
14971 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
14972 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
14973 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
14974 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
14975 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
14976 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
14977 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
14978 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
14979 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
14980 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
14981 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
14982 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
14984 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
14985 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
14986 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
14987 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
14988 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
14990 * doc/tm.texi: Regenerate.
14992 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14993 Alan Hayward <alan.hayward@arm.com>
14994 David Sherwood <david.sherwood@arm.com>
14996 * coretypes.h (opt_scalar_mode): New typedef.
14997 * gdbhooks.py (build_pretty_printers): Handle it.
14998 * machmode.h (mode_iterator::get_2xwider): Add overload for
15000 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
15002 * expr.c (convert_mode_scalar): Likewise.
15003 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15004 * optabs.c (expand_float): Likewise.
15005 (expand_fix): Likewise.
15006 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15008 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15009 Alan Hayward <alan.hayward@arm.com>
15010 David Sherwood <david.sherwood@arm.com>
15012 * optabs.c (expand_float): Explicitly check for scalars before
15013 using a branching expansion.
15014 (expand_fix): Likewise.
15016 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15017 Alan Hayward <alan.hayward@arm.com>
15018 David Sherwood <david.sherwood@arm.com>
15020 * expr.c (convert_mode): Split scalar handling out into...
15021 (convert_mode_scalar): ...this new function. Treat the modes
15024 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15025 Alan Hayward <alan.hayward@arm.com>
15026 David Sherwood <david.sherwood@arm.com>
15028 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
15030 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
15032 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15033 Alan Hayward <alan.hayward@arm.com>
15034 David Sherwood <david.sherwood@arm.com>
15036 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
15037 rather than a machine_mode.
15038 (fixed_from_string): Likewise.
15039 (fixed_convert): Likewise.
15040 (fixed_convert_from_int): Likewise.
15041 (fixed_convert_from_real): Likewise.
15042 (real_convert_from_fixed): Likewise.
15043 * fixed-value.c (fixed_from_double_int): Likewise.
15044 (fixed_from_string): Likewise.
15045 (fixed_convert): Likewise.
15046 (fixed_convert_from_int): Likewise.
15047 (fixed_convert_from_real): Likewise.
15048 (real_convert_from_fixed): Likewise.
15049 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
15051 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15052 Alan Hayward <alan.hayward@arm.com>
15053 David Sherwood <david.sherwood@arm.com>
15055 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
15056 of separate mode class checks. Do not allow vector modes here.
15057 (immed_wide_int_const): Use as_a <scalar_mode>.
15058 * explow.c (trunc_int_for_mode): Likewise.
15059 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
15060 (wi::shwi): Likewise.
15061 (wi::min_value): Likewise.
15062 (wi::max_value): Likewise.
15063 * dwarf2out.c (loc_descriptor): Likewise.
15064 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
15065 for CONST_WIDE_INT.
15067 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15068 Alan Hayward <alan.hayward@arm.com>
15069 David Sherwood <david.sherwood@arm.com>
15071 * tree.h (SCALAR_TYPE_MODE): New macro.
15072 * expr.c (expand_expr_addr_expr_1): Use it.
15073 (expand_expr_real_2): Likewise.
15074 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
15075 (fold_convert_const_fixed_from_int): Likewise.
15076 (fold_convert_const_fixed_from_real): Likewise.
15077 (native_encode_fixed): Likewise
15078 (native_encode_complex): Likewise
15079 (native_encode_vector): Likewise.
15080 (native_interpret_fixed): Likewise.
15081 (native_interpret_real): Likewise.
15082 (native_interpret_complex): Likewise.
15083 (native_interpret_vector): Likewise.
15084 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
15085 (simd_clone_adjust_argument_types): Likewise.
15086 (simd_clone_init_simd_arrays): Likewise.
15087 (simd_clone_adjust): Likewise.
15088 * stor-layout.c (layout_type): Likewise.
15089 * tree.c (build_minus_one_cst): Likewise.
15090 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15091 * tree-inline.c (estimate_move_cost): Likewise.
15092 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
15093 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
15094 (vectorizable_reduction): Likewise.
15095 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
15096 (vect_recog_mixed_size_cond_pattern): Likewise.
15097 (check_bool_pattern): Likewise.
15098 (adjust_bool_pattern): Likewise.
15099 (search_type_for_mask_1): Likewise.
15100 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
15101 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15102 (vectorizable_load): Likewise.
15103 (vectorizable_store): Likewise.
15104 * ubsan.c (ubsan_encode_value): Likewise.
15105 * varasm.c (output_constant): Likewise.
15107 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15108 Alan Hayward <alan.hayward@arm.com>
15109 David Sherwood <david.sherwood@arm.com>
15111 * coretypes.h (scalar_mode): New class.
15112 * machmode.h (scalar_mode): Likewise.
15113 (scalar_mode::includes_p): New function.
15114 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
15115 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
15116 * genmodes.c (get_mode_class): Handle remaining scalar modes.
15117 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
15118 * expmed.c (store_bit_field_1): Likewise.
15119 (extract_bit_field_1): Likewise.
15120 * expr.c (write_complex_part): Likewise.
15121 (read_complex_part): Likewise.
15122 (emit_move_complex_push): Likewise.
15123 (expand_expr_real_2): Likewise.
15124 * function.c (assign_parm_setup_reg): Likewise.
15125 (assign_parms_unsplit_complex): Likewise.
15126 * optabs.c (expand_binop): Likewise.
15127 * rtlanal.c (subreg_get_info): Likewise.
15128 * simplify-rtx.c (simplify_immed_subreg): Likewise.
15129 * varasm.c (output_constant_pool_2): Likewise.
15131 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15132 Alan Hayward <alan.hayward@arm.com>
15133 David Sherwood <david.sherwood@arm.com>
15135 * expmed.c (extract_high_half): Use scalar_int_mode and remove
15137 (expmed_mult_highpart_optab): Likewise.
15138 (expmed_mult_highpart): Likewise.
15140 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15141 Alan Hayward <alan.hayward@arm.com>
15142 David Sherwood <david.sherwood@arm.com>
15144 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
15145 instead of a machine_mode.
15146 (builtin_memset_read_str): Likewise.
15147 * builtins.c (c_readstr): Likewise.
15148 (builtin_memcpy_read_str): Likewise.
15149 (builtin_strncpy_read_str): Likewise.
15150 (builtin_memset_read_str): Likewise.
15151 (builtin_memset_gen_str): Likewise.
15152 (expand_builtin_signbit): Use scalar_int_mode for local variables.
15153 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
15154 instead of a machine_mode.
15155 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
15157 (make_extraction): Likewise.
15158 (try_widen_shift_mode): Take and return scalar_int_modes instead
15160 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
15161 a scalar_int_mode instead of a machine_mode.
15162 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
15163 (avr_addr_space_pointer_mode): Likewise.
15164 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
15165 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
15166 (msp430_unwind_word_mode): Likewise.
15167 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
15168 (spu_addr_space_pointer_mode): Likewise.
15169 (spu_addr_space_address_mode): Likewise.
15170 (spu_libgcc_cmp_return_mode): Likewise.
15171 (spu_libgcc_shift_count_mode): Likewise.
15172 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
15173 (rl78_addr_space_pointer_mode): Likewise.
15174 (fl78_unwind_word_mode): Likewise.
15175 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
15177 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
15178 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
15179 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
15180 (mips_valid_pointer_mode): Likewise.
15181 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
15182 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
15183 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
15185 (ft32_addr_space_address_mode): Likewise.
15186 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
15187 scalar_int_mode instead of a machine_mode.
15188 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
15190 (m32c_addr_space_address_mode): Likewise.
15191 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
15192 (rs6000_eh_return_filter_mode): Likewise.
15193 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
15194 (rs6000_eh_return_filter_mode): Likewise.
15195 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
15196 (s390_libgcc_shift_count_mode): Likewise.
15197 (s390_unwind_word_mode): Likewise.
15198 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
15200 * target.def (mode_rep_extended): Likewise.
15201 (valid_pointer_mode): Likewise.
15202 (addr_space.valid_pointer_mode): Likewise.
15203 (eh_return_filter_mode): Return a scalar_int_mode rather than
15205 (libgcc_cmp_return_mode): Likewise.
15206 (libgcc_shift_count_mode): Likewise.
15207 (unwind_word_mode): Likewise.
15208 (addr_space.pointer_mode): Likewise.
15209 (addr_space.address_mode): Likewise.
15210 * doc/tm.texi: Regenerate.
15211 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
15213 (do_jump): Use scalar_int_mode for local variables.
15214 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
15215 rather than a machine_mode.
15216 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
15217 (scompare_loc_descriptor_wide): Likewise.
15218 (scompare_loc_descriptor_narrow): Likewise.
15219 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
15221 * except.c (sjlj_emit_dispatch_table): Likewise.
15222 (expand_builtin_eh_copy_values): Likewise.
15223 * explow.c (convert_memory_address_addr_space_1): Likewise.
15224 Take a scalar_int_mode rather than a machine_mode.
15225 (convert_memory_address_addr_space): Take a scalar_int_mode rather
15226 than a machine_mode.
15227 (memory_address_addr_space): Use scalar_int_mode for local variables.
15228 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
15229 rather than a machine_mode.
15230 * expmed.c (mask_rtx): Likewise.
15231 (init_expmed_one_conv): Likewise.
15232 (expand_mult_highpart_adjust): Likewise.
15233 (extract_high_half): Likewise.
15234 (expmed_mult_highpart_optab): Likewise.
15235 (expmed_mult_highpart): Likewise.
15236 (expand_smod_pow2): Likewise.
15237 (expand_sdiv_pow2): Likewise.
15238 (emit_store_flag_int): Likewise.
15239 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
15241 (extract_low_bits): Likewise.
15242 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
15244 * expr.c (pieces_addr::adjust): Likewise.
15245 (can_store_by_pieces): Likewise.
15246 (store_by_pieces): Likewise.
15247 (clear_by_pieces_1): Likewise.
15248 (expand_expr_addr_expr_1): Likewise.
15249 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
15250 (expand_expr_real_1): Likewise.
15251 (try_casesi): Likewise.
15252 * final.c (shorten_branches): Likewise.
15253 * fold-const.c (fold_convert_const_int_from_fixed): Change the
15254 type of "mode" to machine_mode.
15255 * internal-fn.c (expand_arith_overflow_result_store): Take a
15256 scalar_int_mode rather than a machine_mode.
15257 (expand_mul_overflow): Use scalar_int_mode for local variables.
15258 * loop-doloop.c (doloop_modify): Likewise.
15259 (doloop_optimize): Likewise.
15260 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
15261 than a machine_mode.
15262 (expand_doubleword_shift_condmove): Likewise.
15263 (expand_doubleword_shift): Likewise.
15264 (expand_doubleword_clz): Likewise.
15265 (expand_doubleword_popcount): Likewise.
15266 (expand_doubleword_parity): Likewise.
15267 (expand_absneg_bit): Use scalar_int_mode for local variables.
15268 (prepare_float_lib_cmp): Likewise.
15269 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
15270 rather than a machine_mode.
15271 (convert_memory_address_addr_space): Likewise.
15272 (get_mode_bounds): Likewise.
15273 (get_address_mode): Return a scalar_int_mode rather than a
15275 * rtlanal.c (get_address_mode): Likewise.
15276 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
15277 than a machine_mode.
15278 * targhooks.c (default_mode_rep_extended): Likewise.
15279 (default_valid_pointer_mode): Likewise.
15280 (default_addr_space_valid_pointer_mode): Likewise.
15281 (default_eh_return_filter_mode): Return a scalar_int_mode rather
15282 than a machine_mode.
15283 (default_libgcc_cmp_return_mode): Likewise.
15284 (default_libgcc_shift_count_mode): Likewise.
15285 (default_unwind_word_mode): Likewise.
15286 (default_addr_space_pointer_mode): Likewise.
15287 (default_addr_space_address_mode): Likewise.
15288 * targhooks.h (default_eh_return_filter_mode): Likewise.
15289 (default_libgcc_cmp_return_mode): Likewise.
15290 (default_libgcc_shift_count_mode): Likewise.
15291 (default_unwind_word_mode): Likewise.
15292 (default_addr_space_pointer_mode): Likewise.
15293 (default_addr_space_address_mode): Likewise.
15294 (default_mode_rep_extended): Take a scalar_int_mode rather than
15296 (default_valid_pointer_mode): Likewise.
15297 (default_addr_space_valid_pointer_mode): Likewise.
15298 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
15300 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
15301 rather than a machine_mode.
15302 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
15303 for local variables.
15304 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15305 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
15306 than a machine_mode.
15308 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15309 Alan Hayward <alan.hayward@arm.com>
15310 David Sherwood <david.sherwood@arm.com>
15312 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
15313 the mode argument to scalar_int_mode.
15314 (do_jump_by_parts_zero_rtx): Likewise.
15315 (do_jump_by_parts_equality_rtx): Likewise.
15316 (do_jump_by_parts_greater): Take a mode argument.
15317 (do_jump_by_parts_equality): Likewise.
15318 (do_jump_1): Update calls accordingly.
15320 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15321 Alan Hayward <alan.hayward@arm.com>
15322 David Sherwood <david.sherwood@arm.com>
15324 * is-a.h (safe_dyn_cast): New function.
15325 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
15326 (jump_table_for_label): Likewise.
15327 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
15328 instead of an rtx_insn *.
15329 (shorten_branches): Use dyn_cast instead of LABEL_P and
15330 JUMP_TABLE_DATA_P. Use jump_table_for_label and
15331 rtx_jump_table_data::get_data_mode.
15332 (final_scan_insn): Likewise.
15334 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15335 Alan Hayward <alan.hayward@arm.com>
15336 David Sherwood <david.sherwood@arm.com>
15338 * combine.c (try_combine): Use is_a <scalar_int_mode> when
15339 trying to combine a full-register integer set with a subreg
15342 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15343 Alan Hayward <alan.hayward@arm.com>
15344 David Sherwood <david.sherwood@arm.com>
15346 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
15347 that is always either address_mode or pointer_mode.
15349 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15350 Alan Hayward <alan.hayward@arm.com>
15351 David Sherwood <david.sherwood@arm.com>
15353 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
15354 when the two are known to be equal.
15356 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15357 Alan Hayward <alan.hayward@arm.com>
15358 David Sherwood <david.sherwood@arm.com>
15360 * simplify-rtx.c (simplify_const_unary_operation): Use
15361 is_a <scalar_int_mode> instead of checking for a nonzero
15362 precision. Forcibly convert op_mode to a scalar_int_mode
15363 in that case. More clearly differentiate the operand and
15364 result modes and use the former when deciding what the value
15365 of a count-bits operation should be. Use is_int_mode instead
15366 of checking for a MODE_INT. Remove redundant check for whether
15367 this mode has a zero precision.
15369 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15370 Alan Hayward <alan.hayward@arm.com>
15371 David Sherwood <david.sherwood@arm.com>
15373 * optabs.c (widen_leading): Change the type of the mode argument
15374 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
15375 (widen_bswap): Likewise.
15376 (expand_parity): Likewise.
15377 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
15378 (expand_ffs): Likewise.
15379 (epand_unop): Check for scalar integer modes before calling the
15382 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15383 Alan Hayward <alan.hayward@arm.com>
15384 David Sherwood <david.sherwood@arm.com>
15386 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
15388 (expand_expr_real_1): Update call accordingly.
15390 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15391 Alan Hayward <alan.hayward@arm.com>
15392 David Sherwood <david.sherwood@arm.com>
15394 * expmed.c (store_bit_field_using_insv): Add op0_mode and
15395 value_mode arguments. Use scalar_int_mode internally.
15396 (store_bit_field_1): Rename the new integer mode from imode
15397 to op0_mode and use it instead of GET_MODE (op0). Update calls
15398 to store_split_bit_field, store_bit_field_using_insv and
15399 store_fixed_bit_field.
15400 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
15401 Use scalar_int_mode internally. Use a bit count rather than a mode
15402 when calculating the largest bit size for get_best_mode.
15403 Update calls to store_split_bit_field and store_fixed_bit_field_1.
15404 (store_fixed_bit_field_1): Add mode and value_mode arguments.
15405 Remove assertion that OP0 has a scalar integer mode.
15406 (store_split_bit_field): Add op0_mode and value_mode arguments.
15407 Update calls to extract_fixed_bit_field.
15408 (extract_bit_field_using_extv): Add an op0_mode argument.
15409 Use scalar_int_mode internally.
15410 (extract_bit_field_1): Rename the new integer mode from imode to
15411 op0_mode and use it instead of GET_MODE (op0). Update calls to
15412 extract_split_bit_field, extract_bit_field_using_extv and
15413 extract_fixed_bit_field.
15414 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
15415 to extract_split_bit_field and extract_fixed_bit_field_1.
15416 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
15417 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
15418 on the target mode.
15419 (extract_split_bit_field): Add an op0_mode argument. Update call
15420 to extract_fixed_bit_field.
15422 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15423 Alan Hayward <alan.hayward@arm.com>
15424 David Sherwood <david.sherwood@arm.com>
15426 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
15427 * explow.c (hard_function_value): Likewise.
15428 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
15429 convert_to_mode call outside the loop.
15430 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
15431 for the mode iterator. Require the mode specified by max_pieces
15433 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
15435 (copy_blkmode_to_reg): Likewise.
15436 (set_storage_via_setmem): Likewise.
15437 * optabs.c (prepare_cmp_insn): Likewise.
15438 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15439 * stor-layout.c (finish_bitfield_representative): Likewise.
15441 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15442 Alan Hayward <alan.hayward@arm.com>
15443 David Sherwood <david.sherwood@arm.com>
15445 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
15446 * expr.c (convert_move): Use them.
15447 (convert_modes): Likewise.
15448 (store_expr_with_bounds): Likewise.
15450 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15451 Alan Hayward <alan.hayward@arm.com>
15452 David Sherwood <david.sherwood@arm.com>
15454 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
15455 parameter for the mode of "x". Remove the "known_x", "known_mode"
15456 and "known_ret" arguments. Change the type of the mode argument
15457 to scalar_int_mode.
15458 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
15459 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
15460 (reg_num_sign_bit_copies_for_combine): Likewise.
15461 * rtlanal.c (nonzero_bits1): Likewise.
15462 (num_sign_bit_copies1): Likewise.
15463 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
15464 (reg_num_sign_bit_copies_general): Likewise.
15465 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
15466 (reg_nonzero_bits_general): Likewise.
15468 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15469 Alan Hayward <alan.hayward@arm.com>
15470 David Sherwood <david.sherwood@arm.com>
15472 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
15473 than in subroutines. Return 1 for non-integer modes.
15474 (cached_num_sign_bit_copies): Change the type of the mode parameter
15475 to scalar_int_mode.
15476 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
15477 classes. Handle CONST_INT_P first and then check whether X also
15478 has a scalar integer mode. Check the same thing for inner registers
15479 of a SUBREG and for values that are being extended or truncated.
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 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
15486 in subroutines. Return the mode mask for non-integer modes.
15487 (cached_nonzero_bits): Change the type of the mode parameter
15488 to scalar_int_mode.
15489 (nonzero_bits1): Likewise. Remove early exit for other mode
15490 classes. Handle CONST_INT_P first and then check whether X
15491 also has a scalar integer mode.
15493 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15494 Alan Hayward <alan.hayward@arm.com>
15495 David Sherwood <david.sherwood@arm.com>
15497 * expr.c (widest_int_mode_for_size): Make the comment match the code.
15498 Return a scalar_int_mode and assert that the size is greater than
15500 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
15501 (op_by_pieces_d::op_by_pieces_d): Likewise.
15502 (op_by_pieces_d::run): Likewise.
15503 (can_store_by_pieces): Likewise.
15505 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15506 Alan Hayward <alan.hayward@arm.com>
15507 David Sherwood <david.sherwood@arm.com>
15509 * combine.c (extract_left_shift): Add a mode argument and update
15511 (make_compound_operation_int): Change the type of the mode parameter
15512 to scalar_int_mode and update the call to extract_left_shift.
15514 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15515 Alan Hayward <alan.hayward@arm.com>
15516 David Sherwood <david.sherwood@arm.com>
15518 * combine.c (simplify_and_const_int): Change the type of the mode
15519 parameter to scalar_int_mode.
15520 (simplify_and_const_int_1): Likewise. Update recursive call.
15522 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15523 Alan Hayward <alan.hayward@arm.com>
15524 David Sherwood <david.sherwood@arm.com>
15526 * combine.c (simplify_compare_const): Check that the mode is a
15527 scalar_int_mode (rather than VOIDmode) before testing its
15529 (simplify_comparison): Move COMPARISON_P handling out of the
15530 loop and restrict the latter part of the loop to scalar_int_modes.
15531 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
15532 and when considering SUBREG_REGs. Use is_int_mode instead of
15533 checking GET_MODE_CLASS against MODE_INT.
15535 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15536 Alan Hayward <alan.hayward@arm.com>
15537 David Sherwood <david.sherwood@arm.com>
15539 * combine.c (try_widen_shift_mode): Move check for equal modes to...
15540 (simplify_shift_const_1): ...here. Use scalar_int_mode for
15541 shift_unit_mode and for modes involved in scalar shifts.
15543 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15544 Alan Hayward <alan.hayward@arm.com>
15545 David Sherwood <david.sherwood@arm.com>
15547 * combine.c (force_int_to_mode): New function, split out from...
15548 (force_to_mode): ...here. Keep xmode up-to-date and use it
15549 instead of GET_MODE (x).
15551 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15552 Alan Hayward <alan.hayward@arm.com>
15553 David Sherwood <david.sherwood@arm.com>
15555 * optabs-query.h (extraction_insn::struct_mode): Change type to
15556 opt_scalar_int_mode and update comment.
15557 (extraction_insn::field_mode): Change type to scalar_int_mode.
15558 (extraction_insn::pos_mode): Likewise.
15559 * combine.c (make_extraction): Update accordingly.
15560 * optabs-query.c (get_traditional_extraction_insn): Likewise.
15561 (get_optab_extraction_insn): Likewise.
15562 * recog.c (simplify_while_replacing): Likewise.
15563 * expmed.c (narrow_bit_field_mem): Change the type of the mode
15564 parameter to opt_scalar_int_mode.
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 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
15571 to a scalar_int_mode instead of a machine_mode.
15572 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
15573 (get_best_mode): Return a boolean and use a pointer argument to store
15574 the selected mode. Replace the limit mode parameter with a bit limit.
15575 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
15576 for the values returned by bit_field_mode_iterator::next_mode.
15577 (store_bit_field): Update call to get_best_mode.
15578 (store_fixed_bit_field): Likewise.
15579 (extract_fixed_bit_field): Likewise.
15580 * expr.c (optimize_bitfield_assignment_op): Likewise.
15581 * fold-const.c (optimize_bit_field_compare): Likewise.
15582 (fold_truth_andor_1): Likewise.
15583 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
15584 Update for new type of m_mode.
15585 (get_best_mode): As above.
15587 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15588 Alan Hayward <alan.hayward@arm.com>
15589 David Sherwood <david.sherwood@arm.com>
15591 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
15592 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
15593 (store_bit_field): Check is_a <scalar_int_mode> before calling
15594 strict_volatile_bitfield_p.
15595 (extract_bit_field): Likewise.
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 * target.def (cstore_mode): Return a scalar_int_mode.
15602 * doc/tm.texi: Regenerate.
15603 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
15604 * targhooks.h (default_cstore_mode): Likewise.
15605 * targhooks.c (default_cstore_mode): Likewise, using a forced
15607 * expmed.c (emit_cstore): Expect the target of the cstore to be
15610 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15611 Alan Hayward <alan.hayward@arm.com>
15612 David Sherwood <david.sherwood@arm.com>
15614 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
15616 (niter_desc): Likewise mode.
15617 (iv_analyze): Add a mode parameter.
15619 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
15620 and change its type to scalar_int_mode.
15621 * loop-iv.c: Update commentary at head of file.
15622 (iv_constant): Pass the mode paraeter before the rtx it describes
15623 and change its type to scalar_int_mode. Remove VOIDmode handling.
15624 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
15625 (iv_extend): Likewise.
15626 (shorten_into_mode): Likewise.
15627 (iv_add): Use scalar_int_mode.
15628 (iv_mult): Likewise.
15629 (iv_shift): Likewise.
15630 (canonicalize_iv_subregs): Likewise.
15631 (get_biv_step_1): Pass the outer_mode parameter before the rtx
15632 it describes and change its mode to scalar_int_mode. Also change
15633 the type of the returned inner_mode to scalar_int_mode.
15634 (get_biv_step): Likewise, turning outer_mode from a pointer
15635 into a direct parameter. Update call to get_biv_step_1.
15636 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
15637 iv_constant and get_biv_step.
15638 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
15639 and change its type to scalar_int_mode. Don't initialise iv->mode
15640 to VOIDmode and remove later checks for its still being VOIDmode.
15641 Update calls to iv_analyze_op and iv_analyze_expr. Check
15642 is_a <scalar_int_mode> when changing the mode under consideration.
15643 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
15644 Update call to iv_analyze_expr.
15645 (iv_analyze_op): Add a mode parameter. Reject subregs whose
15646 inner register is not also a scalar_int_mode. Update call to
15648 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
15649 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
15650 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
15651 separate mode class checks. Update calls to iv_analyze. Remove
15652 fix-up of VOIDmodes after iv_analyze_biv.
15653 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
15654 don't have a scalar_int_mode. Update call to biv_p.
15656 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15657 Alan Hayward <alan.hayward@arm.com>
15658 David Sherwood <david.sherwood@arm.com>
15660 * cfgexpand.c (convert_debug_memory_address): Use
15661 as_a <scalar_int_mode>.
15662 * combine.c (expand_compound_operation): Likewise.
15663 (make_extraction): Likewise.
15664 (change_zero_ext): Likewise.
15665 (simplify_comparison): Likewise.
15666 * cse.c (cse_insn): Likewise.
15667 * dwarf2out.c (minmax_loc_descriptor): Likewise.
15668 (mem_loc_descriptor): Likewise.
15669 (loc_descriptor): Likewise.
15670 * expmed.c (init_expmed_one_mode): Likewise.
15671 (synth_mult): Likewise.
15672 (emit_store_flag_1): Likewise.
15673 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
15674 of a comparison with size.
15675 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
15676 (reduce_to_bit_field_precision): Likewise.
15677 * function.c (expand_function_end): Likewise.
15678 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
15679 * loop-doloop.c (doloop_modify): Likewise.
15680 * optabs.c (expand_binop): Likewise.
15681 (expand_unop): Likewise.
15682 (expand_copysign_absneg): Likewise.
15683 (prepare_cmp_insn): Likewise.
15684 (maybe_legitimize_operand): Likewise.
15685 * recog.c (const_scalar_int_operand): Likewise.
15686 * rtlanal.c (get_address_mode): Likewise.
15687 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15688 (simplify_cond_clz_ctz): Likewise.
15689 * tree-nested.c (get_nl_goto_field): Likewise.
15690 * tree.c (build_vector_type_for_mode): Likewise.
15691 * var-tracking.c (use_narrower_mode): Likewise.
15693 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15694 Alan Hayward <alan.hayward@arm.com>
15695 David Sherwood <david.sherwood@arm.com>
15697 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
15698 * builtins.c (expand_builtin_signbit): Use it.
15699 * cfgexpand.c (expand_debug_expr): Likewise.
15700 * dojump.c (do_jump): Likewise.
15701 (do_compare_and_jump): Likewise.
15702 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
15703 * expmed.c (make_tree): Likewise.
15704 * expr.c (expand_expr_real_2): Likewise.
15705 (expand_expr_real_1): Likewise.
15706 (try_casesi): Likewise.
15707 * fold-const-call.c (fold_const_call_ss): Likewise.
15708 * fold-const.c (unextend): Likewise.
15709 (extract_muldiv_1): Likewise.
15710 (fold_single_bit_test): Likewise.
15711 (native_encode_int): Likewise.
15712 (native_encode_string): Likewise.
15713 (native_interpret_int): Likewise.
15714 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
15715 * internal-fn.c (expand_addsub_overflow): Likewise.
15716 (expand_neg_overflow): Likewise.
15717 (expand_mul_overflow): Likewise.
15718 (expand_arith_overflow): Likewise.
15719 * match.pd: Likewise.
15720 * stor-layout.c (layout_type): Likewise.
15721 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15722 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
15723 * tree-ssanames.c (get_range_info): Likewise.
15724 * tree-switch-conversion.c (array_value_type) Likewise.
15725 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
15726 (vect_recog_divmod_pattern): Likewise.
15727 (vect_recog_mixed_size_cond_pattern): Likewise.
15728 * tree-vrp.c (extract_range_basic): Likewise.
15729 (simplify_float_conversion_using_ranges): Likewise.
15730 * tree.c (int_fits_type_p): Likewise.
15731 * ubsan.c (instrument_bool_enum_load): Likewise.
15732 * varasm.c (mergeable_string_section): Likewise.
15733 (narrowing_initializer_constant_valid_p): Likewise.
15734 (output_constant): Likewise.
15736 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15737 Alan Hayward <alan.hayward@arm.com>
15738 David Sherwood <david.sherwood@arm.com>
15740 * machmode.h (NARROWEST_INT_MODE): New macro.
15741 * expr.c (alignment_for_piecewise_move): Use it instead of
15742 GET_CLASS_NARROWEST_MODE (MODE_INT).
15743 (push_block): Likewise.
15744 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
15746 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15748 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15749 Alan Hayward <alan.hayward@arm.com>
15750 David Sherwood <david.sherwood@arm.com>
15752 * postreload.c (move2add_valid_value_p): Change the type of the
15753 mode parameter to scalar_int_mode.
15754 (move2add_use_add2_insn): Add a mode parameter and use it instead
15756 (move2add_use_add3_insn): Likewise.
15757 (reload_cse_move2add): Update accordingly.
15759 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15760 Alan Hayward <alan.hayward@arm.com>
15761 David Sherwood <david.sherwood@arm.com>
15763 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
15765 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
15766 * optabs.c (expand_unop): Likewise.
15768 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15769 Alan Hayward <alan.hayward@arm.com>
15770 David Sherwood <david.sherwood@arm.com>
15772 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
15773 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
15774 (popcount_loc_descriptor): Likewise.
15775 (bswap_loc_descriptor): Likewise.
15776 (rotate_loc_descriptor): Likewise.
15777 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
15778 calling the functions above.
15780 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15781 Alan Hayward <alan.hayward@arm.com>
15782 David Sherwood <david.sherwood@arm.com>
15784 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
15786 (try_combine): Likewise.
15787 (simplify_if_then_else): Likewise.
15788 * cse.c (cse_insn): Likewise.
15789 * dwarf2out.c (mem_loc_descriptor): Likewise.
15790 * emit-rtl.c (gen_lowpart_common): Likewise.
15791 * simplify-rtx.c (simplify_truncation): Likewise.
15792 (simplify_binary_operation_1): Likewise.
15793 (simplify_const_relational_operation): Likewise.
15794 (simplify_ternary_operation): Likewise.
15795 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
15797 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15798 Alan Hayward <alan.hayward@arm.com>
15799 David Sherwood <david.sherwood@arm.com>
15801 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
15802 * reload.c (push_reload): Likewise.
15803 (find_reloads): Likewise.
15805 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15806 Alan Hayward <alan.hayward@arm.com>
15807 David Sherwood <david.sherwood@arm.com>
15809 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
15810 (make_compound_operation_int): Likewise.
15811 (change_zero_ext): Likewise.
15812 * expr.c (convert_move): Likewise.
15813 (convert_modes): Likewise.
15814 * fwprop.c (forward_propagate_subreg): Likewise.
15815 * loop-iv.c (get_biv_step_1): Likewise.
15816 * optabs.c (widen_operand): Likewise.
15817 * postreload.c (move2add_valid_value_p): Likewise.
15818 * recog.c (simplify_while_replacing): Likewise.
15819 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15820 (simplify_binary_operation_1): Likewise. Remove redundant
15821 mode equality check.
15823 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15824 Alan Hayward <alan.hayward@arm.com>
15825 David Sherwood <david.sherwood@arm.com>
15827 * combine.c (combine_simplify_rtx): Add checks for
15828 is_a <scalar_int_mode>.
15829 (simplify_if_then_else): Likewise.
15830 (make_field_assignment): Likewise.
15831 (simplify_comparison): Likewise.
15832 * ifcvt.c (noce_try_bitop): Likewise.
15833 * loop-invariant.c (canonicalize_address_mult): Likewise.
15834 * simplify-rtx.c (simplify_unary_operation_1): 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 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
15841 is_a <scalar_int_mode> instead of != BLKmode.
15843 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15844 Alan Hayward <alan.hayward@arm.com>
15845 David Sherwood <david.sherwood@arm.com>
15847 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
15848 instead of != VOIDmode.
15849 * combine.c (if_then_else_cond): Likewise.
15850 (change_zero_ext): Likewise.
15851 * dwarf2out.c (mem_loc_descriptor): Likewise.
15852 (loc_descriptor): Likewise.
15853 * rtlanal.c (canonicalize_condition): Likewise.
15854 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
15856 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15857 Alan Hayward <alan.hayward@arm.com>
15858 David Sherwood <david.sherwood@arm.com>
15860 * simplify-rtx.c (simplify_binary_operation_1): Use
15861 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
15863 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15864 Alan Hayward <alan.hayward@arm.com>
15865 David Sherwood <david.sherwood@arm.com>
15867 * wide-int.h (int_traits<unsigned char>) New class.
15868 (int_traits<unsigned short>) Likewise.
15869 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
15870 Use GET_MODE_UNIT_PRECISION and remove redundant test for
15872 * combine.c (set_nonzero_bits_and_sign_copies): Use
15873 is_a <scalar_int_mode>.
15874 (find_split_point): Likewise.
15875 (combine_simplify_rtx): Likewise.
15876 (simplify_logical): Likewise.
15877 (expand_compound_operation): Likewise.
15878 (expand_field_assignment): Likewise.
15879 (make_compound_operation): Likewise.
15880 (extended_count): Likewise.
15881 (change_zero_ext): Likewise.
15882 (simplify_comparison): Likewise.
15883 * dwarf2out.c (scompare_loc_descriptor): Likewise.
15884 (ucompare_loc_descriptor): Likewise.
15885 (minmax_loc_descriptor): Likewise.
15886 (mem_loc_descriptor): Likewise.
15887 (loc_descriptor): Likewise.
15888 * expmed.c (init_expmed_one_mode): Likewise.
15889 * lra-constraints.c (lra_constraint_offset): Likewise.
15890 * optabs.c (prepare_libcall_arg): Likewise.
15891 * postreload.c (move2add_note_store): Likewise.
15892 * reload.c (operands_match_p): Likewise.
15893 * rtl.h (load_extend_op): Likewise.
15894 * rtlhooks.c (gen_lowpart_general): Likewise.
15895 * simplify-rtx.c (simplify_truncation): Likewise.
15896 (simplify_unary_operation_1): Likewise.
15897 (simplify_binary_operation_1): Likewise.
15898 (simplify_const_binary_operation): Likewise.
15899 (simplify_const_relational_operation): Likewise.
15900 (simplify_subreg): Likewise.
15901 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15902 * var-tracking.c (adjust_mems): Likewise.
15903 (prepare_call_arguments): Likewise.
15905 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15906 Alan Hayward <alan.hayward@arm.com>
15907 David Sherwood <david.sherwood@arm.com>
15909 * machmode.h (is_int_mode): New fuction.
15910 * combine.c (find_split_point): Use it.
15911 (combine_simplify_rtx): Likewise.
15912 (simplify_if_then_else): Likewise.
15913 (simplify_set): Likewise.
15914 (simplify_shift_const_1): Likewise.
15915 (simplify_comparison): Likewise.
15916 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
15917 * cse.c (notreg_cost): Likewise.
15918 (cse_insn): Likewise.
15919 * cselib.c (cselib_lookup_1): Likewise.
15920 * dojump.c (do_jump_1): Likewise.
15921 (do_compare_rtx_and_jump): Likewise.
15922 * dse.c (get_call_args): Likewise.
15923 * dwarf2out.c (rtl_for_decl_init): Likewise.
15924 (native_encode_initializer): Likewise.
15925 * expmed.c (emit_store_flag_1): Likewise.
15926 (emit_store_flag): Likewise.
15927 * expr.c (convert_modes): Likewise.
15928 (store_field): Likewise.
15929 (expand_expr_real_1): Likewise.
15930 * fold-const.c (fold_read_from_constant_string): Likewise.
15931 * gimple-ssa-sprintf.c (get_format_string): Likewise.
15932 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
15933 * optabs.c (expand_binop): Likewise.
15934 (expand_unop): Likewise.
15935 (expand_abs_nojump): Likewise.
15936 (expand_one_cmpl_abs_nojump): Likewise.
15937 * simplify-rtx.c (mode_signbit_p): Likewise.
15938 (val_signbit_p): Likewise.
15939 (val_signbit_known_set_p): Likewise.
15940 (val_signbit_known_clear_p): Likewise.
15941 (simplify_relational_operation_1): Likewise.
15942 * tree.c (vector_type_mode): Likewise.
15944 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15945 Alan Hayward <alan.hayward@arm.com>
15946 David Sherwood <david.sherwood@arm.com>
15948 * machmode.h (smallest_mode_for_size): Fix formatting.
15949 (smallest_int_mode_for_size): New function.
15950 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
15951 instead of smallest_mode_for_size.
15952 * combine.c (make_extraction): Likewise.
15953 * config/arc/arc.c (arc_expand_movmem): Likewise.
15954 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
15955 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15956 * config/s390/s390.c (s390_expand_insv): Likewise.
15957 * config/sparc/sparc.c (assign_int_registers): Likewise.
15958 * config/spu/spu.c (spu_function_value): Likewise.
15959 (spu_function_arg): Likewise.
15960 * coverage.c (get_gcov_type): Likewise.
15961 (get_gcov_unsigned_t): Likewise.
15962 * dse.c (find_shift_sequence): Likewise.
15963 * expmed.c (store_bit_field_1): Likewise.
15964 * expr.c (convert_move): Likewise.
15965 (store_field): Likewise.
15966 * internal-fn.c (expand_arith_overflow): Likewise.
15967 * optabs-query.c (get_best_extraction_insn): Likewise.
15968 * optabs.c (expand_twoval_binop_libfunc): Likewise.
15969 * stor-layout.c (layout_type): Likewise.
15970 (initialize_sizetypes): Likewise.
15971 * targhooks.c (default_get_mask_mode): Likewise.
15972 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
15974 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15975 Alan Hayward <alan.hayward@arm.com>
15976 David Sherwood <david.sherwood@arm.com>
15978 * machmode.h (opt_mode::else_blk): New function.
15979 (int_mode_for_mode): Declare.
15980 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
15981 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
15983 * cfgexpand.c (expand_debug_expr): Likewise.
15984 * combine.c (gen_lowpart_or_truncate): Likewise.
15985 (gen_lowpart_for_combine): Likewise.
15986 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
15987 * config/avr/avr.c (avr_to_int_mode): Likewise.
15988 (avr_out_plus_1): Likewise.
15989 (avr_out_plus): Likewise.
15990 (avr_out_round): Likewise.
15991 * config/i386/i386.c (ix86_split_to_parts): Likewise.
15992 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15993 (s390_expand_vcond): Likewise.
15994 * config/spu/spu.c (spu_split_immediate): Likewise.
15995 (spu_expand_mov): Likewise.
15996 * dse.c (get_stored_val): Likewise.
15997 * expmed.c (store_bit_field_1): Likewise.
15998 (convert_extracted_bit_field): Use int_mode_for_mode instead of
16000 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
16001 (extract_low_bits): Likewise.
16002 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
16003 handling rather than repeating the check.
16004 (emit_group_store): Likewise.
16005 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
16006 * optabs.c (expand_absneg_bit): Likewise.
16007 (expand_copysign_absneg): Likewise.
16008 (expand_copysign_bit): Likewise.
16009 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
16010 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
16011 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
16012 * var-tracking.c (prepare_call_arguments): Likewise.
16013 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
16014 int_mode_for_mode instead of mode_for_size.
16015 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
16017 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16018 Alan Hayward <alan.hayward@arm.com>
16019 David Sherwood <david.sherwood@arm.com>
16021 * machmode.h (int_mode_for_size): New function.
16022 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
16023 instead of mode_for_size.
16024 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
16026 * combine.c (expand_field_assignment): Use int_mode_for_size
16027 instead of mode_for_size.
16028 (make_extraction): Likewise.
16029 (simplify_shift_const_1): Likewise.
16030 (simplify_comparison): Likewise.
16031 * dojump.c (do_jump): Likewise.
16032 * dwarf2out.c (mem_loc_descriptor): Likewise.
16033 * emit-rtl.c (init_derived_machine_modes): Likewise.
16034 * expmed.c (flip_storage_order): Likewise.
16035 (convert_extracted_bit_field): Likewise.
16036 * expr.c (copy_blkmode_from_reg): Likewise.
16037 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
16038 * internal-fn.c (expand_mul_overflow): Likewise.
16039 * lower-subreg.c (simple_move): Likewise.
16040 * optabs-libfuncs.c (init_optabs): Likewise.
16041 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16042 * tree.c (vector_type_mode): Likewise.
16043 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
16044 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
16045 * tree-vect-generic.c (expand_vector_parallel): Likewise.
16046 * tree-vect-stmts.c (vectorizable_load): Likewise.
16047 (vectorizable_store): Likewise.
16049 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16050 Alan Hayward <alan.hayward@arm.com>
16051 David Sherwood <david.sherwood@arm.com>
16053 * coretypes.h (pod_mode): New type.
16054 (scalar_int_mode_pod): New typedef.
16055 * machmode.h (pod_mode): New class.
16056 (int_n_data_t::m): Change type to scalar_int_mode_pod.
16057 * genmodes.c (emit_mode_int_n): Update accordingly.
16058 * lower-subreg.h (target_lower_subreg): Change type to
16059 scalar_int_mode_pod.
16060 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
16061 scalar_int_mode_pod.
16063 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16064 Alan Hayward <alan.hayward@arm.com>
16065 David Sherwood <david.sherwood@arm.com>
16067 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
16068 machine_mode to scalar_int_mode.
16069 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
16070 (rs6000_option_override_internal): Remove cast to int.
16071 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
16072 machine_mode to scalar_int_mode.
16073 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
16074 (rs6000_option_override_internal): Remove cast to int.
16075 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
16076 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
16078 * config/s390/s390.c (s390_expand_builtin): Likewise.
16079 * coretypes.h (scalar_int_mode): New type.
16080 (opt_scalar_int_mode): New typedef.
16081 * machmode.h (scalar_int_mode): New class.
16082 (scalar_int_mode::includes_p): New function.
16083 (byte_mode): Change type to scalar_int_mode.
16084 (word_mode): Likewise.
16085 (ptr_mode): Likewise.
16086 * emit-rtl.c (byte_mode): Likewise.
16087 (word_mode): Likewise.
16088 (ptr_mode): Likewise.
16089 (init_derived_machine_modes): Update accordingly.
16090 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
16091 and MODE_PARTIAL_INT.
16092 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
16093 opt_scalar_int_mode.
16095 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16096 Alan Hayward <alan.hayward@arm.com>
16097 David Sherwood <david.sherwood@arm.com>
16099 * target.def (libgcc_floating_mode_supported_p): Take a
16101 * doc/tm.texi: Regenerate.
16102 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
16104 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16105 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
16108 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16109 Alan Hayward <alan.hayward@arm.com>
16110 David Sherwood <david.sherwood@arm.com>
16112 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
16113 * doc/tm.texi: Regenerate.
16114 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
16115 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
16116 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
16117 * targhooks.h (default_floatn_mode): Likewise.
16118 * targhooks.c (default_floatn_mode): Likewise.
16119 * tree.c (build_common_tree_nodes): Update accordingly.
16121 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16122 Alan Hayward <alan.hayward@arm.com>
16123 David Sherwood <david.sherwood@arm.com>
16125 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
16126 (mode_iterator::iterate_p): Likewise.
16127 (mode_iterator::get_wider): Likewise.
16128 * expr.c (init_expr_target): Use opt_scalar_float_mode.
16130 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16131 Alan Hayward <alan.hayward@arm.com>
16132 David Sherwood <david.sherwood@arm.com>
16134 * coretypes.h (opt_scalar_float_mode): New typedef.
16135 * machmode.h (float_mode_for_size): New function.
16136 * emit-rtl.c (double_mode): Delete.
16137 (init_emit_once): Use float_mode_for_size.
16138 * stor-layout.c (layout_type): Likewise.
16139 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
16141 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16142 Alan Hayward <alan.hayward@arm.com>
16143 David Sherwood <david.sherwood@arm.com>
16145 * output.h (assemble_real): Take a scalar_float_mode.
16146 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
16147 * config/arm/arm.md (consttable_4): Likewise.
16148 (consttable_8): Likewise.
16149 (consttable_16): Likewise.
16150 * config/mips/mips.md (consttable_float): Likewise.
16151 * config/s390/s390.c (s390_output_pool_entry): Likewise.
16152 * varasm.c (assemble_real): Take a scalar_float_mode.
16153 (output_constant_pool_2): Update accordingly.
16154 (output_constant): Likewise.
16156 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16157 Alan Hayward <alan.hayward@arm.com>
16158 David Sherwood <david.sherwood@arm.com>
16160 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
16161 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
16162 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
16163 (native_encode_real): Likewise.
16164 (native_interpret_real): Likewise.
16165 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
16166 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16168 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16169 Alan Hayward <alan.hayward@arm.com>
16170 David Sherwood <david.sherwood@arm.com>
16172 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
16173 <scalar_float_mode>. Simplify.
16174 (gen_extend_conv_libfunc): Likewise.
16176 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16177 Alan Hayward <alan.hayward@arm.com>
16178 David Sherwood <david.sherwood@arm.com>
16180 * coretypes.h (scalar_float_mode): New type.
16181 * machmode.h (mode_traits::from_int): Use machine_mode if
16182 USE_ENUM_MODES is defined.
16183 (is_a): New function.
16185 (dyn_cast): Likewise.
16186 (scalar_float_mode): New class.
16187 (scalar_float_mode::includes_p): New function.
16188 (is_float_mode): Likewise.
16189 * gdbhooks.py (MachineModePrinter): New class.
16190 (build_pretty_printer): Use it for scalar_float_mode.
16191 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
16192 (format_helper::format_helper): Turn into a template.
16193 * genmodes.c (get_mode_class): New function.
16194 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
16195 or machine_mode if none.
16196 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
16197 as_a <scalar_float_mode>.
16198 * dwarf2out.c (mem_loc_descriptor): Likewise.
16199 (insert_float): Likewise.
16200 (add_const_value_attribute): Likewise.
16201 * simplify-rtx.c (simplify_immed_subreg): Likewise.
16202 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
16203 (expand_unop): Update accordingly.
16204 (expand_abs_nojump): Likewise.
16205 (expand_copysign_absneg): Take a scalar_float_mode.
16206 (expand_copysign_bit): Likewise.
16207 (expand_copysign): Update accordingly.
16209 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16210 Alan Hayward <alan.hayward@arm.com>
16211 David Sherwood <david.sherwood@arm.com>
16213 * coretypes.h (opt_mode): New class.
16214 * machmode.h (opt_mode): Likewise.
16215 (opt_mode::else_void): New function.
16216 (opt_mode::require): Likewise.
16217 (opt_mode::exists): Likewise.
16218 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
16219 (GET_MODE_2XWIDER_MODE): Likewise.
16220 (mode_iterator::get_wider): Update accordingly.
16221 (mode_iterator::get_2xwider): Likewise.
16222 (mode_iterator::get_known_wider): Likewise, turning into a template.
16223 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
16224 forcing a wider mode to exist.
16225 * config/cr16/cr16.h (LONG_REG_P): Likewise.
16226 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16227 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
16228 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
16229 * lower-subreg.c (init_lower_subreg): Likewise.
16230 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
16231 on the final iteration.
16232 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
16233 a wider mode exists before asking for a move pattern.
16234 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
16235 forcing a wider mode to exist.
16236 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
16237 returning false if no such mode exists.
16238 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
16239 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
16240 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
16241 Avoid checking for a MODE_INT if we already know the mode is not a
16243 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
16244 forcing a wider mode to exist.
16245 (expmed_mult_highpart_optab): Likewise.
16246 (expmed_mult_highpart): Likewise.
16247 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
16249 * lto-streamer-in.c (lto_input_mode_table): Likewise.
16250 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
16251 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16252 * internal-fn.c (expand_mul_overflow): Update use of
16253 GET_MODE_2XWIDER_MODE.
16254 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16255 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
16256 GET_MODE_WIDER_MODE.
16257 (convert_plusminus_to_widen): Likewise.
16258 * tree-switch-conversion.c (array_value_type): Likewise.
16259 * var-tracking.c (emit_note_insn_var_location): Likewise.
16260 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16261 Return false inside rather than outside the loop if no wider mode
16263 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
16264 and GET_MODE_2XWIDER_MODE
16265 (can_compare_p): Use else_void.
16266 * gdbhooks.py (OptMachineModePrinter): New class.
16267 (build_pretty_printer): Use it for opt_mode.
16269 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16270 Alan Hayward <alan.hayward@arm.com>
16271 David Sherwood <david.sherwood@arm.com>
16273 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
16274 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
16276 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16277 Alan Hayward <alan.hayward@arm.com>
16278 David Sherwood <david.sherwood@arm.com>
16280 * machmode.h (mode_traits): New structure.
16281 (get_narrowest_mode): New function.
16282 (mode_iterator::start): Likewise.
16283 (mode_iterator::iterate_p): Likewise.
16284 (mode_iterator::get_wider): Likewise.
16285 (mode_iterator::get_known_wider): Likewise.
16286 (mode_iterator::get_2xwider): Likewise.
16287 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
16288 (FOR_EACH_MODE): Likewise.
16289 (FOR_EACH_MODE_FROM): Likewise.
16290 (FOR_EACH_MODE_UNTIL): Likewise.
16291 (FOR_EACH_WIDER_MODE): Likewise.
16292 (FOR_EACH_2XWIDER_MODE): Likewise.
16293 * builtins.c (expand_builtin_strlen): Use new mode iterators.
16294 * combine.c (simplify_comparison): Likewise
16295 * config/i386/i386.c (type_natural_mode): Likewise.
16296 * cse.c (cse_insn): Likewise.
16297 * dse.c (find_shift_sequence): Likewise.
16298 * emit-rtl.c (init_derived_machine_modes): Likewise.
16299 (init_emit_once): Likewise.
16300 * explow.c (hard_function_value): Likewise.
16301 * expmed.c (extract_fixed_bit_field_1): Likewise.
16302 (extract_bit_field_1): Likewise.
16303 (expand_divmod): Likewise.
16304 (emit_store_flag_1): Likewise.
16305 * expr.c (init_expr_target): Likewise.
16306 (convert_move): Likewise.
16307 (alignment_for_piecewise_move): Likewise.
16308 (widest_int_mode_for_size): Likewise.
16309 (emit_block_move_via_movmem): Likewise.
16310 (copy_blkmode_to_reg): Likewise.
16311 (set_storage_via_setmem): Likewise.
16312 (compress_float_constant): Likewise.
16313 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16314 * optabs-query.c (get_best_extraction_insn): Likewise.
16315 * optabs.c (expand_binop): Likewise.
16316 (expand_twoval_unop): Likewise.
16317 (expand_twoval_binop): Likewise.
16318 (widen_leading): Likewise.
16319 (widen_bswap): Likewise.
16320 (expand_parity): Likewise.
16321 (expand_unop): Likewise.
16322 (prepare_cmp_insn): Likewise.
16323 (prepare_float_lib_cmp): Likewise.
16324 (expand_float): Likewise.
16325 (expand_fix): Likewise.
16326 (expand_sfix_optab): Likewise.
16327 * postreload.c (move2add_use_add2_insn): Likewise.
16328 * reg-stack.c (reg_to_stack): Likewise.
16329 * reginfo.c (choose_hard_reg_mode): Likewise.
16330 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16331 * stor-layout.c (mode_for_size): Likewise.
16332 (smallest_mode_for_size): Likewise.
16333 (mode_for_vector): Likewise.
16334 (finish_bitfield_representative): Likewise.
16335 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
16336 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
16337 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16338 * var-tracking.c (prepare_call_arguments): Likewise.
16340 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16341 Alan Hayward <alan.hayward@arm.com>
16342 David Sherwood <david.sherwood@arm.com>
16344 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
16345 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
16346 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
16347 * machmode.h (mode_size): Move earlier in file.
16348 (mode_precision): Likewise.
16349 (mode_inner): Likewise.
16350 (mode_nunits): Likewise.
16351 (mode_unit_size): Likewise.
16352 (unit_unit_precision): Likewise.
16353 (mode_wider): Likewise.
16354 (mode_2xwider): Likewise.
16355 (machine_mode): New class.
16356 (mode_to_bytes): New function.
16357 (mode_to_bits): Likewise.
16358 (mode_to_precision): Likewise.
16359 (mode_to_inner): Likewise.
16360 (mode_to_unit_size): Likewise.
16361 (mode_to_unit_precision): Likewise.
16362 (mode_to_nunits): Likewise.
16363 (GET_MODE_SIZE): Use mode_to_bytes.
16364 (GET_MODE_BITSIZE): Use mode_to_bits.
16365 (GET_MODE_PRECISION): Use mode_to_precision.
16366 (GET_MODE_INNER): Use mode_to_inner.
16367 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
16368 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
16369 (GET_MODE_NUNITS): Use mode_to_nunits.
16370 * system.h (ALWAYS_INLINE): New macro.
16371 * config/powerpcspe/powerpcspe-c.c
16372 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
16373 int for arg1_mode and arg2_mode.
16375 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16376 Alan Hayward <alan.hayward@arm.com>
16377 David Sherwood <david.sherwood@arm.com>
16379 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
16380 Prefix mode names with E_ in case statements.
16381 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16382 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
16383 (aarch64_split_simd_move): Likewise.
16384 (aarch64_gen_storewb_pair): Likewise.
16385 (aarch64_gen_loadwb_pair): Likewise.
16386 (aarch64_gen_store_pair): Likewise.
16387 (aarch64_gen_load_pair): Likewise.
16388 (aarch64_get_condition_code_1): Likewise.
16389 (aarch64_constant_pool_reload_icode): Likewise.
16390 (get_rsqrte_type): Likewise.
16391 (get_rsqrts_type): Likewise.
16392 (get_recpe_type): Likewise.
16393 (get_recps_type): Likewise.
16394 (aarch64_gimplify_va_arg_expr): Likewise.
16395 (aarch64_simd_container_mode): Likewise.
16396 (aarch64_emit_load_exclusive): Likewise.
16397 (aarch64_emit_store_exclusive): Likewise.
16398 (aarch64_expand_compare_and_swap): Likewise.
16399 (aarch64_gen_atomic_cas): Likewise.
16400 (aarch64_emit_bic): Likewise.
16401 (aarch64_emit_atomic_swap): Likewise.
16402 (aarch64_emit_atomic_load_op): Likewise.
16403 (aarch64_evpc_trn): Likewise.
16404 (aarch64_evpc_uzp): Likewise.
16405 (aarch64_evpc_zip): Likewise.
16406 (aarch64_evpc_ext): Likewise.
16407 (aarch64_evpc_rev): Likewise.
16408 (aarch64_evpc_dup): Likewise.
16409 (aarch64_gen_ccmp_first): Likewise.
16410 (aarch64_gen_ccmp_next): Likewise.
16411 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
16412 (alpha_emit_xfloating_libcall): Likewise.
16413 (emit_insxl): Likewise.
16414 (alpha_arg_type): Likewise.
16415 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
16416 (arc_preferred_simd_mode): Likewise.
16417 (arc_secondary_reload): Likewise.
16418 (get_arc_condition_code): Likewise.
16419 (arc_print_operand): Likewise.
16420 (arc_legitimate_constant_p): Likewise.
16421 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16422 * config/arc/arc.md (casesi_load): Likewise.
16423 (casesi_compact_jump): Likewise.
16424 * config/arc/predicates.md (proper_comparison_operator): Likewise.
16425 (cc_use_register): Likewise.
16426 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16427 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
16428 (arm_init_iwmmxt_builtins): Likewise.
16429 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
16430 (neon_expand_vector_init): Likewise.
16431 (arm_attr_length_move_neon): Likewise.
16432 (maybe_get_arm_condition_code): Likewise.
16433 (arm_emit_vector_const): Likewise.
16434 (arm_preferred_simd_mode): Likewise.
16435 (arm_output_iwmmxt_tinsr): Likewise.
16436 (thumb1_output_casesi): Likewise.
16437 (thumb2_output_casesi): Likewise.
16438 (arm_emit_load_exclusive): Likewise.
16439 (arm_emit_store_exclusive): Likewise.
16440 (arm_expand_compare_and_swap): Likewise.
16441 (arm_evpc_neon_vuzp): Likewise.
16442 (arm_evpc_neon_vzip): Likewise.
16443 (arm_evpc_neon_vrev): Likewise.
16444 (arm_evpc_neon_vtrn): Likewise.
16445 (arm_evpc_neon_vext): Likewise.
16446 (arm_validize_comparison): Likewise.
16447 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
16448 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
16449 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
16450 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
16451 (c6x_preferred_simd_mode): Likewise.
16452 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
16453 (epiphany_rtx_costs): Likewise.
16454 * config/epiphany/predicates.md (proper_comparison_operator):
16456 * config/frv/frv.c (condexec_memory_operand): Likewise.
16457 (frv_emit_move): Likewise.
16458 (output_move_single): Likewise.
16459 (output_condmove_single): Likewise.
16460 (frv_hard_regno_mode_ok): Likewise.
16461 (frv_matching_accg_mode): Likewise.
16462 * config/h8300/h8300.c (split_adds_subs): Likewise.
16463 (h8300_rtx_costs): Likewise.
16464 (h8300_print_operand): Likewise.
16465 (compute_mov_length): Likewise.
16466 (output_logical_op): Likewise.
16467 (compute_logical_op_length): Likewise.
16468 (compute_logical_op_cc): Likewise.
16469 (h8300_shift_needs_scratch_p): Likewise.
16470 (output_a_shift): Likewise.
16471 (compute_a_shift_length): Likewise.
16472 (compute_a_shift_cc): Likewise.
16473 (expand_a_rotate): Likewise.
16474 (output_a_rotate): Likewise.
16475 * config/i386/i386.c (classify_argument): Likewise.
16476 (function_arg_advance_32): Likewise.
16477 (function_arg_32): Likewise.
16478 (function_arg_64): Likewise.
16479 (function_value_64): Likewise.
16480 (ix86_gimplify_va_arg): Likewise.
16481 (ix86_legitimate_constant_p): Likewise.
16482 (put_condition_code): Likewise.
16483 (split_double_mode): Likewise.
16484 (ix86_avx256_split_vector_move_misalign): Likewise.
16485 (ix86_expand_vector_logical_operator): Likewise.
16486 (ix86_split_idivmod): Likewise.
16487 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
16488 (ix86_build_const_vector): Likewise.
16489 (ix86_build_signbit_mask): Likewise.
16490 (ix86_match_ccmode): Likewise.
16491 (ix86_cc_modes_compatible): Likewise.
16492 (ix86_expand_branch): Likewise.
16493 (ix86_expand_sse_cmp): Likewise.
16494 (ix86_expand_sse_movcc): Likewise.
16495 (ix86_expand_int_sse_cmp): Likewise.
16496 (ix86_expand_vec_perm_vpermi2): Likewise.
16497 (ix86_expand_vec_perm): Likewise.
16498 (ix86_expand_sse_unpack): Likewise.
16499 (ix86_expand_int_addcc): Likewise.
16500 (ix86_split_to_parts): Likewise.
16501 (ix86_vectorize_builtin_gather): Likewise.
16502 (ix86_vectorize_builtin_scatter): Likewise.
16503 (avx_vpermilp_parallel): Likewise.
16504 (inline_memory_move_cost): Likewise.
16505 (ix86_tieable_integer_mode_p): Likewise.
16506 (x86_maybe_negate_const_int): Likewise.
16507 (ix86_expand_vector_init_duplicate): Likewise.
16508 (ix86_expand_vector_init_one_nonzero): Likewise.
16509 (ix86_expand_vector_init_one_var): Likewise.
16510 (ix86_expand_vector_init_concat): Likewise.
16511 (ix86_expand_vector_init_interleave): Likewise.
16512 (ix86_expand_vector_init_general): Likewise.
16513 (ix86_expand_vector_set): Likewise.
16514 (ix86_expand_vector_extract): Likewise.
16515 (emit_reduc_half): Likewise.
16516 (ix86_emit_i387_round): Likewise.
16517 (ix86_mangle_type): Likewise.
16518 (ix86_expand_round_sse4): Likewise.
16519 (expand_vec_perm_blend): Likewise.
16520 (canonicalize_vector_int_perm): Likewise.
16521 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
16522 (expand_vec_perm_1): Likewise.
16523 (expand_vec_perm_interleave3): Likewise.
16524 (expand_vec_perm_even_odd_pack): Likewise.
16525 (expand_vec_perm_even_odd_1): Likewise.
16526 (expand_vec_perm_broadcast_1): Likewise.
16527 (ix86_vectorize_vec_perm_const_ok): Likewise.
16528 (ix86_expand_vecop_qihi): Likewise.
16529 (ix86_expand_mul_widen_hilo): Likewise.
16530 (ix86_expand_sse2_abs): Likewise.
16531 (ix86_expand_pextr): Likewise.
16532 (ix86_expand_pinsr): Likewise.
16533 (ix86_preferred_simd_mode): Likewise.
16534 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
16535 * config/i386/sse.md (*andnot<mode>3): Likewise.
16536 (<mask_codefor><code><mode>3<mask_name>): Likewise.
16537 (*<code><mode>3): Likewise.
16538 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
16539 (ia64_expand_atomic_op): Likewise.
16540 (ia64_arg_type): Likewise.
16541 (ia64_mode_to_int): Likewise.
16542 (ia64_scalar_mode_supported_p): Likewise.
16543 (ia64_vector_mode_supported_p): Likewise.
16544 (expand_vec_perm_broadcast): Likewise.
16545 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
16546 (iq2000_function_arg_advance): Likewise.
16547 (iq2000_function_arg): Likewise.
16548 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
16549 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
16550 (m68k_libcall_value): Likewise.
16551 (m68k_function_value): Likewise.
16552 (sched_attr_op_type): Likewise.
16553 * config/mcore/mcore.c (mcore_output_move): Likewise.
16554 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
16556 (microblaze_function_arg): Likewise.
16557 * config/mips/mips.c (mips16_build_call_stub): Likewise.
16558 (mips_print_operand): Likewise.
16559 (mips_mode_ok_for_mov_fmt_p): Likewise.
16560 (mips_vector_mode_supported_p): Likewise.
16561 (mips_preferred_simd_mode): Likewise.
16562 (mips_expand_vpc_loongson_even_odd): Likewise.
16563 (mips_expand_vec_unpack): Likewise.
16564 (mips_expand_vi_broadcast): Likewise.
16565 (mips_expand_vector_init): Likewise.
16566 (mips_expand_vec_reduc): Likewise.
16567 (mips_expand_msa_cmp): Likewise.
16568 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
16569 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
16570 (cc_flags_for_mode): Likewise.
16571 * config/msp430/msp430.c (msp430_print_operand): Likewise.
16572 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
16573 (nds32_output_casesi_pc_relative): Likewise.
16574 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16575 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
16576 (nvptx_gen_unpack): Likewise.
16577 (nvptx_gen_pack): Likewise.
16578 (nvptx_gen_shuffle): Likewise.
16579 (nvptx_gen_wcast): Likewise.
16580 (nvptx_preferred_simd_mode): Likewise.
16581 * config/pa/pa.c (pa_secondary_reload): Likewise.
16582 * config/pa/predicates.md (base14_operand): Likewise.
16583 * config/powerpcspe/powerpcspe-c.c
16584 (altivec_resolve_overloaded_builtin): Likewise.
16585 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
16587 (rs6000_preferred_simd_mode): Likewise.
16588 (output_vec_const_move): Likewise.
16589 (rs6000_expand_vector_extract): Likewise.
16590 (rs6000_split_vec_extract_var): Likewise.
16591 (reg_offset_addressing_ok_p): Likewise.
16592 (rs6000_legitimate_offset_address_p): Likewise.
16593 (rs6000_legitimize_address): Likewise.
16594 (rs6000_emit_set_const): Likewise.
16595 (rs6000_const_vec): Likewise.
16596 (rs6000_emit_move): Likewise.
16597 (spe_build_register_parallel): Likewise.
16598 (rs6000_darwin64_record_arg_recurse): Likewise.
16599 (swap_selector_for_mode): Likewise.
16600 (spe_init_builtins): Likewise.
16601 (paired_init_builtins): Likewise.
16602 (altivec_init_builtins): Likewise.
16603 (do_load_for_compare): Likewise.
16604 (rs6000_generate_compare): Likewise.
16605 (rs6000_expand_float128_convert): Likewise.
16606 (emit_load_locked): Likewise.
16607 (emit_store_conditional): Likewise.
16608 (rs6000_output_function_epilogue): Likewise.
16609 (rs6000_handle_altivec_attribute): Likewise.
16610 (rs6000_function_value): Likewise.
16611 (emit_fusion_gpr_load): Likewise.
16612 (emit_fusion_p9_load): Likewise.
16613 (emit_fusion_p9_store): Likewise.
16614 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
16615 (fusion_gpr_mem_load): Likewise.
16616 (fusion_addis_mem_combo_load): Likewise.
16617 (fusion_addis_mem_combo_store): Likewise.
16618 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
16619 (fusion_gpr_mem_load): Likewise.
16620 (fusion_addis_mem_combo_load): Likewise.
16621 (fusion_addis_mem_combo_store): Likewise.
16622 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16624 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
16625 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16626 (rs6000_preferred_simd_mode): Likewise.
16627 (output_vec_const_move): Likewise.
16628 (rs6000_expand_vector_extract): Likewise.
16629 (rs6000_split_vec_extract_var): Likewise.
16630 (reg_offset_addressing_ok_p): Likewise.
16631 (rs6000_legitimate_offset_address_p): Likewise.
16632 (rs6000_legitimize_address): Likewise.
16633 (rs6000_emit_set_const): Likewise.
16634 (rs6000_const_vec): Likewise.
16635 (rs6000_emit_move): Likewise.
16636 (rs6000_darwin64_record_arg_recurse): Likewise.
16637 (swap_selector_for_mode): Likewise.
16638 (paired_init_builtins): Likewise.
16639 (altivec_init_builtins): Likewise.
16640 (rs6000_expand_float128_convert): Likewise.
16641 (emit_load_locked): Likewise.
16642 (emit_store_conditional): Likewise.
16643 (rs6000_output_function_epilogue): Likewise.
16644 (rs6000_handle_altivec_attribute): Likewise.
16645 (rs6000_function_value): Likewise.
16646 (emit_fusion_gpr_load): Likewise.
16647 (emit_fusion_p9_load): Likewise.
16648 (emit_fusion_p9_store): Likewise.
16649 * config/rx/rx.c (rx_gen_move_template): Likewise.
16650 (flags_from_mode): Likewise.
16651 * config/s390/predicates.md (s390_alc_comparison): Likewise.
16652 (s390_slb_comparison): Likewise.
16653 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
16654 (s390_vector_mode_supported_p): Likewise.
16655 (s390_cc_modes_compatible): Likewise.
16656 (s390_match_ccmode_set): Likewise.
16657 (s390_canonicalize_comparison): Likewise.
16658 (s390_emit_compare_and_swap): Likewise.
16659 (s390_branch_condition_mask): Likewise.
16660 (s390_rtx_costs): Likewise.
16661 (s390_secondary_reload): Likewise.
16662 (__SECONDARY_RELOAD_CASE): Likewise.
16663 (s390_expand_cs): Likewise.
16664 (s390_preferred_simd_mode): Likewise.
16665 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
16666 * config/sh/sh.c (sh_print_operand): Likewise.
16667 (dump_table): Likewise.
16668 (sh_secondary_reload): Likewise.
16669 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16670 * config/sh/sh.md (casesi_worker_1): Likewise.
16671 (casesi_worker_2): Likewise.
16672 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
16673 (fcc_comparison_operator): Likewise.
16674 * config/sparc/sparc.c (sparc_expand_move): Likewise.
16675 (emit_soft_tfmode_cvt): Likewise.
16676 (sparc_preferred_simd_mode): Likewise.
16677 (output_cbranch): Likewise.
16678 (sparc_print_operand): Likewise.
16679 (sparc_expand_vec_perm_bmask): Likewise.
16680 (vector_init_bshuffle): Likewise.
16681 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
16682 (spu_vector_mode_supported_p): Likewise.
16683 (spu_expand_insv): Likewise.
16684 (spu_emit_branch_or_set): Likewise.
16685 (spu_handle_vector_attribute): Likewise.
16686 (spu_builtin_splats): Likewise.
16687 (spu_builtin_extract): Likewise.
16688 (spu_builtin_promote): Likewise.
16689 (spu_expand_sign_extend): Likewise.
16690 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
16691 (tilegx_simd_int): Likewise.
16692 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
16693 (tilepro_simd_int): Likewise.
16694 * config/v850/v850.c (const_double_split): Likewise.
16695 (v850_print_operand): Likewise.
16696 (ep_memory_offset): Likewise.
16697 * config/vax/vax.c (vax_rtx_costs): Likewise.
16698 (vax_output_int_move): Likewise.
16699 (vax_output_int_add): Likewise.
16700 (vax_output_int_subtract): Likewise.
16701 * config/visium/predicates.md (visium_branch_operator): Likewise.
16702 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
16703 (visium_print_operand_address): Likewise.
16704 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16705 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
16706 (xtensa_expand_conditional_branch): Likewise.
16707 (xtensa_copy_incoming_a7): Likewise.
16708 (xtensa_output_literal): Likewise.
16709 * dfp.c (decimal_real_maxval): Likewise.
16710 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16712 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16713 Alan Hayward <alan.hayward@arm.com>
16714 David Sherwood <david.sherwood@arm.com>
16716 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
16717 (mode_nunits_inline): Likewise.
16718 (mode_inner_inline): Likewise.
16719 (mode_unit_size_inline): Likewise.
16720 (mode_unit_precision_inline): Likewise.
16721 (emit_insn_modes_h): Likewise. Also emit a #define of the
16723 (emit_mode_wider): Add an E_ prefix to mode names.
16724 (emit_mode_complex): Likewise.
16725 (emit_mode_inner): Likewise.
16726 (emit_mode_adjustments): Likewise.
16727 (emit_mode_int_n): Likewise.
16728 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
16729 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
16730 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
16731 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
16732 (CRC32_BUILTIN, ENTRY): Likewise.
16733 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
16734 (aarch64_pop_regs): Likewise.
16735 (aarch64_process_components): Likewise.
16736 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
16737 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
16738 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
16739 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
16740 * config/arm/arm.c (arm_init_libfuncs): Likewise.
16741 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
16743 * config/i386/i386-builtin.def (pcmpestr): Likewise.
16744 (pcmpistr): Likewise.
16745 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
16746 * config/mmix/mmix.c (mmix_output_condition): Likewise.
16747 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
16749 * config/rl78/rl78.c (mduc_regs): Likewise.
16750 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16751 (htm_expand_builtin): Likewise.
16752 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
16753 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
16754 * config/xtensa/xtensa.c (print_operand): Likewise.
16755 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
16756 (NUM_MODE_VECTOR_INT): Likewise.
16757 * genoutput.c (null_operand): Likewise.
16758 (output_operand_data): Likewise.
16759 * genrecog.c (print_parameter_value): Likewise.
16760 * lra.c (debug_operand_data): Likewise.
16762 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16764 * dwarf2out.c (scompare_loc_descriptor_wide)
16765 (scompare_loc_descriptor_narrow): New functions, split out from...
16766 (scompare_loc_descriptor): ...here.
16767 * expmed.c (emit_store_flag_int): New function, split out from...
16768 (emit_store_flag): ...here.
16770 2017-08-30 Richard Biener <rguenther@suse.de>
16772 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
16773 (dwarf2out_early_finish): Move setting of AT_pubnames from
16774 early debug output to early finish.
16776 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16778 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
16779 and -mdata-region to the assembler.
16781 2017-08-30 Richard Biener <rguenther@suse.de>
16783 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
16785 (gen_subprogram_die): Add DW_AT_object_pointer only early.
16786 (dwarf2out_early_global_decl): Only generate a DIE for the
16787 abstract origin if it doesn't already exist or is a declaration DIE.
16788 (resolve_addr): Do not add the linkage name twice when
16789 generating a stub DIE for the DW_TAG_GNU_call_site target.
16791 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16793 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16794 Use machine_mode rather than int for arg1_mode.
16796 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
16799 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
16800 that the second argument of the built-in functions to unpack
16801 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
16802 switch statement instead a lot of if statements.
16803 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
16804 Allow 64-bit values to be in Altivec registers as well as
16805 traditional floating point registers.
16806 (pack<mode>, FMOVE128_VSX iterator): Likewise.
16808 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
16810 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
16811 MAX_REGS_PER_ADDRESS == 1.
16813 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
16815 * config/i386/i386.opt (flag_fentry): Do not init to -1.
16816 * config/i386/i386.c (ix86_option_override_internal): Simplify
16817 setting of opts->x_flag_entry.
16819 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16820 Jakub Jelinek <jakub@redhat.com>
16821 Richard Biener <rguenther@suse.de>
16823 PR tree-optimization/81503
16824 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
16825 folded constant fits in the target type; reorder tests for clarity.
16827 2017-08-29 Martin Liska <mliska@suse.cz>
16829 * passes.def: Include pass_lower_switch.
16830 * stmt.c (dump_case_nodes): Remove and move to
16831 tree-switch-conversion.
16832 (case_values_threshold): Likewise.
16833 (expand_switch_as_decision_tree_p): Likewise.
16834 (emit_case_decision_tree): Likewise.
16835 (expand_case): Likewise.
16836 (balance_case_nodes): Likewise.
16837 (node_has_low_bound): Likewise.
16838 (node_has_high_bound): Likewise.
16839 (node_is_bounded): Likewise.
16840 (emit_case_nodes): Likewise.
16841 (struct simple_case_node): New struct.
16842 (add_case_node): Remove.
16843 (emit_case_dispatch_table): Use vector instead of case_list.
16844 (reset_out_edges_aux): Remove.
16845 (compute_cases_per_edge): Likewise.
16846 (expand_case): Build list of simple_case_node.
16847 (expand_sjlj_dispatch_table): Use it.
16848 * tree-switch-conversion.c (struct case_node): Moved from
16849 stmt.c and adjusted.
16850 (emit_case_nodes): Likewise.
16851 (node_has_low_bound): Likewise.
16852 (node_has_high_bound): Likewise.
16853 (node_is_bounded): Likewise.
16854 (case_values_threshold): Likewise.
16855 (reset_out_edges_aux): Likewise.
16856 (compute_cases_per_edge): Likewise.
16857 (add_case_node): Likewise.
16858 (dump_case_nodes): Likewise.
16859 (balance_case_nodes): Likewise.
16860 (expand_switch_as_decision_tree_p): Likewise.
16861 (emit_jump): Likewise.
16862 (emit_case_decision_tree): Likewise.
16863 (try_switch_expansion): Likewise.
16864 (do_jump_if_equal): Likewise.
16865 (emit_cmp_and_jump_insns): Likewise.
16866 (fix_phi_operands_for_edge): New function.
16867 (record_phi_operand_mapping): Likewise.
16868 (class pass_lower_switch): New pass.
16869 (pass_lower_switch::execute): New function.
16870 (make_pass_lower_switch): Likewise.
16871 (conditional_probability):
16872 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
16873 * tree-pass.h: Add make_pass_lower_switch.
16875 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16878 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
16881 2017-08-29 Richard Biener <rguenther@suse.de>
16883 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
16884 we do not add a DW_AT_inline attribute twice.
16885 (gen_subprogram_die): Remove code setting DW_AT_inline on
16886 DECL_ABSTRACT_P nodes.
16888 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
16890 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
16891 calls to internal functions.
16892 (gimplify_modify_expr): Likewise.
16893 * tree-call-cdce.c (use_internal_fn): Likewise.
16894 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
16895 (convert_to_divmod): Set the nothrow flag.
16896 * tree-if-conv.c (predicate_mem_writes): Likewise.
16897 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16898 (vectorizable_call): Likewise.
16899 (vectorizable_store): Likewise.
16900 (vectorizable_load): Likewise.
16901 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
16902 (vect_recog_mask_conversion_pattern): Likewise.
16904 2017-08-29 Martin Liska <mliska@suse.cz>
16907 * gcc.c (driver_handle_option): Add new argument.
16908 * opts-common.c (handle_option): Pass
16909 target_option_override_hook.
16910 * opts-global.c (lang_handle_option): Add new option.
16911 (set_default_handlers): Add new argument.
16912 (decode_options): Likewise.
16913 * opts.c (target_handle_option): Likewise.
16914 (common_handle_option): Call target_option_override_hook.
16915 * opts.h (struct cl_option_handler_func): Add hook for
16916 target option override.
16917 (struct cl_option_handlers): Likewise.
16918 (set_default_handlers): Add new argument.
16919 (decode_options): Likewise.
16920 (common_handle_option): Likewise.
16921 (target_handle_option): Likewise.
16922 * toplev.c (toplev::main): Pass targetm.target_option.override
16925 2017-08-29 Richard Biener <rguenther@suse.de>
16926 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
16928 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
16929 life to the active subtree.
16931 2017-08-28 Jeff Law <law@redhat.com>
16933 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
16934 derive_equivalences.
16935 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
16936 Code moved into....
16937 (edge_info::derive_equivalences): New private member function
16939 * tree-ssa-dom.c (class edge_info): Changed from a struct
16940 to a class. Add ctor/dtor, methods and data members.
16941 (edge_info::edge_info): Renamed from allocate_edge_info.
16942 Initialize additional members.
16943 (edge_info::~edge_info): New.
16944 (free_dom_edge_info): Delete the edge info.
16945 (record_edge_info): Use new class & associated member functions.
16946 Tighten forms for testing for edge equivalences.
16947 (record_temporary_equivalences): Iterate over the simple
16948 equivalences rather than assuming there's only one per edge.
16949 (cprop_into_successor_phis): Iterate over the simple
16950 equivalences rather than assuming there's only one per edge.
16951 (optimize_stmt): Use operand_equal_p rather than pointer
16952 equality for mini-DSE code.
16954 2017-08-28 Nathan Sidwell <nathan@acm.org>
16956 * gcc.c (execute): Fold SIGPIPE handling into switch
16957 statement. Adjust internal error message.
16959 2017-08-28 Richard Biener <rguenther@suse.de>
16962 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
16963 Do nothing for removed DIEs.
16965 2017-08-28 Richard Biener <rguenther@suse.de>
16967 PR tree-optimization/81977
16968 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
16971 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
16974 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
16975 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
16976 using targetm.gen_mem_thread_fence.
16978 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
16981 * config/i386/i386.md (*<btsc><mode>): Change operand 2
16982 predicate to register_operand. Reorder operands.
16983 (*btr<mode>): Ditto.
16984 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
16985 (*btr<mode>_mask): Ditto.
16987 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
16989 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
16990 * config/rs6000/xmmintrin.h: New file.
16991 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
16993 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16996 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
16997 parameter and_insn and return it.
16998 (recombine_lvx_pattern): Insert a copy to ensure availability of
16999 the base register of the copied masking operation at the point of
17000 the instruction replacement.
17001 (recombine_stvx_pattern): Likewise.
17003 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
17005 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
17006 undocumented switches.
17007 (-mpower9-dform-vector): Likewise.
17008 (-mpower9-dform): Likewise.
17009 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
17010 comments to delete references to -mpower9-dform* switches.
17011 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
17012 Delete reference to -mpower9-dform* switches, test for
17013 -mpower9-vector instead.
17014 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
17015 (OTHER_P9_VECTOR_MASKS): Likewise.
17016 (POWERPC_MASKS): Likewise.
17017 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
17018 tests against -mpower9-dform* to -mpower9-vector. Delete code
17019 that checked for -mpower9-dform* consistancy with other options.
17020 Add test for -mpower9-misc to enable other power9 switches.
17021 (rs6000_init_hard_regno_mode_ok): Likewise.
17022 (rs6000_option_override_internal): Likewise.
17023 (rs6000_emit_prologue): Likewise.
17024 (rs6000_emit_epilogue): Likewise.
17025 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
17026 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
17027 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
17029 (emit_fusion_p9_store): Likewise.
17030 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
17031 resetting these macros if the assembler does not support ISA 3.0
17033 (TARGET_P9_DFORM_VECTOR): Likewise.
17034 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
17035 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
17037 2017-08-25 Alan Modra <amodra@gmail.com>
17039 PR rtl-optimization/81747
17040 * cse.c (cse_extended_basic_block): Don't attempt to record
17041 equivalences for degenerate conditional branches that branch
17042 to their fall-through.
17044 2017-08-24 Martin Sebor <msebor@redhat.com>
17046 PR middle-end/81908
17047 * gimple-fold.c (size_must_be_zero_p): New function.
17048 (gimple_fold_builtin_memory_op): Call it.
17050 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
17052 * config/rs6000/mm_malloc.h: New file.
17054 2017-08-24 Bin Cheng <bin.cheng@arm.com>
17056 PR tree-optimization/81913
17057 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
17058 analysis when either IVs in condition can wrap.
17060 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
17062 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
17063 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
17065 2017-08-24 Richard Biener <rguenther@suse.de>
17068 * targhooks.c (default_target_can_inline_p): Properly
17069 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
17070 is present and always compare.
17071 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
17072 infer -mfpmath=sse from TARGET_SSE_P.
17073 (ix86_can_inline_p): Properly use target_option_default_node when
17074 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
17076 2017-08-24 Richard Biener <rguenther@suse.de>
17079 * dwarf2out.c (output_die): Handle flag_generate_offload like
17081 (output_comp_unit): Likewise.
17082 (gen_array_type_die): Likewise.
17083 (dwarf2out_early_finish): Likewise.
17084 (note_variable_value_in_expr): Likewise.
17085 (dwarf2out_finish): Likewise. Adjust assert.
17086 * cgraphunit.c (symbol_table::compile): Move setting of
17087 flag_generate_offload earlier ...
17088 (symbol_table::finalize_compilation_unit): ... here, before
17089 early debug finalization.
17091 2017-08-24 Richard Biener <rguenther@suse.de>
17093 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
17094 and ipa-fnsummary.h.
17095 (ix86_can_inline_p): When ix86_fpmath flags do not match
17096 check whether the callee uses FP math at all.
17098 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
17100 PR middle-end/81931
17101 * tree-ssanames.c (get_nonzero_bits): Use element_precision
17102 instead of TYPE_PRECISION.
17104 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
17105 Alan Hayward <alan.hayward@arm.com>
17106 David Sherwood <david.sherwood@arm.com>
17108 * combine.c (make_extraction): Use subreg_offset_from_lsb.
17110 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
17112 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
17114 (ix86_frame::stack_realign_allocate): New field.
17115 (struct machine_frame_state): Modify comments.
17116 (machine_frame_state::sp_realigned_fp_end): New field.
17117 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
17118 layout calculation.
17119 (sp_valid_at): Add assertion to assure no attempt to access invalid
17120 offset of a realigned stack.
17121 (fp_valid_at): Likewise.
17122 (choose_baseaddr): Modify comments.
17123 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
17124 ix86_expand_prologue.
17125 (ix86_expand_prologue): Modify stack realignment and allocation.
17126 (ix86_expand_epilogue): Modify comments.
17127 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
17128 avx2_runtime, avx512f, and avx512f_runtime.
17130 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
17132 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
17133 (mstackrealign): Do not init to -1.
17134 * config/i386/i386.c (ix86_option_override_internal):
17135 Check opts_set, not opts when setting default value of
17136 opts->x_ix86_force_align_arg_pointer.
17138 2017-08-23 Richard Biener <rguenther@suse.de>
17140 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
17141 lang_hooks.decl_printable_name.
17142 * print-rtl-function.c (print_rtx_function): Likewise.
17143 * tree-pretty-print.c (dump_function_header): Likewise.
17145 2017-08-23 Richard Biener <rguenther@suse.de>
17148 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
17149 -g0 at compile-time.
17151 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17153 PR middle-end/19706
17154 * doc/sourcebuild.texi (Other hardware attributes):
17157 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17159 PR middle-end/19706
17160 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
17161 Fix single-use check.
17163 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
17165 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
17167 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17169 * doc/install.texi: Modify to add more details on running selected
17172 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17174 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
17175 is combined with -mabi=ms.
17176 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
17179 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17181 PR tree-optimization/81488
17182 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
17183 and cached_basis fields.
17184 (MAX_SPREAD): New constant.
17185 (alloc_cand_and_find_basis): Initialize new fields.
17186 (clear_visited): New function.
17187 (create_phi_basis_1): Rename from create_phi_basis, set visited
17188 and cached_basis fields.
17189 (create_phi_basis): New wrapper function.
17190 (phi_add_costs_1): Rename from phi_add_costs, add spread
17191 parameter, set visited field, short-circuit when limits reached.
17192 (phi_add_costs): New wrapper function.
17193 (record_phi_increments_1): Rename from record_phi_increments, set
17195 (record_phi_increments): New wrapper function.
17196 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
17197 (phi_incr_cost): New wrapper function.
17198 (all_phi_incrs_profitable_1): Rename from
17199 all_phi_incrs_profitable, set visited field.
17200 (all_phi_incrs_profitable): New wrapper function.
17202 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17203 Alan Hayward <alan.hayward@arm.com>
17204 David Sherwood <david.sherwood@arm.com>
17206 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
17207 that takes the outer and inner modes.
17208 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
17209 comparison as the canonical test for a paradoxical subreg.
17210 * combine.c (simplify_set): Use paradoxical_subreg_p.
17211 (make_extraction): Likewise.
17212 (force_to_mode): Likewise.
17213 (rtx_equal_for_field_assignment_p): Likewise.
17214 (gen_lowpart_for_combine): Likewise.
17215 (simplify_comparison): Likewise.
17216 * cse.c (equiv_constant): Likewise.
17217 * expmed.c (store_bit_field_1): Likewise.
17218 * final.c (alter_subreg): Likewise.
17219 * fwprop.c (propagate_rtx): Likewise.
17220 (forward_propagate_subreg): Likewise.
17221 * ira-conflicts.c (ira_build_conflicts): Likewise.
17222 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
17223 * lra-constraints.c (curr_insn_transform): Likewise.
17224 (split_reg): Likewise.
17225 * lra-eliminations.c (move_plus_up): Likewise.
17226 (lra_eliminate_regs_1): Likewise.
17227 * recog.c (general_operand): Likewise.
17228 * ree.c (combine_reaching_defs): Likewise.
17229 * reload.c (push_reload): Likewise.
17230 (find_reloads): Likewise.
17231 * reload1.c (elimination_effects): Likewise.
17232 (compute_reload_subreg_offset): Likewise.
17233 (choose_reload_regs): Likewise.
17234 * rtlanal.c (subreg_lsb_1): Likewise.
17235 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17236 (simplify_subreg): Likewise.
17237 * var-tracking.c (track_loc_p): Likewise.
17238 * emit-rtl.c (byte_lowpart_offset): Likewise.
17239 (paradoxical_subreg_p): Delete out-of-line definition.
17241 2017-08-22 Jeff Law <law@redhat.com>
17243 PR tree-optimization/81741
17244 PR tree-optimization/71947
17245 * tree-ssa-dom.c: Include tree-inline.h.
17246 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
17247 equivalences if one is more expensive to compute than the other.
17248 * tree-ssa-scopedtables.h (class const_or_copies): Make
17249 record_const_or_copy_raw method private.
17250 (class avail_exprs_stack): New method simplify_binary_operation.
17251 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
17252 avail_exprs_stack::simplify_binary_operation as needed.
17253 (avail_exprs_stack::simplify_binary_operation): New function.
17255 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
17257 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
17258 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
17259 (DOT_SYMBOLS): Likewise.
17260 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
17261 (RELOCATABLE_NEEDS_FIXUP): Likewise.
17262 (RS6000_ABI_NAME): Likewise.
17263 (TARGET_CMODEL): Likewise.
17264 (TOC_SECTION_ASM_OP): Likewise.
17265 (SET_CMODEL): New macro.
17266 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
17268 2017-08-22 Richard Biener <rguenther@suse.de>
17270 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
17271 to allow for free-lang-data replacements similar to verify_type_variant.
17273 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17274 Alan Hayward <alan.hayward@arm.com>
17275 David Sherwood <david.sherwood@arm.com>
17277 * config/aarch64/aarch64.md (casesi): Use DImode rather than
17278 VOIDmode for the LABEL_REF.
17280 2017-08-22 Richard Biener <rguenther@suse.de>
17282 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
17284 2017-08-22 Richard Biener <rguenther@suse.de>
17286 * common.opt (feliminate-dwarf2-dups): Ignore.
17287 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
17288 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
17289 same_die_p_wrap, compute_section_prefix,
17290 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
17291 (comdat_symbol_id, comdat_symbol_number): Likewise.
17292 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
17294 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
17295 (output_die): Mark unreachable path unreachable.
17296 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
17297 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
17298 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
17299 (dwarf2out_early_finish): Likewise.
17301 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
17303 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
17305 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
17308 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
17309 not VAR_P. Filter attribute warnings with OPT_Wattributes.
17310 (avr_attribute_table) <io, io_low, address>: Initialize
17311 .decl_required with true.
17313 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
17315 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
17316 undocumented debugging options.
17317 (-mvsx-scalar-double): Likewise.
17318 (-mallow-df-permute): Likewise.
17319 (-mvectorize-builtins): Likewise.
17320 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17321 (rs6000_builtin_vectorized_function): Likewise.
17322 (rs6000_builtin_md_vectorized_function): Likewise.
17323 (rs6000_opt_vars): Likewise.
17325 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
17328 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
17329 (*btrq_imm): Rename from *btrq.
17330 (*btcq_imm): Rename from *btcq.
17331 (btsc): New code attribute.
17332 (*<btsc><mode>): New insn pattern.
17333 (*btr<mode>): Ditto.
17334 (*<btsc><mode>_mask): New insn_and_split pattern.
17335 (*btr<mode>_mask): Ditto.
17337 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17338 Alan Hayward <alan.hayward@arm.com>
17339 David Sherwood <david.sherwood@arm.com>
17341 * function.c (pad_below): Simplify padding calculation.
17343 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17344 Alan Hayward <alan.hayward@arm.com>
17345 David Sherwood <david.sherwood@arm.com>
17347 * target.def (function_prologue): Remove frame size argument.
17348 (function_epilogue): Likewise.
17349 * doc/tm.texi: Regenerate.
17350 * final.c (final_start_function): Update call to function_prologue.
17351 (final_end_function): Update call to function_epilogue.
17352 (default_function_pro_epilogue): Remove frame size argument.
17353 * output.h (default_function_pro_epilogue): Likewise.
17354 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
17355 (arm_output_function_prologue): Likewise.
17356 * config/frv/frv.c (frv_function_prologue): Likewise.
17357 (frv_function_epilogue): Likewise.
17358 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
17359 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
17360 (ia64_output_function_epilogue): Likewise.
17361 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
17362 (m32r_output_function_epilogue): Likewise.
17363 * config/microblaze/microblaze.c (microblaze_function_prologue)
17364 (microblaze_function_epilogue): Likewise.
17365 * config/mips/mips.c (mips_output_function_prologue): Likewise.
17366 (mips_output_function_epilogue): Likewise.
17367 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
17368 (mmix_target_asm_function_epilogue): Likewise.
17369 * config/msp430/msp430.c (msp430_start_function): Likewise.
17370 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
17371 (nds32_asm_function_epilogue): Likewise.
17372 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
17373 * config/pa/pa.c (pa_output_function_prologue): Likewise.
17374 (pa_output_function_epilogue): Likewise.
17375 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
17376 (rs6000_output_function_epilogue): Likewise.
17377 * config/rl78/rl78.c (rl78_start_function): Likewise.
17378 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
17379 (rs6000_output_function_epilogue): Likewise.
17380 * config/rx/rx.c (rx_output_function_prologue): Likewise.
17381 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
17382 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
17383 (sparc_asm_function_epilogue): Likewise.
17385 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17387 * tree.h (type_has_mode_precision_p): New function.
17388 * convert.c (convert_to_integer_1): Use it.
17389 * expr.c (expand_expr_real_2): Likewise.
17390 (expand_expr_real_1): Likewise.
17391 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
17392 * match.pd: Likewise.
17393 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
17394 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
17395 * tree-tailcall.c (process_assignment): Likewise.
17396 * tree-vect-loop.c (vectorizable_reduction): Likewise.
17397 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
17398 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
17399 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17400 (vectorizable_assignment): Likewise.
17401 (vectorizable_shift): Likewise.
17402 (vectorizable_operation): Likewise.
17403 * tree-vrp.c (register_edge_assert_for_2): Likewise.
17405 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
17407 * match.pd: Add pow (C, x) simplification.
17409 2017-08-21 Richard Biener <rguenther@suse.de>
17411 PR tree-optimization/81900
17412 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
17413 for blocks with abnormal predecessors.
17414 (compute_antic): Do not set visited flag prematurely.
17416 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
17419 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
17421 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17423 * stor-layout.h (vector_type_mode): Move to...
17424 * tree.h (vector_type_mode): ...here.
17425 * stor-layout.c (vector_type_mode): Move to...
17426 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
17428 2017-08-21 Richard Biener <rguenther@suse.de>
17430 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
17431 register_external_die hooks.
17432 (debug_false_tree_charstarstar_uhwistar): Declare.
17433 (debug_nothing_tree_charstar_uhwi): Likewise.
17434 * debug.c (do_nothing_debug_hooks): Adjust.
17435 (debug_false_tree_charstarstar_uhwistar): New do nothing.
17436 (debug_nothing_tree_charstar_uhwi): Likewise.
17437 * dbxout.c (dbx_debug_hooks): Adjust.
17438 (xcoff_debug_hooks): Likewise.
17439 * sdbout.c (sdb_debug_hooks): Likewise.
17440 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
17441 * dwarf2out.c (macinfo_label_base): New global.
17442 (dwarf2out_register_external_die): New function for the
17443 register_external_die hook.
17444 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
17445 (dwarf2_debug_hooks): Use them.
17446 (dwarf2_lineno_debug_hooks): Adjust.
17447 (struct die_struct): Add with_offset flag.
17448 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
17449 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
17450 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
17451 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
17452 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
17453 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
17454 defining section names for the early LTO debug variants.
17455 (reset_indirect_string): New helper.
17456 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
17457 (print_dw_val): Add support for offsetted symbol references.
17458 (get_ultimate_context): Split out from is_cxx.
17459 (is_cxx): Use get_ultimate_context.
17460 (is_fortran): Add decl overload.
17461 (compute_comp_unit_symbol): Split out worker from
17462 compute_section_prefix.
17463 (compute_section_prefix): Call compute_comp_unit_symbol and
17464 set comdat_type_p here.
17465 (output_die): Skip DIE symbol output for the LTO added one.
17466 Handle DIE symbol references with offset.
17467 (output_comp_unit): Guard section name mangling properly.
17468 For LTO debug sections emit a symbol at the section beginning
17469 which we use to refer to its DIEs.
17470 (add_abstract_origin_attribute): For DIEs registered via
17471 dwarf2out_register_external_die directly refer to the early
17472 DIE rather than indirectly through the shadow one we created.
17473 Remove obsolete call to dwarf2out_abstract_function for
17474 non-function/block origins.
17475 (gen_array_type_die): When generating early LTO debug do
17476 not emit DW_AT_string_length.
17477 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
17478 late when in LTO. As suggested place a gcc_unreachable for
17479 the DECL_ABSTRACT_P case.
17480 (gen_subprogram_die): Avoid another specification DIE
17481 for early built declarations/definitions for the late LTO case.
17482 (gen_variable_die): Add type references for late duplicated VLA dies
17484 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
17485 we have the abstract instance already.
17486 (process_scope_var): Adjust decl DIE contexts in LTO which
17487 first puts them in limbo.
17488 (gen_decl_die): Do not generate type DIEs late apart from
17489 types for VLAs or for decls we do not yet have a DIE. Do not
17490 call dwarf2out_abstract_function late.
17491 (dwarf2out_early_global_decl): Make sure to create DIEs
17492 for abstract instances of a decl first.
17493 (dwarf2out_late_global_decl): Adjust comment.
17494 (output_macinfo_op): With multiple macro sections use
17495 macinfo_label_base to distinguish labels.
17496 (output_macinfo): Likewise. Update macinfo_label_base.
17497 Pass in the line info label.
17498 (note_variable_value_in_expr): When generating LTO resolve
17499 all variable values here by generating DIEs as needed.
17500 (init_sections_and_labels): Add early LTO debug flag parameter
17501 and generate different sections and names if set. Add generation
17502 counter for the labels so we can have multiple of them.
17503 (reset_dies): Helper to allow DIEs to be output multiple times.
17504 (dwarf2out_finish): When outputting DIEs to the fat part of an
17505 LTO object first reset DIEs.
17506 (dwarf2out_early_finish): Output early DIEs when generating LTO.
17507 (modified_type_die): Check for decl_ultimate_origin being self
17509 (gen_type_die_with_usage): Likewise.
17510 (gen_typedef_die): Allow decl_ultimate_origin being self.
17511 (set_decl_abstract_flags): Remove.
17512 (set_block_abstract_flags): Likewise.
17513 (dwarf2out_abstract_function): Treat the early generated DIEs
17514 as the abstract copy and only add DW_AT_inline and
17515 DW_AT_artificial here and call set_decl_origin_self.
17516 If the DIE has an abstract origin don't do anything.
17517 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
17518 if we have none yet (Go fails to build one, PR78628).
17519 (variably_modified_type_p): Prevent endless recursion for Ada
17520 cyclic pointer types.
17521 * lto-streamer-in.c: Include debug.h.
17522 (dref_queue): New global.
17523 (lto_read_tree_1): Stream in DIE references.
17524 (lto_input_tree): Register DIE references.
17525 (input_function): Stream DECL_DEBUG_ARGS.
17526 * lto-streamer-out.c: Include debug.h.
17527 (lto_write_tree_1): Output DIE references.
17528 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
17529 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
17530 (output_function): Stream DECL_DEBUG_ARGS.
17531 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
17532 Stream DECL_ABSTRACT_ORIGIN.
17533 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
17534 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
17535 DECL_CONTEXT for file-scope decls.
17536 * lto-streamer.h (struct dref_entry): Declare.
17537 (dref_queue): Likewise.
17538 * cfgexpand.c (pass_expand::execute): Do not call the
17539 outlining_inline_function hook here.
17540 * lto-wrapper.c (debug_obj): New global.
17541 (tool_cleanup): Unlink it if required.
17542 (debug_objcopy): New function.
17543 (run_gcc): Handle early debug sections in the IL files by
17544 extracting them to separate files, partially linkin them and
17545 feeding the result back as result to the linker.
17546 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
17547 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
17548 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
17549 sections into a separate segment.
17550 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
17552 (darwin_asm_dwarf_section): Likewise.
17553 (darwin_asm_output_dwarf_offset): Likewise.
17554 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
17556 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17557 Alan Hayward <alan.hayward@arm.com>
17558 David Sherwood <david.sherwood@arm.com>
17560 * read-md.h (md_reader::record_potential_iterator_use): Replace
17561 pointer argument with an rtx and an index.
17562 * read-rtl.c (iterator_group::apply_iterator): Likewise.
17563 (apply_mode_iterator): Likewise.
17564 (apply_code_iterator): Likewise.
17565 (apply_int_iterator): Likewise.
17566 (apply_subst_iterator): Likewise.
17567 (record_iterator_use): Likewise.
17568 (record_attribute_use): Likewise.
17569 (md_reader::record_potential_iterator_use): Likewise. Update calls
17570 to record_iterator_use and apply_iterator.
17571 (iterator_use): Replace ptr with x and index.
17572 (attribute_use): Likewise.
17573 (apply_attribute_uses): Update calls to apply_iterator.
17574 (apply_iterators): Likewise. Update initialization of iterator_use.
17575 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
17576 and record_potential_iterator_use.
17577 (rtx_reader::read_rtx_operand): Likewise.
17579 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17580 Alan Hayward <alan.hayward@arm.com>
17581 David Sherwood <david.sherwood@arm.com>
17583 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
17586 2017-08-21 Richard Biener <rguenther@suse.de>
17588 PR middle-end/81884
17589 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
17590 at struct end conservatively when comparing common bases.
17592 2017-08-21 Richard Biener <rguenther@suse.de>
17594 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
17595 (mem_ref_in_stmt): Remove.
17596 (determine_max_movement): Use ref index to get at the reference.
17597 (invariantness_dom_walker::before_dom_children): Deal with
17598 lim data already initialized.
17599 (gather_mem_refs_stmt): Initialize lim data and record ref index.
17601 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
17603 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
17604 (TARGET_ISA_ROUND): Ditto.
17605 (TARGET_ROUND): Ditto.
17606 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
17607 * config/i386/i386.md: Ditto.
17608 * config/i386/sse.md: Ditto.
17609 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
17610 with OPTION_MASK_ISA_SSE4_1.
17612 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
17615 * doc/extend.texi (x86 Built-in Functions): Correct the name of
17616 __builtin_ia32_lzcnt_u16.
17618 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
17621 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
17622 (rs6000_set_current_function): Rewrite function to use it.
17624 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
17627 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
17628 and TYPE_WARN_IF_NOT_ALIGN.
17629 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
17630 (handle_warn_if_not_align): New.
17631 (place_union_field): Call handle_warn_if_not_align.
17632 (place_field): Call handle_warn_if_not_align.
17633 Copy TYPE_WARN_IF_NOT_ALIGN.
17634 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
17635 (layout_type): Likewise.
17636 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
17638 (tree_decl_common): Add warn_if_not_align.
17639 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
17640 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
17641 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
17642 (DECL_WARN_IF_NOT_ALIGN): Likewise.
17643 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
17644 * doc/extend.texi: Document warn_if_not_aligned attribute.
17645 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
17647 2017-08-17 Martin Liska <mliska@suse.cz>
17650 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
17651 (get_data_dependence): Use it as pointer type.
17652 (distribute_loop): Likewise.
17654 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17656 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
17657 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
17658 (p8_vmrgow_<mode>_direct): New define_insn.
17659 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
17660 handle endianness for vmrgew and vmrgow permute patterns.
17662 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
17664 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
17665 * config/rs6000/rs6000-cpus.def: Remove comment.
17666 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
17667 (POWERPC_MASKS): Likewise.
17668 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
17669 use of TARGET_VSX_TIMODE.
17670 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
17671 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
17672 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
17673 (rs6000_option_override_internal): Remove dead code.
17674 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
17675 (rs6000_legitimize_reload_address): Likewise.
17676 (rs6000_legitimate_address_p): Likewise.
17677 (rs6000_opt_masks): Delete "vsx-timode".
17678 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
17679 from function comment.
17680 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
17681 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
17682 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
17684 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
17685 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
17686 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
17687 (VSX_TI): Likewise.
17689 (define_peephole2): Likewise.
17691 2017-08-17 Martin Sebor <msebor@redhat.com>
17694 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
17695 past the end of an array.
17696 (test_pp_format): Add test cases.
17698 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
17700 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
17701 missing ECF_NOTHROW flags.
17703 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
17706 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
17707 operands residing in integer registers.
17708 (*vsx_le_perm_load_<mode>): Likewise.
17709 (*vsx_le_perm_store_<mode>): Likewise.
17710 (define_peephole2): Add peepholes to optimize the above.
17712 2017-08-17 Marek Polacek <polacek@redhat.com>
17714 PR middle-end/81814
17715 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
17716 to mimic what shorten_compare did. Change the return type to bool.
17717 (fold_cond_expr_with_comparison): Update call to
17718 operand_equal_for_comparison_p.
17719 (fold_ternary_loc): Likewise.
17721 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
17723 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
17725 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
17727 2017-08-17 Richard Biener <rguenther@suse.de>
17729 * tree-ssa-structalias.c (solve_graph): When propagating
17730 to successors update the graphs succ edges and avoid duplicate work.
17732 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
17735 * config/i386/i386.c (ix86_option_override_internal): Save target
17736 specific options after ix86_stack_protector_guard_reg was changed.
17738 2017-08-17 Richard Biener <rguenther@suse.de>
17740 PR tree-optimization/81827
17741 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
17742 (new_var_info): Initialize it conservatively.
17743 (get_call_vi): Mark register vars.
17744 (new_scalar_tmp_constraint_exp): Likewise.
17745 (handle_rhs_call): Likewise.
17746 (handle_const_call): Likewise.
17747 (create_function_info_for): Likewise.
17748 (solve_constraints): Sort varinfos to separate register from
17749 non-register vars to pack points-to solution bitmaps during
17752 2017-08-17 Marek Polacek <polacek@redhat.com>
17754 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
17756 2017-08-17 Richard Biener <rguenther@suse.de>
17758 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
17759 to true when overflow is undefined and we saturated the result.
17761 2017-08-17 Alan Modra <amodra@gmail.com>
17764 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
17765 Don't use store multiple if only one reg needs saving.
17766 (interesting_frame_related_regno): New function.
17767 (rs6000_frame_related): Don't emit frame info for regs that
17769 (rs6000_emit_epilogue): Likewise.
17771 2017-08-16 Nathan Sidwell <nathan@acm.org>
17773 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
17774 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
17775 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
17776 * tree.c (free_lang_data_in_type): Use else-if chain. Always
17777 clear TYPE_LANG_1. Remove obsolete member-function stripping.
17778 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
17779 (verify_type): Adjust for TYPE_BINFO move.
17780 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
17781 process TYPE_BINFO directly.
17782 (hash_tree): Likewise.
17783 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17785 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17788 2017-08-16 David Malcolm <dmalcolm@redhat.com>
17790 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
17792 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
17795 * config/i386/i386.md (*anddi_1_btr): Change predicates of
17796 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
17797 Add ix86_binary_operator_ok to insn constraint.
17798 (*iordi_1_bts): Ditto.
17799 (*xordi_1_btc): Ditto.
17800 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
17801 Update corresponding peephole2 pattern.
17805 2017-08-16 Bin Cheng <bin.cheng@arm.com>
17807 PR tree-optimization/81832
17808 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
17809 copy loop header which has IFN_LOOP_DIST_ALIAS call.
17811 2017-08-16 Marek Polacek <polacek@redhat.com>
17814 * fold-const.c (fold_indirect_ref_1): Restore original behavior
17815 regarding size_zero_node.
17817 2017-08-16 Martin Liska <mliska@suse.cz>
17820 * config.gcc: Respect previously set extra_objs in case
17823 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
17825 PR tree-optimization/81835
17826 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
17827 the phi SSA_NAME. Check that the condition in a COND_EXPR does
17828 not depend on the phi.
17830 2017-08-16 Alan Modra <amodra@gmail.com>
17832 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
17835 2017-08-16 Alan Modra <amodra@gmail.com>
17837 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
17838 (save_reg_p): ..into this. Update all callers.
17839 (first_reg_to_save): Simplify.
17841 2017-08-16 Alan Modra <amodra@gmail.com>
17843 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
17846 2017-08-15 Joseph Myers <joseph@codesourcery.com>
17850 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
17851 constant count if that count is less than 32.
17853 2017-08-15 Nathan Sidwell <nathan@acm.org>
17855 * gcc.c (execute): Emit friendlier message if inferior is killed
17856 by an external cause.
17858 2017-08-15 Richard Biener <rguenther@suse.de>
17860 PR tree-optimization/81790
17861 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
17862 CONSTRUCTORs from simplifying and VN.
17864 2017-08-14 Martin Sebor <msebor@redhat.com>
17866 * builtin-attrs.def: Add comments.
17868 2017-08-14 Martin Sebor <msebor@redhat.com>
17871 * doc/extend.texi (attribute nonstring): Document new attribute.
17873 2017-08-14 Martin Sebor <msebor@redhat.com>
17876 * tree-diagnostic.c (default_tree_printer): Handle %G.
17877 * gimple-pretty-print.h (percent_G_format): Declare new function.
17878 * gimple-pretty-print.c (percent_G_format): Define.
17879 * tree-pretty-print.c (percent_K_format): Add argument.
17881 2017-08-14 Martin Sebor <msebor@redhat.com>
17883 PR translation/79998
17884 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
17885 Remove a stray space.
17887 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
17890 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
17891 (*iordi_1_bts): Ditto.
17892 (*xordi_1_btc): Ditto.
17894 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17897 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
17898 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17900 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
17901 quoted strings, and make more translator-friendly.
17902 (darwin_rs6000_override_options): Likewise.
17903 (rs6000_option_override_internal): Likewise.
17904 (rs6000_return_in_memory): Fix overlong line.
17905 (init_cmulative_args): Use quoted strings, and make more
17906 translator-friendly.
17907 (rs6000_pass_by_reference): Fix overlong line.
17908 (def_builtin): Use quoted strings.
17909 (altivec_expand_predicate_builtin): Use quoted strings, and make
17910 more translator-friendly.
17911 (htm_expand_builtin): Use quoted strings.
17912 (cpu_expand_builtin): Use quoted strings, and make more
17913 translator-friendly.
17914 (altivec_expand_builtin): Likewise.
17915 (paired_expand_predicate_builtin): Likewise.
17916 (rs6000_invalid_builtin): Likewise.
17917 (builtin_function_type): Use quoted strings.
17918 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
17919 more translator-friendly.
17920 (rs6000_trampoline_init): Likewise.
17921 (rs6000_handle_altivec_attribute): Likewise.
17922 (rs6000_inner_target_options): Use quoted strings.
17923 (rs6000_disable_incompatible_switches): Likewise.
17924 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
17925 strings, and make more translator-friendly.
17926 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
17928 2017-08-14 Bin Cheng <bin.cheng@arm.com>
17930 PR tree-optimization/81799
17931 * tree-loop-distribution.c (version_loop_by_alias_check): Force
17932 cond_expr to simple gimple operand.
17934 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
17936 PR middle-end/46932
17937 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
17939 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
17943 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
17944 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
17945 TARGET_GASISR_PROLOGUES.
17946 * config/avr/avr.c (avr_option_override): Same.
17947 (avr_pass_pre_proep::execute): Same.
17949 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
17952 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
17953 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
17955 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
17957 * config/i386/i386.md (*load_tp_<mode>): Redefine as
17958 define_insn_and_split. Split to a memory load from 0 in
17959 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
17960 using PTR mode iterator.
17961 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
17962 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
17963 (*add_tp_<mode>): Redefine as define_insn_and_split.
17964 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
17965 address space. Merge with *add_tp_x32 using PTR mode iterator.
17966 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
17967 Split to an add with a memory load from 0 in
17968 DEFAULT_TLS_SEG_REG address space.
17970 2017-08-12 Andrew Pinski <apinski@cavium.com>
17972 * config/aarch64/aarch64-option-extensions.def (rdma):
17973 Fix feature string to what Linux prints out in /proc/cpuinfo.
17975 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
17978 * dwarf2out.c (modified_type_die): For C typedef types that have
17979 an ultimate origin, process the ultimate origin instead of the
17981 (gen_typedef_die): Assert that input DECLs have no ultimate
17983 (gen_type_die_with_usage): For typedef variants that have an
17984 ultimate origin, just call gen_decl_die on the original DECL.
17985 (process_scope_var): Avoid creating DIEs for local typedefs and
17986 concrete static variables.
17988 2017-08-12 Alan Modra <amodra@gmail.com>
17992 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
17993 match gnu-user.h startfile.
17994 (ENDFILE_LINUX_SPEC): Similarly.
17996 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
17999 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
18001 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
18003 2017-08-11 Tamar Christina <tamar.christina@arm.com>
18004 * config/aarch64/aarch64.md (mov<mode>): Change.
18005 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
18006 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
18007 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
18009 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
18011 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
18012 for storage order barriers.
18014 2017-08-11 Martin Liska <mliska@suse.cz>
18017 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
18018 variables of void type.
18020 2017-08-11 Martin Liska <mliska@suse.cz>
18022 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
18023 TARGET_SUPPORTS_ALIASES.
18024 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
18025 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
18026 (optimize_weakref): Likewise.
18027 * symtab.c (symtab_node::noninterposable_alias): Likewise.
18028 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
18029 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
18031 2017-08-11 Martin Liska <mliska@suse.cz>
18034 * config/i386/i386.c (make_resolver_func): Do complete
18035 refactoring of the function.
18037 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
18040 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
18041 * config/i386/i386.c (ix86_stack_protect_guard): Use
18042 ix86_stack_protect_guard_symbol_str to generate varible declaration.
18043 * doc/invoke.texi (x86 Options): Document
18044 -mstack-protector-guard-symbol= option.
18046 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
18048 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
18049 * config/i386/i386.c (ix86_split_stack_guard): New function.
18050 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
18051 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
18052 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
18053 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
18054 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
18055 (split_stack_space_check): Call ix86_split_stack_guard.
18057 2017-08-10 Martin Sebor <msebor@redhat.com>
18059 * print-tree.c (print_node): Print location using the established
18061 Replace spaces with colons.
18062 (debug_raw, debug): Ditto.
18064 2017-08-10 Martin Sebor <msebor@redhat.com>
18067 * pretty-print.c (pp_format): Correct the handling of %s precision.
18069 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
18072 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18074 (ix86_finalize_stack_frame_flags): This. Also clear
18075 frame_pointer_needed if -fno-omit-frame-pointer is used without
18077 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18078 with ix86_finalize_stack_frame_flags.
18079 (ix86_expand_epilogue): Likewise.
18080 (ix86_expand_split_stack_prologue): Likewise.
18081 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
18083 2017-08-10 Martin Liska <mliska@suse.cz>
18086 * c-attribs.c (handle_target_attribute):
18087 Report warning for an empty string argument of target attribute.
18089 2017-08-09 Jakub Jelinek <jakub@redhat.com>
18092 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
18094 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
18095 or DECL_NONLOCAL labels.
18096 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
18097 or DECL_NONLOCAL labels here.
18099 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
18101 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
18102 to indicate when early gimple folding has been disabled.
18103 (rs6000_gimple_fold_builtin): Add debug content.
18104 (rs6000_invalid_builtin): Fix whitespace.
18105 (rs6000_expand_builtin): Fix whitespace.
18106 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
18108 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
18111 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
18112 SAVE_MULTIPLE if not all the registers that saves, should be saved.
18114 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
18116 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
18117 (qdf24xx): Likewise.
18118 * config/aarch64/aarch64.md: Include falkor.md.
18119 * config/aarch64/falkor.md: New.
18121 2017-08-09 Marek Polacek <polacek@redhat.com>
18124 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
18125 * diagnostic.c (emit_diagnostic): Add a comment.
18126 (emit_diagnostic_valist): New function.
18128 2017-08-09 Marek Polacek <polacek@redhat.com>
18131 * input.c (make_location): New overload.
18132 * input.h (make_location): Declare.
18134 2017-08-08 Alan Modra <amodra@gmail.com>
18135 H.J. Lu <hongjiu.lu@intel.com>
18138 * gcc.c (NO_PIE_SPEC): Delete.
18139 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
18141 (LINK_PIE_SPEC): ..to here.
18142 (LINK_COMMAND_SPEC): Support -no-pie.
18143 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
18144 chain of crtbegin*.o selection, update for PIE_SPEC changes and
18146 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
18147 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
18148 (ENDFILE_CRTEND_SPEC): Similarly.
18150 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
18153 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
18154 (mstack-protector-guard-offset=): Ditto.
18155 * config/i386/i386.c (ix86_option_override): Handle
18156 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
18158 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
18159 ix86_stack_protect_guard_offset variables.
18160 (TARGET_STACK_PROTECT_GUARD): Always define.
18161 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
18162 and -mstack-protector-guard-offset= options.
18164 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18166 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
18167 boundary case for the last candidate.
18169 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18171 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
18172 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
18174 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18176 PR middle-end/19706
18177 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
18178 * config/aarch64/aarch64-builtins.c
18179 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
18180 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
18181 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
18183 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18184 Andrew Pinski <pinskia@gmail.com>
18186 PR middle-end/19706
18187 * internal-fn.def (XORSIGN): New.
18188 * optabs.def (xorsign_optab): New.
18189 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
18190 (convert_expand_mult_copysign): New.
18191 (pass_optimize_widening_mul::execute): Call
18192 convert_expand_mult_copysign.
18194 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18196 PR tree-optimization/81354
18197 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
18198 Insert on edges rather than explicitly creating landing pads.
18199 (analyze_candidates_and_replace): Commit edge inserts.
18201 2017-08-08 Richard Biener <rguenther@suse.de>
18203 PR middle-end/81719
18204 * tree-ssa-loop-niter.c: Include tree-dfa.h.
18205 (expand_simple_operations): Also look through ADDR_EXPRs with
18206 MEM_REF bases treating them as POINTER_PLUS_EXPR.
18208 2017-08-08 Richard Biener <rguenther@suse.de>
18210 PR tree-optimization/81723
18211 * tree-vect-slp.c (struct bst_traits): New hash traits.
18212 (bst_fail): New global.
18213 (vect_build_slp_tree_2): New worker, split out from ...
18214 (vect_build_slp_tree): ... this now wrapping it with using
18215 bst_fail set to cache SLP tree build fails. Properly handle
18217 (vect_analyze_slp_instance): Allocate and free bst_fail.
18219 2017-08-08 Martin Liska <mliska@suse.cz>
18222 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
18223 LABEL_DECLs that can be from a different function.
18225 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18227 PR tree-optimization/81744
18228 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
18229 loop's number of iterations.
18231 2017-08-08 Martin Liska <mliska@suse.cz>
18233 * asan.c: Include header files.
18234 * attribs.c (build_decl_attribute_variant): New function moved
18236 (build_type_attribute_qual_variant): Likewise.
18237 (cmp_attrib_identifiers): Likewise.
18238 (simple_cst_list_equal): Likewise.
18239 (omp_declare_simd_clauses_equal): Likewise.
18240 (attribute_value_equal): Likewise.
18241 (comp_type_attributes): Likewise.
18242 (build_type_attribute_variant): Likewise.
18243 (lookup_ident_attribute): Likewise.
18244 (remove_attribute): Likewise.
18245 (merge_attributes): Likewise.
18246 (merge_type_attributes): Likewise.
18247 (merge_decl_attributes): Likewise.
18248 (merge_dllimport_decl_attributes): Likewise.
18249 (handle_dll_attribute): Likewise.
18250 (attribute_list_equal): Likewise.
18251 (attribute_list_contained): Likewise.
18252 * attribs.h (lookup_attribute): New function moved from tree.[ch].
18253 (lookup_attribute_by_prefix): Likewise.
18254 * bb-reorder.c: Include header files.
18255 * builtins.c: Likewise.
18256 * calls.c: Likewise.
18257 * cfgexpand.c: Likewise.
18258 * cgraph.c: Likewise.
18259 * cgraphunit.c: Likewise.
18260 * convert.c: Likewise.
18261 * dwarf2out.c: Likewise.
18262 * final.c: Likewise.
18263 * fold-const.c: Likewise.
18264 * function.c: Likewise.
18265 * gimple-expr.c: Likewise.
18266 * gimple-fold.c: Likewise.
18267 * gimple-pretty-print.c: Likewise.
18268 * gimple.c: Likewise.
18269 * gimplify.c: Likewise.
18270 * hsa-common.c: Likewise.
18271 * hsa-gen.c: Likewise.
18272 * internal-fn.c: Likewise.
18273 * ipa-chkp.c: Likewise.
18274 * ipa-cp.c: Likewise.
18275 * ipa-devirt.c: Likewise.
18276 * ipa-fnsummary.c: Likewise.
18277 * ipa-inline.c: Likewise.
18278 * ipa-visibility.c: Likewise.
18280 * lto-cgraph.c: Likewise.
18281 * omp-expand.c: Likewise.
18282 * omp-general.c: Likewise.
18283 * omp-low.c: Likewise.
18284 * omp-offload.c: Likewise.
18285 * omp-simd-clone.c: Likewise.
18286 * opts-global.c: Likewise.
18287 * passes.c: Likewise.
18288 * predict.c: Likewise.
18289 * sancov.c: Likewise.
18290 * sanopt.c: Likewise.
18291 * symtab.c: Likewise.
18292 * toplev.c: Likewise.
18293 * trans-mem.c: Likewise.
18294 * tree-chkp.c: Likewise.
18295 * tree-eh.c: Likewise.
18296 * tree-into-ssa.c: Likewise.
18297 * tree-object-size.c: Likewise.
18298 * tree-parloops.c: Likewise.
18299 * tree-profile.c: Likewise.
18300 * tree-ssa-ccp.c: Likewise.
18301 * tree-ssa-live.c: Likewise.
18302 * tree-ssa-loop.c: Likewise.
18303 * tree-ssa-sccvn.c: Likewise.
18304 * tree-ssa-structalias.c: Likewise.
18305 * tree-ssa.c: Likewise.
18306 * tree-streamer-in.c: Likewise.
18307 * tree-vectorizer.c: Likewise.
18308 * tree-vrp.c: Likewise.
18309 * tsan.c: Likewise.
18310 * ubsan.c: Likewise.
18311 * varasm.c: Likewise.
18312 * varpool.c: Likewise.
18313 * tree.c: Remove functions moved to attribs.[ch].
18314 * tree.h: Likewise.
18315 * config/aarch64/aarch64.c: Add attrs.h header file.
18316 * config/alpha/alpha.c: Likewise.
18317 * config/arc/arc.c: Likewise.
18318 * config/arm/arm.c: Likewise.
18319 * config/avr/avr.c: Likewise.
18320 * config/bfin/bfin.c: Likewise.
18321 * config/c6x/c6x.c: Likewise.
18322 * config/cr16/cr16.c: Likewise.
18323 * config/cris/cris.c: Likewise.
18324 * config/darwin.c: Likewise.
18325 * config/epiphany/epiphany.c: Likewise.
18326 * config/fr30/fr30.c: Likewise.
18327 * config/frv/frv.c: Likewise.
18328 * config/ft32/ft32.c: Likewise.
18329 * config/h8300/h8300.c: Likewise.
18330 * config/i386/winnt.c: Likewise.
18331 * config/ia64/ia64.c: Likewise.
18332 * config/iq2000/iq2000.c: Likewise.
18333 * config/lm32/lm32.c: Likewise.
18334 * config/m32c/m32c.c: Likewise.
18335 * config/m32r/m32r.c: Likewise.
18336 * config/m68k/m68k.c: Likewise.
18337 * config/mcore/mcore.c: Likewise.
18338 * config/microblaze/microblaze.c: Likewise.
18339 * config/mips/mips.c: Likewise.
18340 * config/mmix/mmix.c: Likewise.
18341 * config/mn10300/mn10300.c: Likewise.
18342 * config/moxie/moxie.c: Likewise.
18343 * config/msp430/msp430.c: Likewise.
18344 * config/nds32/nds32-isr.c: Likewise.
18345 * config/nds32/nds32.c: Likewise.
18346 * config/nios2/nios2.c: Likewise.
18347 * config/nvptx/nvptx.c: Likewise.
18348 * config/pa/pa.c: Likewise.
18349 * config/pdp11/pdp11.c: Likewise.
18350 * config/powerpcspe/powerpcspe.c: Likewise.
18351 * config/riscv/riscv.c: Likewise.
18352 * config/rl78/rl78.c: Likewise.
18353 * config/rx/rx.c: Likewise.
18354 * config/s390/s390.c: Likewise.
18355 * config/sh/sh.c: Likewise.
18356 * config/sol2.c: Likewise.
18357 * config/sparc/sparc.c: Likewise.
18358 * config/spu/spu.c: Likewise.
18359 * config/stormy16/stormy16.c: Likewise.
18360 * config/tilegx/tilegx.c: Likewise.
18361 * config/tilepro/tilepro.c: Likewise.
18362 * config/v850/v850.c: Likewise.
18363 * config/vax/vax.c: Likewise.
18364 * config/visium/visium.c: Likewise.
18365 * config/xtensa/xtensa.c: Likewise.
18367 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18370 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
18371 constraints since the -mupper-regs-* switches have been
18373 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
18374 into a vector from a double word element that was extracted from
18375 another vector, and eliminate extra XXPERMDI instructions.
18376 (vsx_concat_<mode>_2): Likewise.
18377 (vsx_concat_<mode>_3): Likewise.
18378 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
18379 concat to allow optimizing inserts from previous extracts.
18381 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
18383 * config/i386/i386.c (ix86_stack_protect_guard): Generate
18384 memory reference to a SSP offset in TLS address space.
18385 (ix86_print_operand) <case '@'>: Remove.
18386 (ix86_print_operand_punct_valid_p): Remove '@' code.
18387 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
18388 UNSPEC_SP_TLS_TEST.
18389 (stack_tls_protect_set_<mode>): Remove.
18390 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
18391 (stack_tls_protect_test_<mode>): Remove.
18392 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
18394 2017-08-07 Olivier Hainque <hainque@adacore.com>
18397 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
18399 2017-08-07 Douglas Rupp <rupp@adacore.com>
18401 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
18402 variable was referenced as multidir in command.
18404 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18407 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
18410 2017-08-07 Martin Liska <mliska@suse.cz>
18412 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
18413 * config/rl78/rl78.c: Add include of attribs.h.
18414 * config/sh/sh.c: Likewise.
18415 * config/v850/v850.c: Likewise.
18417 2017-08-07 Tom de Vries <tom@codesourcery.com>
18419 PR middle-end/78266
18420 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
18422 2017-08-07 Martin Liska <mliska@suse.cz>
18424 * config/mips/mips.c: Include attribs.h.
18426 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
18429 * doc/invoke.texi: Document change in behvaior for -Ofast for
18432 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
18434 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
18436 (aarch64_pop_regs): Likewise.
18437 (aarch64_gen_load_pair): Likewise.
18438 (aarch64_save_callee_saves): Likewise.
18439 (aarch64_restore_callee_saves): Likewise.
18441 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18443 * config/i386/i386.c: Revert the last change.
18445 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18448 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18450 (ix86_finalize_stack_frame_flags): This. Also clear
18451 frame_pointer_needed if -fno-omit-frame-pointer is used without
18453 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18454 with ix86_finalize_stack_frame_flags.
18455 (ix86_expand_epilogue): Likewise.
18456 (ix86_expand_split_stack_prologue): Likewise.
18458 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18461 * config/i386/i386.c (get_builtin_code_for_version): Set priority
18462 to P_AES for Westmere.
18464 2017-08-07 Jonathan Yong <10walls@gmail.com>
18466 * config/i386/mingw.opt (fset-stack-executable): Removed.
18467 * config/i386/cygming.opt (fset-stack-executable): Moved
18469 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
18471 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
18473 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
18475 2017-08-07 Marek Polacek <polacek@redhat.com>
18477 PR middle-end/81737
18478 * fold-const.c (fold_indirect_ref_1): Check type_domain.
18480 2017-08-07 Martin Liska <mliska@suse.cz>
18482 * attribs.h (canonicalize_attr_name): New function.
18483 (cmp_attribs): Move from c-format.c and adjusted.
18484 (is_attribute_p): Moved from tree.h.
18485 * tree-inline.c: Add new includes.
18486 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
18487 (private_is_attribute_p): Remove.
18488 (private_lookup_attribute): Likewise.
18489 (private_lookup_attribute_by_prefix): Simplify.
18490 (remove_attribute): Use is_attribute_p.
18491 * tree.h: Remove removed declarations.
18493 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18495 PR middle-end/81698
18496 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
18497 instead of computing it in the function. Formatting fix.
18498 (expand_case): Don't rely on default_edge being the first edge,
18499 clear it if removing it, pass default_edge to
18500 emit_case_dispatch_table.
18501 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
18504 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
18506 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
18507 insn in the function, emit NOP after the insn.
18509 2017-08-06 Tom de Vries <tom@codesourcery.com>
18511 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
18514 2017-08-06 Tom de Vries <tom@codesourcery.com>
18516 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
18519 2017-08-04 Yury Gribov <tetra2005@gmail.com>
18521 PR tree-optimization/57371
18522 * match.pd: New pattern.
18524 2017-08-04 Marek Polacek <polacek@redhat.com>
18526 PR middle-end/81695
18527 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
18528 perform the computation in offset_int.
18530 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18532 PR tree-optimization/81136
18533 * tree-vectorizer.h: Include tree-hash-traits.h.
18534 (vec_base_alignments): New typedef.
18535 (vec_info): Add a base_alignments field.
18536 (vect_record_base_alignments): Declare.
18537 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
18539 (DR_IS_CONDITIONAL_IN_STMT): New macro.
18540 (create_data_ref): Add an is_conditional_in_stmt argument.
18541 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
18542 the is_conditional_in_stmt field.
18543 (data_ref_loc): Add an is_conditional_in_stmt field.
18544 (get_references_in_stmt): Set the is_conditional_in_stmt field.
18545 (find_data_references_in_stmt): Update call to create_data_ref.
18546 (graphite_find_data_references_in_stmt): Likewise.
18547 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
18548 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
18549 (vect_record_base_alignment): New function.
18550 (vect_record_base_alignments): Likewise.
18551 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
18552 for nested statements even if we fail to compute a misalignment.
18553 Use pooled base alignments for unconditional references.
18554 (vect_find_same_alignment_drs): Compare base addresses instead
18556 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
18557 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
18559 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18561 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
18562 Add an explicit name for the enum. Use auto_vec for slp_instances
18563 and grouped_stores.
18564 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
18566 (_bb_vec_info): Add a constructor and destructor.
18567 (vinfo_for_stmt): Return NULL for uids of -1 as well.
18568 (destroy_loop_vec_info): Delete.
18569 (vect_destroy_datarefs): Likewise.
18570 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
18571 (vec_info::vec_info): New function.
18572 (vec_info::~vec_info): Likewise.
18573 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
18574 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
18575 destroy_loop_vec_info.
18576 * tree-vect-loop.c (new_loop_vec_info): Replace with...
18577 (_loop_vec_info::_loop_vec_info): ...this.
18578 (destroy_loop_vec_info): Replace with...
18579 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
18580 the stmt_vec_infos. Leave handling of vec_info information to its
18581 destructor. Remove explicit vector releases.
18582 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
18583 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
18584 * tree-vect-slp.c (new_bb_vec_info): Replace with...
18585 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
18586 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
18587 (destroy_bb_vec_info): Replace with...
18588 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
18589 information to its destructor.
18590 (vect_slp_analyze_bb_1): Use new and delete instead of
18591 new_bb_vec_info and destroy_bb_vec_info.
18592 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
18595 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18597 * tree-data-ref.h (subscript): Add access_fn field.
18598 (data_dependence_relation): Add could_be_independent_p.
18599 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
18600 (same_access_functions): Move to tree-data-ref.c.
18601 * tree-data-ref.c (ref_contains_union_access_p): New function.
18602 (access_fn_component_p): Likewise.
18603 (access_fn_components_comparable_p): Likewise.
18604 (dr_analyze_indices): Add a reference to access_fn_component_p.
18605 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
18607 (constant_access_functions): Likewise.
18608 (add_other_self_distances): Likewise.
18609 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
18610 (initialize_data_dependence_relation): Use XCNEW and remove
18611 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
18612 of access functions that have the same type. Allow the
18613 subsequence to end with different bases in some circumstances.
18614 Record the chosen access functions in SUB_ACCESS_FN.
18615 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
18616 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
18617 (subscript_dependence_tester_1): Likewise dra and drb.
18618 (build_classic_dist_vector): Update calls accordingly.
18619 (subscript_dependence_tester): Likewise.
18620 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
18621 DDR_COULD_BE_INDEPENDENT_P.
18622 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
18623 comp_alias_ddrs instead of may_alias_ddrs.
18624 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
18626 (vect_analyze_data_ref_dependence): Use it if
18627 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
18628 distance vectors if that fails.
18629 (dependence_distance_ge_vf): New function.
18630 (vect_prune_runtime_alias_test_list): Use it. Don't clear
18631 LOOP_VINFO_MAY_ALIAS_DDRS.
18633 2017-08-04 Richard Biener <rguenther@suse.de>
18635 PR middle-end/81705
18636 * fold-const.c (fold_binary_loc): Properly restrict
18637 minus_var0 && minus_var1 case when associating undefined overflow
18640 2017-08-04 Tom de Vries <tom@codesourcery.com>
18642 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
18644 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18646 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18647 Don't start diagnostic messages with a capital letter.
18648 * config/rs6000/rs6000.c (rs6000_option_override_internal):
18650 (rs6000_invalid_builtin): Likewise.
18651 (rs6000_trampoline_init): Likewise.
18653 2017-08-03 Jakub Jelinek <jakub@redhat.com>
18656 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
18657 after setting changeable df flags.
18659 2017-08-03 Richard Biener <rguenther@suse.de>
18661 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
18662 up if the use is in USE - X.
18664 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
18666 * toplev.c (dumpfile.h): New include.
18667 (internal_error_reentered): New static function. Use it...
18668 (internal_error_function): ...here to handle reentered internal_error.
18670 2017-08-03 Richard Biener <rguenther@suse.de>
18672 PR middle-end/81148
18673 * fold-const.c (split_tree): Add minus_var and minus_con
18674 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
18675 here but always use minus_*.
18676 (associate_trees): Assert we never associate with MINUS_EXPR
18677 and NULL first operand. Do not recurse for PLUS_EXPR operands
18678 when associating as MINUS_EXPR either.
18679 (fold_binary_loc): Track minus_var and minus_con.
18681 2017-08-03 Tom de Vries <tom@codesourcery.com>
18684 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
18685 ACCEL_COMPILER, apply finish_options on
18686 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
18688 2017-08-03 Tom de Vries <tom@codesourcery.com>
18691 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
18692 function_entry_patch_area_size > 0.
18694 2017-08-03 Jakub Jelinek <jakub@redhat.com>
18697 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
18698 instead of 10??LU, perform unit multiplication in wide_int,
18699 don't change alloc_object_size_limit if the limit is larger
18702 PR tree-optimization/81655
18703 PR tree-optimization/81588
18704 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
18705 the case when ranges[i].low and high are 1 for unsigned type with
18708 PR middle-end/81052
18709 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
18710 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
18712 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18714 * tree-vrp.h: Add include guard.
18716 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
18719 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
18720 (ud2): New insn pattern.
18721 * config/i386/i386.c (ix86_expand_epilogue):
18722 For naked functions, generate ud2 instead of trap insn.
18724 2017-08-02 Marek Polacek <polacek@redhat.com>
18727 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
18729 2017-08-02 Tom de Vries <tom@codesourcery.com>
18730 Cesar Philippidis <cesar@codesourcery.com>
18732 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
18733 Add missing edge probabilities.
18735 2017-08-02 Tamar Christina <tamar.christina@arm.com>
18737 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
18738 Correct endianness.
18740 2017-08-02 Jakub Jelinek <jakub@redhat.com>
18742 PR middle-end/79499
18743 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
18744 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
18745 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
18746 prologue_seq sequences - if any.
18748 2017-08-02 Richard Biener <rguenther@suse.de>
18750 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
18751 via vectors if supported, integer extracts via punning if supported
18752 or otherwise vector extracts.
18754 2017-08-02 Richard Biener <rguenther@suse.de>
18756 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
18758 (bitmap_insert_into_set): ... this.
18760 2017-08-02 Richard Biener <rguenther@suse.de>
18762 PR tree-optimization/81633
18764 2015-08-17 Alan Hayward <alan.hayward@arm.com>
18766 PR tree-optimization/71752
18767 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
18769 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
18771 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
18772 (machine_function::call_ms2sysv_pad_out): Remove field.
18773 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
18774 (ix86_compute_frame_layout): Likewise.
18776 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
18779 * config/i386/i386.c (ix86_set_func_type): Disallow naked
18780 attribute with interrupt attribute.
18782 2017-08-01 Andrew Pinski <apinski@cavium.com>
18784 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
18785 BIT_INSERT_EXPR's operand 1
18786 to see if the types precision matches.
18788 2017-08-01 Martin Liska <mliska@suse.cz>
18790 PR middle-end/70140
18791 * builtins.c (expand_builtin_memcpy_args): Remove.
18792 (expand_builtin_memcpy): Call newly added function
18793 expand_builtin_memory_copy_args.
18794 (expand_builtin_memcpy_with_bounds): Likewise.
18795 (expand_builtin_mempcpy): Remove last argument.
18796 (expand_builtin_mempcpy_with_bounds): Likewise.
18797 (expand_builtin_memory_copy_args): New function created from
18798 expand_builtin_mempcpy_args with small modifications.
18799 (expand_builtin_mempcpy_args): Remove.
18800 (expand_builtin_stpcpy): Remove unused argument.
18801 (expand_builtin): Likewise.
18802 (expand_builtin_with_bounds): Likewise.
18804 2017-08-01 Martin Liska <mliska@suse.cz>
18807 Make mempcpy more optimal (PR middle-end/70140).
18809 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18812 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
18813 __builtin_vec_cmpne verify both arguments are compatible vectors
18814 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
18815 verify arg1_type is a pointer or array type. For __builtin_vec_st,
18816 move computation of aligned to after checking the argument types.
18820 * config/rs6000/vsx.md (vextract_fp_from_shorth,
18821 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
18824 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
18825 Doug Rupp <rupp@adacore.com>
18826 Olivier Hainque <hainque@adacore.com>
18828 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
18829 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
18831 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
18832 for TARGET_OS_CPP_BUILTIN.
18833 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
18834 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
18836 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
18837 passing required abi options to the assembler for EABI configurations.
18838 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
18839 of .text.hot and .text.unlikely sections for kernel modules when
18840 using ARM style exceptions.
18841 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
18842 options. Add EXTRA_CC1_SPEC.
18843 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
18845 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
18847 (ARM_TARGET2_DWARF_FORMAT): Define.
18848 * config/arm/t-vxworks: Adjust multilib control to removal of the
18849 Diab command line options.
18851 2017-08-01 Martin Liska <mliska@suse.cz>
18853 PR gcov-profile/81561
18854 * gcov.c (unblock): Make unblocking safe as we need to preserve
18855 index correspondence of blocks and block_lists.
18857 2017-08-01 Richard Biener <rguenther@suse.de>
18859 PR tree-optimization/81181
18860 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
18861 (compute_antic): ... end of iteration here.
18863 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
18865 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
18866 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
18867 (ftree-slp-vectorize): Likewise.
18868 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
18869 can no longer be set independent of flag_tree_loop_vectorize.
18870 * omp-general.c (emp_max_vf): Likewise.
18871 * opts.c (enable_fdo_optimizations): Remove references to
18872 flag_tree_vectorize, these are now implicit.
18873 (common_handle_option): Remove handling for OPT_ftree_vectorize,
18874 and leave it for the options machinery.
18876 2017-08-01 Martin Liska <mliska@suse.cz>
18878 PR middle-end/70140
18879 * builtins.c (expand_builtin_memcpy_args): Remove.
18880 (expand_builtin_memcpy): Call newly added function
18881 expand_builtin_memory_copy_args.
18882 (expand_builtin_memcpy_with_bounds): Likewise.
18883 (expand_builtin_mempcpy): Remove last argument.
18884 (expand_builtin_mempcpy_with_bounds): Likewise.
18885 (expand_builtin_memory_copy_args): New function created from
18886 expand_builtin_mempcpy_args with small modifications.
18887 (expand_builtin_mempcpy_args): Remove.
18888 (expand_builtin_stpcpy): Remove unused argument.
18889 (expand_builtin): Likewise.
18890 (expand_builtin_with_bounds): Likewise.
18892 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18895 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
18896 print "ds:" only for immediates in generic address space.
18898 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18901 * config/i386/i386.c (ix86_funciton_naked): New prototype.
18902 (ix86_function_ok_for_sibcall): Return false for naked functions.
18904 2017-08-01 Richard Biener <rguenther@suse.de>
18906 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
18907 (compute_antic): Seed worklist with exit block predecessors.
18908 * cfganal.c (dfs_find_deadend): For a cycle return the source
18909 of the edge closing it.
18911 2017-08-01 Tamar Christina <tamar.christina@arm.com>
18913 * config/aarch64/aarch64.c
18914 (aarch64_can_const_movi_rtx_p): Move 0 check.
18916 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18918 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
18919 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
18921 * match.pd: Ditto in address comparison pattern.
18923 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18925 PR tree-optimization/81627
18926 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
18927 closed ssa form for store-store chain.
18929 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18931 PR tree-optimization/81620
18932 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
18933 for store-store chain.
18935 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18937 PR tree-optimization/81588
18938 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
18939 ranges[i].in_p, invert comparison code ccode. For >/>=,
18940 swap rhs1 and rhs2 and comparison code unconditionally,
18941 for </<= don't do that. Don't swap rhs1/rhs2 again if
18942 ranges[i].in_p, instead invert comparison code ccode if
18943 opcode or oe->rank is BIT_IOR_EXPR.
18946 * optabs.def (vec_extract_optab, vec_init_optab): Change from
18947 a direct optab to conversion optab.
18948 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
18949 with GET_MODE_INNER as last argument instead of optab_handler.
18950 * expmed.c (extract_bit_field_1): Likewise. Use vector from
18951 vector extraction if possible and optab is available.
18952 * expr.c (store_constructor): Use convert_optab_handler instead
18953 of optab_handler. Use vector initialization from smaller
18954 vectors if possible and optab is available.
18955 * tree-vect-stmts.c (vectorizable_load): Likewise.
18956 * doc/md.texi (vec_extract, vec_init): Document that the optabs
18957 now have two modes.
18958 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
18959 of vec_init from half-sized vectors with the same element mode.
18960 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
18961 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
18962 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
18963 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
18964 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
18965 after mode in gen_vec_extract* calls.
18966 (vec_extract<mode>): Renamed to ...
18967 (vec_extract<mode><ssescalarmodelower>): ... this.
18968 (vec_extract<mode><ssehalfvecmodelower>): New expander.
18969 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
18970 element mode after mode in gen_vec_init* calls.
18971 (VEC_INIT_HALF_MODE): New mode iterator.
18972 (vec_init<mode>): Renamed to ...
18973 (vec_init<mode><ssescalarmodelower>): ... this.
18974 (vec_init<mode><ssehalfvecmodelower>): New expander.
18975 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
18976 (vec_extractv2sfsf): ... this.
18977 (vec_initv2sf): Renamed to ...
18978 (vec_initv2sfsf): ... this.
18979 (vec_extractv2si): Renamed to ...
18980 (vec_extractv2sisi): ... this.
18981 (vec_initv2si): Renamed to ...
18982 (vec_initv2sisi): ... this.
18983 (vec_extractv4hi): Renamed to ...
18984 (vec_extractv4hihi): ... this.
18985 (vec_initv4hi): Renamed to ...
18986 (vec_initv4hihi): ... this.
18987 (vec_extractv8qi): Renamed to ...
18988 (vec_extractv8qiqi): ... this.
18989 (vec_initv8qi): Renamed to ...
18990 (vec_initv8qiqi): ... this.
18991 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
18992 (vec_init<mode>): Renamed to ...
18993 (vec_init<mode><VEC_base_l>): ... this.
18994 (vec_extract<mode>): Renamed to ...
18995 (vec_extract<mode><VEC_base_l>): ... this.
18996 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
18997 (vec_initv2sfsf): ... this.
18998 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
18999 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
19000 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
19001 element mode after mode in gen_vec_init* calls.
19002 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
19003 (vec_init<mode><Vel>): ... this.
19004 (vec_extract<mode>): Renamed to ...
19005 (vec_extract<mode><Vel>): ... this.
19006 * config/aarch64/iterators.md (Vel): New mode attribute.
19007 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
19008 Add element mode after mode in gen_vec_extract* calls.
19009 * config/s390/vector.md (non_vec_l): New mode attribute.
19010 (vec_extract<mode>): Renamed to ...
19011 (vec_extract<mode><non_vec_l>): ... this.
19012 (vec_init<mode>): Renamed to ...
19013 (vec_init<mode><non_vec_l>): ... this.
19014 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
19015 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
19017 * config/arm/iterators.md (V_elem_l): New mode attribute.
19018 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
19019 (vec_extract<mode><V_elem_l>): ... this.
19020 (vec_extractv2di): Renamed to ...
19021 (vec_extractv2didi): ... this.
19022 (vec_init<mode>): Renamed to ...
19023 (vec_init<mode><V_elem_l>): ... this.
19024 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
19025 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
19026 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
19027 Add element mode after gen_vec_extract* calls.
19028 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
19029 (vec_init<mode><unitmode>): ... this.
19030 (vec_extract<mode>): Renamed to ...
19031 (vec_extract<mode><unitmode>): ... this.
19032 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
19033 (vec_init<mode><unitmode>): ... this.
19034 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
19035 (vec_initv2sfsf): ... this.
19036 (vec_extractv2sf): Renamed to ...
19037 (vec_extractv2sfsf): ... this.
19038 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
19039 Add element mode after gen_vec_extract* calls.
19040 * config/mips/mips.md (unitmode): New mode iterator.
19041 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
19042 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
19043 * config/spu/spu.md (inner_l): New mode attribute.
19044 (vec_init<mode>): Renamed to ...
19045 (vec_init<mode><inner_l>): ... this.
19046 (vec_extract<mode>): Renamed to ...
19047 (vec_extract<mode><inner_l>): ... this.
19048 * config/sparc/sparc.md (veltmode): New mode iterator.
19049 (vec_init<VMALL:mode>): Renamed to ...
19050 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
19051 * config/ia64/vect.md (vec_initv2si): Renamed to ...
19052 (vec_initv2sisi): ... this.
19053 (vec_initv2sf): Renamed to ...
19054 (vec_initv2sfsf): ... this.
19055 (vec_extractv2sf): Renamed to ...
19056 (vec_extractv2sfsf): ... this.
19057 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
19058 (vec_init<mode>): Renamed to ...
19059 (vec_init<mode><VEC_base_l>): ... this.
19060 (vec_extract<mode>): Renamed to ...
19061 (vec_extract<mode><VEC_base_l>): ... this.
19062 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
19063 (vec_initv2sfsf): ... this.
19064 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
19065 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
19066 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
19067 gen_vec_init* calls.
19069 2017-08-01 Richard Biener <rguenther@suse.de>
19071 PR tree-optimization/81297
19072 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
19073 TREE_OVERFLOW from INTEGER_CSTs.
19075 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
19077 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
19079 2017-07-31 Carl Love <cel@us.ibm.com>
19081 * config/rs6000/rs6000-c: Add support for built-in functions
19082 vector signed char vec_xl_be (signed long long, signed char *);
19083 vector unsigned char vec_xl_be (signed long long, unsigned char *);
19084 vector signed int vec_xl_be (signed long long, signed int *);
19085 vector unsigned int vec_xl_be (signed long long, unsigned int *);
19086 vector signed long long vec_xl_be (signed long long, signed long long *);
19087 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
19088 vector signed short vec_xl_be (signed long long, signed short *);
19089 vector unsigned short vec_xl_be (signed long long, unsigned short *);
19090 vector double vec_xl_be (signed long long, double *);
19091 vector float vec_xl_be (signed long long, float *);
19092 * config/rs6000/altivec.h (vec_xl_be): Add #define.
19093 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
19094 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
19096 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
19097 (altivec_expand_builtin): Add switch statement to call
19098 altivec_expand_xl_be for each builtin.
19099 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
19100 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
19101 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
19102 __builtin_vsx_le_be_v16qi.
19103 * doc/extend.texi: Update the built-in documentation file for the
19104 new built-in functions.
19106 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19109 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
19111 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
19113 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19115 * config.gcc: Add z14.
19116 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
19117 CPU model numbers for z13s and z14.
19118 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
19120 * config/s390/s390-opts.h (enum processor_type): Rename
19121 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19122 * config/s390/s390.c (processor_table): Add field for CPU name to
19123 be passed to Binutils.
19124 (s390_asm_output_machine_for_arch): Use the new field in
19125 processor_table for Binutils.
19126 (s390_expand_builtin): Replace arch12 with z14.
19127 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19128 (s390_get_sched_attrmask): Likewise.
19129 (s390_get_unit_mask): Likewise.
19130 * config/s390/s390.opt: Add z14 to processor_type enum.
19132 2017-07-31 Martin Jambor <mjambor@suse.cz>
19135 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
19136 regardless of optimization level.
19138 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
19139 Martin Liska <mliska@suse.cz>
19141 * predict.def: Remove old comment and adjust probability.
19142 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
19143 PREDICT statements.
19145 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19148 * config/i386/i386.c (ix86_function_naked): New function.
19149 (ix86_can_use_return_insn_p): Return false for naked functions.
19150 (ix86_expand_prologue): Skip prologue for naked functions.
19151 (ix86_expand_epilogue): Skip epilogue for naked functions
19152 and emit trap instruction.
19153 (ix86_warn_func_return): New function.
19154 (ix86_attribute_table): Add "naked" attribute specification.
19155 (TARGET_WARN_FUNC_RETURN): Define.
19156 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
19158 2017-07-31 Martin Liska <mliska@suse.cz>
19160 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
19161 (dump_gimple_bb_header): Always dump BB info.
19162 (pp_cfg_jump): Do not append info about BB when dumping a jump.
19164 2017-07-31 Martin Liska <mliska@suse.cz>
19167 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
19168 also with current_function_decl non-null equality.
19170 2017-07-31 Jakub Jelinek <jakub@redhat.com>
19173 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
19174 change type to the element type, instead add eltype variable and
19175 use it where we are interested in the element type.
19177 PR tree-optimization/81603
19178 * ipa-polymorphic-call.c
19179 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
19180 offset arithmetic in offset_int, bail out if the resulting bit offset
19181 doesn't fit into shwi.
19183 2017-07-31 Martin Liska <mliska@suse.cz>
19185 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
19186 (gimplify_save_expr): Fix comment.
19188 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
19191 * config/i386/i386.c (ix86_function_arg): Update arguments for
19193 (ix86_compute_frame_layout): Set the initial stack offset to
19194 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
19195 INCOMING_FRAME_SP_OFFSET.
19196 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
19197 stack before exception handler returns.
19198 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
19199 the 'ERROR_CODE' for exception handler.
19201 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
19203 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
19204 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
19205 (ASM_OUTPUT_REG_POP): Ditto.
19206 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
19207 instead of asm_fprintf to output pure string.
19209 2017-07-29 Jakub Jelinek <jakub@redhat.com>
19211 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
19212 to imported_module_or_decl hook.
19213 (debug_nothing_tree_tree_tree_bool): Remove.
19214 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
19215 * debug.c (do_nothing_debug_hooks): Use
19216 debug_nothing_tree_tree_tree_bool_bool instead of
19217 debug_nothing_tree_tree_tree_bool.
19218 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19219 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
19220 * sdbout.c (sdb_debug_hooks): Likewise.
19221 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
19222 (gen_namespace_die): Add DW_AT_export_symbols attribute if
19224 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
19225 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
19226 attribute, don't add anything.
19228 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19230 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
19231 (fold_build2_stat_loc): Likewise.
19232 (fold_build3_stat_loc): Likewise.
19233 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
19234 (fold_build1_loc): Remove macro.
19235 (fold_build2_loc): Likewise.
19236 (fold_build3_loc): Likewise.
19238 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19240 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
19241 (gimple_build_debug_bind_source_stat): Likewise.
19242 * gimple.h (gimple_build_debug_bind): Remove macro.
19243 (gimple_build_debug_bind_source): Likewise.
19245 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19247 * bitmap.c (bitmap_alloc): Adjust.
19248 (bitmap_gc_alloc): Likewise.
19249 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
19251 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19253 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
19254 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
19255 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
19256 (bitmap_gc_alloc_stat): Likewise.
19257 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
19259 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19261 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
19262 * rtl.h (shallow_copy_rtx): Remove macro.
19264 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19266 * emit-rtl.c (gen_raw_REG): Adjust.
19267 * gengenrtl.c (gendef): Likewise.
19268 * rtl.c (rtx_alloc_stat): Remove _stat from name.
19269 * rtl.h (rtx_alloc): Remove macro.
19271 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19273 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
19274 (build_tree_list_stat): Likewise.
19275 * tree.h (build_tree_list): Remove macro.
19276 (build_tree_list_vec): Likewise.
19278 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19280 * tree.c (make_vector_stat): Remove _stat from name.
19281 (build_vector_stat): Likewise.
19282 * tree.h (make_vector_stat): Remove macro.
19283 (build_vector_stat): Likewise.
19285 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19287 * tree.h (build_var_debug_value): Remove prototype.
19289 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19291 * tree.c (tree_cons_stat): Remove _stat from name.
19292 * tree.h (tree_cons): Remove macro.
19294 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19296 * tree.c (build_vl_exp_stat): Remove _stat from name.
19297 * tree.h (build_vl_exp): Remove macro.
19299 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19301 * tree.c (build_decl_stat): Remove _stat from name.
19302 * tree.h (build_decl): Remove macro.
19304 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19306 * gimple.c (gimple_build_with_ops_stat): Adjust.
19307 (gimple_alloc_stat): Remove _stat from name.
19308 * gimple.h (gimple_alloc): Remove macro.
19310 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19312 * tree.c (make_tree_vec_stat): Remove _stat from name.
19313 (grow_tree_vec_stat): Likewise.
19314 * tree.h (make_tree_vec_stat): Adjust prototype.
19315 (grow_tree_vec_stat): Likewise.
19316 (make_tree_vec): Remove macro.
19317 (grow_tree_vec): Likewise.
19319 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19321 * fold-const.c (fold_build1_stat_loc): Adjust.
19322 (fold_build2_stat_loc): Likewise.
19323 (fold_build3_stat_loc): Likewise.
19324 * tree.c (build0_stat): Remove _stat from name.
19325 (build1_stat): Likewise.
19326 (build2_stat): Likewise.
19327 (build3_stat): Likewise.
19328 (build4_stat): Likewise.
19329 (build5_stat): Likewise.
19330 * tree.h (build1_loc): Remove macro, and rename _stat function
19332 (build2_loc): Likewise.
19333 (build3_loc): Likewise.
19334 (build4_loc): Likewise.
19335 (build5_loc): Likewise.
19337 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19339 * tree.c (make_int_cst_stat): Remove _stat from name.
19340 * tree.h (make_int_cst_stat): Adjust prototype.
19341 (make_int_cst): Remove macro.
19343 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19345 * tree.c (make_tre_binfo_stat): Remove _stat from name.
19346 * tree.h (make_tree_binfo_stat): Adjust prototype.
19347 (make_tree_binfo): Remove.
19349 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19351 * tree.c (copy_node_stat): Rename to copy_node.
19352 (build_distinct_type_copy): Adjust.
19353 * tree.h (copy_node_stat): Adjust prototype.
19354 (copy_node): Remove macro.
19356 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19358 * tree.c (make_node_stat): rename to make_node.
19359 (build_tree_list_stat): Adjust.
19360 (build0_stat): Likewise.
19361 (build2_stat): Likewise.
19362 (build3_stat): Likewise.
19363 (build4_stat): Likewise.
19364 (build5_stat): Likewise.
19365 (build_decl_stat): Likewise.
19366 * tree.h (make_node_stat): Adjust prototype.
19367 (make_node): remove macro.
19369 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
19371 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
19372 (PPC_FEATURE2_SCV): Likewise.
19373 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
19375 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19377 * config/aarch64/aarch64.c
19378 (aarch64_internal_mov_immediate): Add new special pattern.
19379 * config/aarch64/aarch64.md (*movdi_aarch64):
19380 Add reg/32bit const mov case.
19382 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19383 Richard Sandiford <richard.sandiford@linaro.org>
19385 * config/aarch64/aarch64.md (mov<mode>): Generalize.
19386 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
19387 Add integer and movi cases.
19388 (movi-split-hf-df-sf split, fp16): New.
19389 (enabled): Added TARGET_FP_F16INST.
19390 * config/aarch64/iterators.md (GPF_HF): New.
19391 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
19393 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19395 * config/aarch64/aarch64.c
19396 (aarch64_simd_container_mode): Add prototype.
19397 (aarch64_expand_mov_immediate): Add HI support.
19398 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
19399 (aarch64_can_const_movi_rtx_p): New.
19400 (aarch64_preferred_reload_class):
19401 Remove restrictions of using FP registers for certain SIMD operations.
19402 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
19403 (aarch64_valid_floating_const): Add integer move validation.
19404 (aarch64_simd_imm_scalar_p): Remove.
19405 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
19406 (aarch64_legitimate_constant_p): Expand list of supported cases.
19407 * config/aarch64/aarch64-protos.h
19408 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
19409 (aarch64_reinterpret_float_as_int): New.
19410 (aarch64_simd_imm_scalar_p): Remove.
19411 * config/aarch64/constraints.md (Uvi): New.
19412 (Dd): Split into Ds and new Dd.
19413 * config/aarch64/aarch64.md (*movsi_aarch64):
19415 (*movdi_aarch64): Add SIMD mov case.
19417 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19419 * tree-predcom.c: (struct chain): Handle store-store chain in which
19420 stores for elimination only store loop invariant values.
19421 (execute_pred_commoning_chain): Ditto.
19422 (prepare_initializers_chain_store_elim): Ditto.
19423 (prepare_finalizers): Ditto.
19424 (is_inv_store_elimination_chain): New function.
19425 (initialize_root_vars_store_elim_1): New function.
19427 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19429 * tree-predcom.c: Revise general description of the pass.
19430 (enum chain_type): New enum type for store elimination.
19431 (struct chain): New field supporting store elimination.
19432 (struct component): Ditto.
19433 (dump_chain): Dump store-stores chain.
19434 (release_chain): Release resources.
19435 (split_data_refs_to_components): Compute and create component
19436 contains only stores for elimination.
19437 (get_chain_last_ref_at): New function.
19438 (make_invariant_chain): Initialization.
19439 (make_rooted_chain): Specify chain type in parameter and record it.
19440 (add_looparound_copies): Skip for store-stores chain.
19441 (determine_roots_comp): Compute type of chain and pass it to
19443 (initialize_root_vars_store_elim_2): New function.
19444 (finalize_eliminated_stores): New function.
19445 (remove_stmt): Handle store for elimination.
19446 (execute_pred_commoning_chain): Execute predictive commoning on
19447 store-store chains.
19448 (determine_unroll_factor): Skip unroll for store-stores chain.
19449 (prepare_initializers_chain_store_elim): New function.
19450 (prepare_initializers_chain): Hanlde store-store chain.
19451 (prepare_finalizers_chain, prepare_finalizers): New function.
19452 (tree_predictive_commoning_loop): Return integer value indicating
19453 if loop is unrolled or lcssa form is corrupted.
19454 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
19456 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19458 * tree-predcom.c (initialize_root): Delete.
19459 (execute_pred_commoning_chain): Initialize root vars and replace
19460 reference of non-combined chain directly, rather than call above
19463 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19465 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
19466 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
19468 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19470 * tree-predcom.c (struct chain): New field init_seq.
19471 (release_chain): Release init_seq.
19472 (prepare_initializers_chain): Record intialization stmts in above
19474 (insert_init_seqs): New function.
19475 (tree_predictive_commoning_loop): Call insert_init_seqs.
19477 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19479 * tree-predcom.c (determine_roots_comp): Skip trivial components.
19481 2017-07-28 Richard Biener <rguenther@suse.de>
19483 * match.pd: Remove superfluous :c.
19484 * genmatch.c (simplify::id): Add member.
19485 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
19487 (current_id): New global.
19488 (dt_node::parent): Move from ...
19489 (dt_operand::parent): ... here. Add for_id member.
19490 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
19491 (decision_tree::find_node): Relax order requirement when
19492 merging DT_TRUE nodes to ones inbetween the current simplify
19493 and the one we try to merge with. Add diagnostic whenever
19494 we need to enforce pattern order by not merging.
19495 (decision_tree::insert): Set current_id.
19496 (decision_tree::print_node): Dump parent node and for_id.
19497 (parser::last_id): Add member.
19498 (parser::push_simplify): Assign unique id.
19499 (parser::parser): Initialize last_id.
19501 2017-07-28 Martin Liska <mliska@suse.cz>
19504 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
19505 gimple_build_debug_bind.
19507 2017-07-28 Richard Biener <rguenther@suse.de>
19509 PR tree-optimization/81502
19510 * match.pd: Add pattern combining BIT_INSERT_EXPR with
19512 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
19514 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
19515 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
19516 for BIT_FIELD_REF args.
19517 * fold-const.c (make_bit_field_ref): Likewise.
19518 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
19520 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19523 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
19524 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
19525 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
19526 Or it into SANITIZER_UNDEFINED.
19527 * ubsan.c: Include gimple-fold.h and varasm.h.
19528 (ubsan_expand_ptr_ifn): New function.
19529 (instrument_pointer_overflow): New function.
19530 (maybe_instrument_pointer_overflow): New function.
19531 (instrument_object_size): Formatting fix.
19532 (pass_ubsan::execute): Call instrument_pointer_overflow
19533 and maybe_instrument_pointer_overflow.
19534 * internal-fn.c (expand_UBSAN_PTR): New function.
19535 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
19536 * sanitizer.def (__ubsan_handle_pointer_overflow,
19537 __ubsan_handle_pointer_overflow_abort): New builtins.
19538 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
19539 * internal-fn.def (UBSAN_PTR): New internal function.
19540 * opts.c (sanitizer_opts): Add pointer-overflow.
19541 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
19542 * fold-const.c (build_range_check): Compute pointer range check in
19543 integral type if pointer arithmetics would be needed. Formatting
19546 2017-07-28 Martin Liska <mliska@suse.cz>
19549 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
19550 parameters that are of a variable-length.
19552 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
19554 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
19556 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
19557 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
19558 (CRT_CALL_STATIC_FUNCTION): Likewise.
19559 (ASM_DEFAULT_SPEC): New define.
19560 (ASM_SPEC32): Likewise.
19561 (ASM_SPEC64): Likewise.
19562 (ASM_SPEC_COMMON): Likewise.
19563 (ASM_SPEC): Likewise.
19564 (INVALID_64BIT): Likewise.
19565 (LINK_OS_DEFAULT_SPEC): Likewise.
19566 (LINK_OS_SPEC32): Likewise.
19567 (LINK_OS_SPEC64): Likewise.
19568 (POWERPC_LINUX): Likewise.
19569 (PTRDIFF_TYPE): Likewise.
19570 (RESTORE_FP_PREFIX): Likewise.
19571 (RESTORE_FP_SUFFIX): Likewise.
19572 (SAVE_FP_PREFIX): Likewise.
19573 (SAVE_FP_SUFFIX): Likewise.
19574 (SIZE_TYPE): Likewise.
19575 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
19576 (TARGET_64BIT): Likewise.
19577 (TARGET_64BIT): Likewise.
19578 (TARGET_AIX): Likewise.
19579 (WCHAR_TYPE_SIZE): Likewise.
19580 (WCHAR_TYPE): Undefine.
19581 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
19582 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
19583 (CPP_OS_RTEMS_SPEC): Delete.
19584 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
19585 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
19587 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
19589 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19591 PR tree-optimization/81578
19592 * tree-parloops.c (build_new_reduction): Bail out if
19593 reduction_code isn't one of the standard OpenMP reductions.
19594 Move the details printing after that decision.
19596 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19598 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
19599 related to reload_in_progress.
19600 (splat_input_operand): Likewise.
19601 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
19603 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
19605 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
19606 (TARGET_INSTANTIATE_DECLS): Likewise.
19607 (legitimate_indexed_address_p): Delete reload_in_progress code.
19608 (rs6000_debug_legitimate_address_p): Likewise.
19609 (rs6000_eliminate_indexed_memrefs): Likewise.
19610 (rs6000_emit_le_vsx_store): Likewise.
19611 (rs6000_emit_move_si_sf_subreg): Likewise.
19612 (rs6000_emit_move): Likewise.
19613 (register_to_reg_type): Likewise.
19614 (rs6000_pre_atomic_barrier): Likewise.
19615 (rs6000_machopic_legitimize_pic_address): Likewise.
19616 (rs6000_allocate_stack_temp): Likewise.
19617 (rs6000_address_for_fpconvert): Likewise.
19618 (rs6000_address_for_altivec): Likewise.
19619 (rs6000_secondary_memory_needed_rtx): Delete function.
19620 (rs6000_check_sdmode): Likewise.
19621 (rs6000_alloc_sdmode_stack_slot): Likewise.
19622 (rs6000_instantiate_decls): Likewise.
19623 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
19624 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
19625 Delete reload_in_progress.
19626 (*vec_reload_and_plus_<mptrsize>): Likewise.
19627 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
19628 (vsx_div_v2di): Likewise.
19629 (vsx_udiv_v2di): Likewise.
19631 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19633 * config/rs6000/rs6000.opt (mlra): Replace with stub.
19634 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
19635 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
19636 (rs6000_debug_reg_global): Delete print of LRA status.
19637 (rs6000_option_override_internal): Delete dead LRA related code.
19638 (rs6000_lra_p): Delete function.
19639 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
19641 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
19643 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
19644 * config/riscv/rtems.h: New file.
19646 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19647 Sudakshina Das <sudi.das@arm.com>
19649 * config/aarch64/aarch64.md
19650 (define_split for and<mode>3nr_compare): Move
19651 non aarch64_logical_operand to a register.
19652 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
19653 register immediate operand to a register.
19654 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
19656 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19658 PR middle-end/81564
19659 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
19661 2017-07-27 Richard Biener <rguenther@suse.de>
19663 PR tree-optimization/81573
19664 PR tree-optimization/81494
19665 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
19666 multi defuse cycle case.
19668 2017-07-27 Richard Biener <rguenther@suse.de>
19670 PR tree-optimization/81571
19671 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
19674 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
19676 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
19677 earlier and only if MASK_FPU is set. Adjust formatting.
19679 2017-07-27 Martin Liska <mliska@suse.cz>
19681 * opt-functions.awk: Add validation of value of Init.
19682 * optc-gen.awk: Pass new argument.
19684 2017-07-27 Martin Liska <mliska@suse.cz>
19686 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
19687 Fix wrong condition.
19689 2017-07-27 Martin Liska <mliska@suse.cz>
19691 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
19692 BBs and edges seen by autoFDO.
19694 2017-07-27 Richard Biener <rguenther@suse.de>
19696 PR tree-optimization/81502
19697 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
19698 with incompatible but same sized type.
19699 (execute_update_addresses_taken): Likewise.
19701 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
19703 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
19704 flag_tree_loop_vectorize rather than flag_tree_vectorize.
19706 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19709 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
19710 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
19711 Change s_operand to memory_operand.
19713 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
19715 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
19716 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
19717 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
19718 Emit instructions rather than returning an expression. Handle TFmode
19719 and KFmode by casting to TImode.
19720 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
19721 (rs6000_emit_le_vsx_store): Likewise.
19722 * config/rs6000/vsx.md (VSX_TI): New iterator.
19723 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
19724 (*vsx_le_undo_permute_<mode>): Likewise.
19725 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
19726 emit the split sequence.
19727 (*vsx_le_perm_store_<mode>): Likewise.
19729 2017-07-27 Jakub Jelinek <jakub@redhat.com>
19731 PR tree-optimization/81555
19732 PR tree-optimization/81556
19733 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
19734 if true, force CHANGED for the recursive invocation.
19735 (reassociate_bb): Remember original length of ops array, pass
19736 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
19738 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
19739 attributes for noipa attribute. For naked attribute use
19740 lookup_attribute first before lookup_attribute_spec.
19741 * final.c (rest_of_handle_final): Disable IPA RA for functions with
19743 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
19744 for functions with noipa attribute.
19745 (cgraph_externally_visible_p): Return true for functions with noipa
19747 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
19748 for functions with noipa attribute.
19749 * doc/extend.texi: Document noipa function attribute.
19750 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
19751 also for functions with noipa attribute.
19752 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
19754 2017-07-26 Andrew Pinski <apinski@cavium.com>
19756 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
19757 vec_unalign_load_cost and vec_unalign_store_cost.
19759 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
19761 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
19762 -mvsx-small-integer option.
19763 (ISA_3_0_MASKS_IEEE): Likewise.
19764 (OTHER_VSX_VECTOR_MASKS): Likewise.
19765 (POWERPC_MASKS): Likewise.
19766 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
19767 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
19768 code, only testing for DImode being allowed in non-VSX floating
19770 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
19771 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
19773 (rs6000_option_override_internal): Delete -mvsx-small-integer.
19774 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
19775 TARGET_P8_VECTOR test.
19776 (rs6000_secondary_reload_simple_move): Likewise.
19777 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
19778 since TARGET_P9_VECTOR was already tested.
19779 (rs6000_opt_masks): Remove -mvsx-small-integer.
19780 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
19781 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19783 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
19784 test for TARGET_VEXTRACTUB was used, and that uses
19786 (p9 extract splitter): Likewise.
19787 (vsx_extract_<mode>_di_p9): Likewise.
19788 (vsx_extract_<mode>_store_p9): Likewise.
19789 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
19790 for TARGET_P9_VECTOR was used. Delete code that is now dead with
19791 the elimination of TARGET_VSX_SMALL_INTEGER.
19792 (vsx_extract_<mode>_p8): Likewise.
19793 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
19794 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
19795 (vsx_set_<mode>_p9): Likewise.
19796 (vsx_set_v4sf_p9): Likewise.
19797 (vsx_set_v4sf_p9_zero): Likewise.
19798 (vsx_insert_extract_v4sf_p9): Likewise.
19799 (vsx_insert_extract_v4sf_p9_2): Likewise.
19800 * config/rs6000/rs6000.md (sign extend splitter): Change
19801 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
19802 (floatsi<mode>2_lfiwax_mem): Likewise.
19803 (floatunssi<mode>2_lfiwzx_mem): Likewise.
19804 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
19805 since a test for TARGET_P9_VECTOR was used.
19806 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19807 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
19808 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19809 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
19810 TARGET_P8_VECTOR test.
19811 (fix_trunc<mode>si2_stfiwx): Likewise.
19812 (fix_trunc<mode>si2_internal): Likewise.
19813 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
19814 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19816 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19817 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
19818 TARGET_P8_VECTOR test.
19819 (fixuns_trunc<mode>si2_stfiwx): Likewise.
19820 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
19821 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19823 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19824 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
19825 since a test for TARGET_P9_VECTOR was used.
19826 (splitter for loading small constants): Likewise.
19828 2017-07-26 Andrew Pinski <apinski@cavium.com>
19830 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
19833 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
19836 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
19837 (fp_valid_at): Likewise.
19839 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
19841 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
19842 (qdf24xx_addrcost_table): Likewise.
19843 (cortexa57_tunings): Update to use generic_branch_cost.
19844 (cortexa72_tunings): Likewise.
19845 (cortexa73_tunings): Likewise.
19846 (qdf24xx_tunings): Likewise.
19848 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
19850 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
19851 (thunderx2t99_branch_cost): Likewise.
19852 (cortexa35_tunings): Update to use generic_branch_cost.
19853 (cortexa53_tunings): Likewise.
19854 (cortexa57_tunings): Likewise.
19855 (cortexa72_tunings): Likewise.
19856 (cortexa73_tunings): Likewise.
19857 (thunderx2t99_tunings): Likewise.
19859 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19861 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
19862 (sparc_option_override): Honour MASK_FSMULD.
19863 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
19864 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
19865 * config/sparc/sparc.opt (mfsmuld): New option.
19866 * doc/invoke.texi (mfsmuld): Document option.
19868 2017-07-26 Marek Polacek <polacek@redhat.com>
19870 PR middle-end/70992
19871 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
19873 2017-07-26 Richard Biener <rguenther@suse.de>
19875 * gimple-match-head.c (do_valueize): Return OP if valueize
19877 (get_def): New helper to get at the def stmt of a SSA name
19878 if valueize allows.
19879 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
19880 do_valueize to get at the def stmt.
19881 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
19883 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
19885 PR middle-end/46932
19886 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
19888 2017-07-26 Martin Liska <mliska@suse.cz>
19891 * function.c (expand_function_start): Make expansion of
19892 nonlocal_goto_save_area after parm_birth_insn.
19894 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19896 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
19897 from all CPU target flags enable members.
19899 2017-07-26 Richard Biener <rguenther@suse.de>
19901 * genmatch.c (dt_simplify::gen): Make iterator vars const.
19902 (decision_tree::gen): Make 'type' const.
19903 (write_predicate): Likewise.
19905 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
19907 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
19908 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
19909 (rs6000_option_override_internal): Likewise.
19910 (rs6000_expand_vector_set): Likewise.
19911 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
19912 (TARGET_UPPER_REGS_SF): Likewise.
19913 (TARGET_UPPER_REGS_DI): Likewise.
19914 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
19915 (TARGET_DIRECT_MOVE_64BIT): Likewise.
19916 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
19917 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19918 (Splitters for DI constants in Altivec registers): Likewise.
19919 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
19920 (vsx_set_v4sf_p9): Likewise.
19921 (vsx_set_v4sf_p9_zero): Likewise.
19922 (vsx_insert_extract_v4sf_p9): Likewise.
19923 (vsx_insert_extract_v4sf_p9_2): Likewise.
19925 2017-07-25 Carl Love <cel@us.ibm.com>
19927 * doc/extend.texi: Update the built-in documentation file for the
19928 existing built-in functions
19929 vector signed char vec_cnttz (vector signed char);
19930 vector unsigned char vec_cnttz (vector unsigned char);
19931 vector signed short vec_cnttz (vector signed short);
19932 vector unsigned short vec_cnttz (vector unsigned short);
19933 vector signed int vec_cnttz (vector signed int);
19934 vector unsigned int vec_cnttz (vector unsigned int);
19935 vector signed long long vec_cnttz (vector signed long long);
19936 vector unsigned long long vec_cnttz (vector unsigned long long);
19938 2017-07-25 Andrew Pinski <apinski@cavium.com>
19940 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
19941 accesses where the use is for the first operand of a BIT_INSERT.
19943 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19946 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
19947 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
19949 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19951 * config/i386/gstabs.h: Delete.
19952 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
19954 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
19956 * config/i386/i386.c (ix86_decompose_address): Do not check for
19957 register RTX when looking at index_reg or base_reg.
19958 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
19960 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
19962 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
19963 to update EH info here.
19965 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19967 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
19969 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19971 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
19973 2017-07-25 Torsten Duwe <duwe@suse.de>
19975 * common.opt: Introduce -fpatchable-function-entry
19976 command line option, and its variables function_entry_patch_area_size
19977 and function_entry_patch_area_start.
19978 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
19979 including a two-value parser.
19980 * target.def (print_patchable_function_entry): New target hook.
19981 * targhooks.h (default_print_patchable_function_entry): New function.
19982 * targhooks.c (default_print_patchable_function_entry): Likewise.
19983 * toplev.c (process_options): Switch off IPA-RA if
19984 patchable function entries are being generated.
19985 * varasm.c (assemble_start_function): Look at the
19986 patchable-function-entry command line switch and current
19987 function attributes and maybe generate NOP instructions by
19988 calling the print_patchable_function_entry hook.
19989 * doc/extend.texi: Document patchable_function_entry attribute.
19990 * doc/invoke.texi: Document -fpatchable_function_entry
19991 command line option.
19992 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
19994 * doc/tm.texi: Re-generate.
19996 2017-07-25 Jakub Jelinek <jakub@redhat.com>
19999 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
20000 TARGET_AVX512DQ rather than TARGET_AVX512BW.
20002 2017-07-25 Tamar Christina <tamar.christina@arm.com>
20004 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
20006 2017-07-25 Richard Biener <rguenther@suse.de>
20008 PR tree-optimization/81455
20009 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
20010 not walk in cycles when looking for guards.
20012 2017-07-25 Richard Biener <rguenther@suse.de>
20014 PR tree-optimization/81529
20015 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
20016 when optimizing backedge uses.
20018 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
20020 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
20022 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
20023 to dl_section_ref. On AIX, append an expression to subtract
20024 the size of the section length to dl_section_ref.
20026 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
20028 * configure.ac: If any of the config.* scripts fail, exit 1.
20029 * configure: Regenerate.
20031 2017-07-25 Richard Biener <rguenther@suse.de>
20033 PR middle-end/81546
20034 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
20035 of immediate uses, be more verbose on errors.
20037 2017-07-25 Richard Biener <rguenther@suse.de>
20039 PR tree-optimization/81510
20040 * tree-vect-loop.c (vect_is_simple_reduction): When the
20041 reduction stmt is not inside the loop bail out.
20043 2017-07-25 Richard Biener <rguenther@suse.de>
20045 PR tree-optimization/81303
20046 * tree-vect-loop-manip.c (vect_loop_versioning): Build
20047 profitability check against LOOP_VINFO_NITERSM1.
20049 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20051 * domwalk.c (cmp_bb_postorder): Simplify.
20052 (sort_bbs_postorder): New function. Use it...
20053 (dom_walker::walk): ...here to optimize common cases.
20055 2017-07-25 Martin Liska <mliska@suse.cz>
20058 * ipa-visibility.c (function_and_variable_visibility): Make the
20059 redirection just on target that supports aliasing.
20060 Fix GNU coding style.
20062 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
20065 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
20067 * config/arm/rtems.h: Likewise.
20068 * config/bfin/rtems.h: Likewise.
20069 * config/i386/rtemself.h: Likewise.
20070 * config/lm32/rtems.h: Likewise.
20071 * config/m32c/rtems.h: Likewise.
20072 * config/m68k/rtemself.h: Likewise.
20073 * config/microblaze/rtems.h: Likewise.
20074 * config/mips/rtems.h: Likewise.
20075 * config/moxie/rtems.h: Likewise.
20076 * config/nios2/rtems.h: Likewise.
20077 * config/powerpcspe/rtems.h: Likewise.
20078 * config/rs6000/rtems.h: Likewise.
20079 * config/rtems.h: Likewise.
20080 * config/sh/rtems.h: Likewise.
20081 * config/sh/rtemself.h: Likewise.
20082 * config/sparc/rtemself.h: Likewise.
20084 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20087 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
20088 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
20089 * tree-ssa-structalias.c (alias_get_name): Same.
20091 2017-07-25 Bin Cheng <bin.cheng@arm.com>
20094 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
20095 instructions if no du chain is found.
20097 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20099 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
20101 2017-07-25 Richard Biener <rguenther@suse.de>
20103 PR middle-end/81505
20104 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
20107 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
20109 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
20110 upper-regs options.
20111 (ISA_2_7_MASKS_SERVER): Likewise.
20112 (ISA_3_0_MASKS_IEEE): Likewise.
20113 (OTHER_P8_VECTOR_MASKS): Likewise.
20114 (OTHER_VSX_VECTOR_MASKS): Likewise.
20115 (POWERPC_MASKS): Likewise.
20116 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
20117 duplicate list of options.
20118 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
20119 explicit -mupper-regs options.
20120 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
20121 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
20122 alias for -mupper-regs-df.
20123 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20124 (rs6000_init_hard_regno_mode_ok): Likewise.
20125 (rs6000_option_override_internal): Likewise.
20126 (rs6000_opt_masks): Likewise.
20127 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
20128 options in terms of whether -mvsx or -mpower8-vector was used.
20129 (TARGET_UPPER_REGS_DI): Likewise.
20130 (TARGET_UPPER_REGS_SF): Likewise.
20131 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
20132 -mupper-regs-* options.
20134 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
20136 * passes.c (emergency_dump_function): Print some empty lines and a
20137 header before the RTL dump.
20139 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
20141 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
20143 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
20146 * config/aarch64/aarch64.c (aarch64_classify_symbol):
20147 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
20149 2017-07-24 Carl Love <cel@us.ibm.com>
20151 * config/rs6000/rs6000-c.c: Add support for built-in functions
20152 vector float vec_extract_fp32_from_shorth (vector unsigned short);
20153 vector float vec_extract_fp32_from_shortl (vector unsigned short);
20154 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
20155 vec_extract_fp_from_shortl): Add defines for the two builtins.
20156 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
20157 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
20159 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
20160 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
20161 * doc/extend.texi: Update the built-in documentation file for the
20162 new built-in function.
20164 2017-07-24 Jakub Jelinek <jakub@redhat.com>
20167 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
20169 * doc/generic.texi: Likewise.
20170 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
20171 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
20173 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
20175 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
20176 (aarch64_mls_elt_merge<mode>): Likewise.
20178 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
20180 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
20181 having __cxa_atexit.
20183 2017-07-23 Michael Collison <michael.collison@arm.com>
20185 * config/arm/arm.c (arm_option_override): Deprecate
20186 use of -mstructure-size-boundary.
20187 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
20188 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
20190 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20193 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
20194 Reduce cost estimate for direct moves.
20196 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
20199 * config/i386/i386.c (ix86_option_override_internal): Disable
20200 BMI, BMI2 and TBM instructions for -m16.
20202 2017-07-21 Carl Love <cel@us.ibm.com>
20204 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20205 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20206 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20207 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20208 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20209 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20210 VMULOSW): New enum "unspec" values.
20211 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20212 altivec_vmulosw): New patterns.
20213 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20214 VMULOSW): Add definitions.
20216 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
20218 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
20219 (qdf24xx): Likewise.
20220 * config/aarch64/aarch64-options-extensions.def (rdma); New.
20221 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
20222 (AARCH64_FL_V8_1): Renumber.
20223 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
20224 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
20225 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
20226 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
20227 rdma to feature modifiers list.
20229 2017-07-21 Yury Gribov <tetra2005@gmail.com>
20231 PR middle-end/56727
20232 * ipa-visibility (function_and_variable_visibility): Convert
20233 recursive PLT call to direct call if appropriate.
20235 2017-07-21 Andrew Pinski <apinski@cavium.com>
20237 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
20238 operand 1 to see if the types precision matches.
20239 * fold-const.c (operand_equal_p): Likewise.
20241 2017-07-21 Richard Biener <rguenther@suse.de>
20243 PR tree-optimization/81303
20244 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
20245 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
20246 (vect_peeling_hash_get_lowest_cost): Adjust.
20247 (vect_enhance_data_refs_alignment): Likewise. Use
20248 vect_get_peeling_costs_all_drs to compute the penalty for no
20249 peeling to match up costs.
20251 2017-07-21 Richard Biener <rguenther@suse.de>
20253 PR tree-optimization/81500
20254 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
20255 we didn't identify a reduction path.
20257 2017-07-21 Tom de Vries <tom@codesourcery.com>
20258 Cesar Philippidis <cesar@codesourcery.com>
20260 PR gcov-profile/81442
20261 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
20264 2017-07-21 Tom de Vries <tom@codesourcery.com>
20267 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
20269 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
20270 nvptx_override_options_after_change.
20272 2017-07-21 Ulrich Drepper <drepper@redhat.com>
20274 * dwarf2out.c (output_file_names): Avoid double testing for
20275 dwarf_version >= 5.
20277 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
20279 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
20281 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20283 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
20285 (try_crossjump_to_edge): Do not punt on partitioned functions.
20287 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20289 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
20290 Put all BBs reachable only via paths crossing cold region to cold
20292 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
20294 2016-07-21 Richard Biener <rguenther@suse.de>
20296 PR tree-optimization/81303
20297 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
20298 into account prologue and epilogue iterations when raising
20299 min_profitable_iters to sth at least covering one vector iteration.
20301 2017-07-21 Tamar Christina <tamar.christina@arm.com>
20303 * config/arm/arm.c (arm_test_cpu_arch_dat):
20306 2017-07-20 Nathan Sidwell <nathan@acm.org>
20308 Remove TYPE_METHODS.
20309 * tree.h (TYPE_METHODS): Delete.
20310 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
20311 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
20312 (dbxout_type_methods): Scan TYPE_FIELDS.
20313 (dbxout_type): Don't check TYPE_METHODS here.
20314 * function.c (use_register_for_decl): Always ignore register for
20315 class types when not optimizing.
20316 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
20317 * tree.c (free_lang_data_in_type): Stitch out member functions and
20318 templates from TYPE_FIELDS.
20319 (build_distinct_type_copy, verify_type_variant,
20320 verify_type): Member fns are on TYPE_FIELDS.
20321 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
20322 * tree-pretty-print.c (dump_generic_node): Likewise.
20324 2017-07-20 Jakub Jelinek <jakub@redhat.com>
20327 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
20328 V2TImode and V4TImode.
20329 (ix86_expand_vector_extract): Likewise.
20330 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
20331 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
20332 (ssescalarmode): Handle V4TImode and V2TImode.
20333 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
20334 (*vec_extractv2ti, *vec_extractv4ti): New insns.
20335 (VEXTRACTI128_MODE): New mode iterator.
20336 (splitter for *vec_extractv?ti first element): New.
20337 (VEC_INIT_MODE): New mode iterator.
20338 (vec_init<mode>): Consolidate 3 expanders into one using
20339 VEC_INIT_MODE mode iterator.
20341 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20343 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
20344 non_spilled_static_chain_regno_p.
20346 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20348 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
20350 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
20352 * bb-reorder.c (connect_traces): Allow copying of blocks within
20355 2017-07-20 Richard Biener <rguenther@suse.de>
20357 * gimple.h (gimple_phi_result): Add gphi * overload.
20358 (gimple_phi_result_ptr): Likewise.
20359 (gimple_phi_arg): Likewise. Adjust index assert to only
20360 allow actual argument accesses rather than all slots available
20362 (gimple_phi_arg_def): Add gphi * overload.
20363 * tree-phinodes.c (make_phi_node): Initialize only actual
20365 (resize_phi_node): Clear memory not covered by old node,
20366 do not initialize excess argument slots.
20367 (reserve_phi_args_for_new_edge): Initialize new argument slot
20370 2017-07-20 Bin Cheng <bin.cheng@arm.com>
20372 PR tree-optimization/81388
20374 2016-07-21 Bin Cheng <bin.cheng@arm.com>
20376 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
20377 by removing computation of may_be_zero.
20379 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20380 Tom de Vries <tom@codesourcery.com>
20382 PR middle-end/81030
20383 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
20384 when gimple level profile disagrees with what RTL expander did.
20386 2017-07-20 Richard Biener <rguenther@suse.de>
20388 PR tree-optimization/61171
20389 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
20390 (vect_analyze_stmt): Add slp instance parameter.
20391 (vectorizable_reduction): Likewise.
20392 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
20393 (vect_is_simple_reduction): Deal with chains not detected
20394 as SLP reduction chain, specifically not properly associated
20395 chains containing a mix of plus/minus.
20396 (get_reduction_op): Remove.
20397 (get_initial_defs_for_reduction): Simplify, pass in whether
20398 this is a reduction chain, pass in the SLP node for the PHIs.
20399 (vect_create_epilog_for_reduction): Get the SLP instance as
20401 (vectorizable_reduction): Get the SLP instance as arg.
20402 During analysis remember the SLP node with the PHIs in the
20403 instance. Simplify getting at the vectorized reduction PHIs.
20404 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
20405 through SLP instance.
20406 (vect_slp_analyze_operations): Likewise.
20407 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
20408 (vect_transform_stmt): Likewise.
20410 2017-07-20 Tom de Vries <tom@codesourcery.com>
20412 PR tree-optimization/81489
20413 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
20414 read of phi arg location to before loop that modifies phi.
20416 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20418 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
20421 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20423 PR middle-end/81331
20424 * except.c (execute): Fix ordering issue.
20426 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20428 PR rtl-optimization/81423
20429 * combine.c (make_compound_operation_int): Don't try to optimize
20430 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
20432 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20434 PR rtl-optimization/81423
20435 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
20436 with a constant that is -1 in the truncated to mode.
20438 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20440 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
20441 (determine_unlikely_bbs): ... here.
20442 * predict.h (propagate_unlikely_bbs_forward): Declare.
20443 * cfgexpand.c (pass_expand::execute): Use it.
20444 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
20446 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
20447 propagate_unlikely_bbs_forward.
20449 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20451 PR middle-end/81331
20452 * except.c (maybe_add_nop_after_section_switch): New function.
20455 2017-07-19 Tom de Vries <tom@codesourcery.com>
20457 * gimple.h (gimple_phi_set_arg): Make assert more strict.
20459 2017-07-19 Tom de Vries <tom@codesourcery.com>
20461 * gimple.h (gimple_phi_arg): Make assert more strict.
20463 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
20465 * config.gcc (powerpc*-*-*): Add mmintrin.h.
20466 * config/rs6000/mmintrin.h: New file.
20467 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
20469 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20471 PR tree-optimization/81346
20472 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
20474 2017-07-19 Tom de Vries <tom@codesourcery.com>
20476 * config/nvptx/nvptx.md (VECIM): Add V2DI.
20478 2017-07-19 Tom de Vries <tom@codesourcery.com>
20480 * config/nvptx/nvptx-modes.def: Add V2DImode.
20481 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
20482 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
20483 (nvptx_output_mov_insn): Handle lack of mov.b128.
20484 (nvptx_print_operand): Handle 'H' and 'L' codes.
20485 (nvptx_vector_mode_supported): Allow V2DImode.
20486 (nvptx_preferred_simd_mode): New function.
20487 (nvptx_data_alignment): New function.
20488 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
20489 nvptx_preferred_simd_mode.
20490 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
20492 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
20494 2017-07-19 Tom de Vries <tom@codesourcery.com>
20496 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
20497 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
20498 (nvptx_vector_mode_supported): New function. Allow V2SImode.
20499 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
20500 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
20501 (mov<VECIM>_insn): New define_insn.
20502 (define_expand "mov<VECIM>): New define_expand.
20504 2017-07-19 Tom de Vries <tom@codesourcery.com>
20506 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
20508 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20510 PR tree-optimization/81346
20511 * fold-const.h (fold_div_compare, range_check_type): Declare.
20512 * fold-const.c (range_check_type): New function.
20513 (build_range_check): Use range_check_type.
20514 (fold_div_compare): No longer static, rewritten into
20515 a match.pd helper function.
20516 (fold_comparison): Don't call fold_div_compare here.
20517 * match.pd (X / C1 op C2): New optimization using fold_div_compare
20518 as helper function.
20520 2017-07-19 Nathan Sidwell <nathan@acm.org>
20522 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
20523 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
20524 * tree.c (find_decls_types_r, verify_type): Use
20525 TYPE_{MIN,MAX}_VALUE_RAW.
20526 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
20527 (hash_tree): Likewise.
20528 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20530 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20533 2017-07-18 Tom de Vries <tom@codesourcery.com>
20535 PR middle-end/81464
20536 * omp-expand.c (expand_omp_for_static_chunk): Handle
20537 equal-argument loop exit phi.
20539 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
20542 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
20543 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
20544 operand 2 predicate.
20545 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
20546 operand 2 predicate.
20547 (ror,rol -> rorx splitters): Use const_int_operand as
20548 operand 2 predicate.
20550 2017-06-18 Richard Biener <rguenther@suse.de>
20552 PR tree-optimization/81410
20553 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
20554 the gap in the ! slp_perm SLP case after each group.
20556 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20558 PR middle-end/81463
20559 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
20562 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20564 PR middle-end/81462
20565 * predict.c (set_even_probabilities): Cleanup; do not affect
20566 probabilities that are already known.
20567 (combine_predictions_for_bb): Call even when count is set.
20569 2017-07-18 Nathan Sidwell <nathan@acm.org>
20571 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
20574 2017-07-18 Bin Cheng <bin.cheng@arm.com>
20577 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
20578 optimization for loop niter analysis.
20580 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
20583 * config/avr/avr.c (avr_optimize_casesi): Don't use
20584 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
20586 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
20588 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
20589 body_cost_vec from _vect_peel_extended_info.
20590 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
20591 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
20594 2017-07-18 Bin Cheng <bin.cheng@arm.com>
20596 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
20598 2017-07-18 Richard Biener <rguenther@suse.de>
20600 PR tree-optimization/80620
20601 PR tree-optimization/81403
20602 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
20603 info when re-using a VN table entry.
20605 2017-07-18 Richard Biener <rguenther@suse.de>
20607 PR tree-optimization/81418
20608 * tree-vect-loop.c (vectorizable_reduction): Properly compute
20609 vectype_in. Verify that with lane-reducing reduction operations
20610 we have a single def-use cycle.
20612 2017-07-17 Carl Love <cel@us.ibm.com>
20614 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
20616 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20617 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20618 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20619 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20620 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20621 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20622 VMULOSW): New enum "unspec" values.
20623 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20624 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20625 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20626 altivec_vmulosw): New patterns.
20627 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20628 VMULOSW): Add definitions.
20630 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
20632 * config/alpha/alpha.c: Include predict.h.
20634 2017-07-17 Yury Gribov <tetra2005@gmail.com>
20636 * tree-vrp.c (compare_assert_loc): Fix comparison function
20637 to return predictable results.
20639 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20641 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
20643 (subdi3): Likewise.
20644 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
20645 * doc/invoke.texi (mexpand-adddi): Update text.
20647 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20649 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
20650 that also clobbers the CC register. The old expand code is moved
20652 (*arc_clzsi2): ... here.
20653 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
20654 the CC register. The old expand code is moved to ...
20655 (arc_ctzsi2): ... here.
20657 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20659 * config/arc/arc.opt (mindexed-loads): Use initial value
20660 TARGET_INDEXED_LOADS_DEFAULT.
20661 (mauto-modify-reg): Use initial value
20662 TARGET_AUTO_MODIFY_REG_DEFAULT.
20663 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
20664 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
20665 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
20666 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
20668 2017-07-17 Martin Liska <mliska@suse.cz>
20671 * opts.c (finish_options): Do not allow -fgnu-tm
20672 w/ -fsanitize={kernel-,}address. Say sorry.
20674 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20677 * tree-loop-distribution.c (classify_partition): Only assert on
20678 numer of iterations.
20679 (merge_dep_scc_partitions): Delete prameter. Update function call.
20680 (distribute_loop): Remove code handling loop with unknown niters.
20681 (pass_loop_distribution::execute): Skip loop with unknown niters.
20683 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20686 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
20687 function sort_partitions_by_post_order.
20689 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20691 PR tree-optimization/81374
20692 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
20693 the max index of basic blocks, rather than number of basic blocks.
20695 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20697 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
20699 (arc_legitimate_pic_operand_p): Likewise.
20700 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
20702 (arc_needs_pcl_p): Likewise.
20703 (arc_legitimate_pc_offset_p): Likewise.
20704 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
20705 function is also used in constrains.md.
20706 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
20707 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
20708 PLUS. Only return true/false in known cases, otherwise assert.
20709 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
20710 is already called in arc_legitimate_constant_p.
20711 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
20713 (LEGITIMATE_PIC_OPERAND_P): Use
20714 arc_raw_symbolic_reference_mentioned_p function.
20715 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
20720 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20721 Andrew Burgess <andrew.burgess@embecosm.com>
20723 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
20724 (arc_return_address_register): New function.
20725 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
20726 (arc_handle_fndecl_attribute): Add naked attribute.
20727 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
20728 (TARGET_WARN_FUNC_RETURN): Likewise.
20729 (arc_allocate_stack_slots_for_args): New function.
20730 (arc_warn_func_return): Likewise.
20731 (machine_function): Change type fn_type.
20732 (arc_compute_function_type): Consider new naked function type,
20733 change function return type.
20734 (arc_must_save_register): Adapt to handle new
20735 arc_compute_function_type's return type.
20736 (arc_expand_prologue): Likewise.
20737 (arc_expand_epilogue): Likewise.
20738 (arc_return_address_regs): Delete.
20739 (arc_return_address_register): New function.
20740 (arc_epilogue_uses): Use above function.
20741 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
20742 (arc_function_type): Change encoding, add naked type.
20743 (ARC_INTERRUPT_P): Change to handle the new encoding.
20744 (ARC_FAST_INTERRUPT_P): Likewise.
20745 (ARC_NORMAL_P): Define.
20746 (ARC_NAKED_P): Likewise.
20747 (arc_compute_function_type): Delete prototype.
20748 * config/arc/arc.md (in_ret_delay_slot): Use
20749 arc_return_address_register function.
20750 (simple_return): Likewise.
20751 (p_return_i): Likewise.
20753 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20755 PR tree-optimization/81428
20756 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
20757 can't be built for those types.
20759 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
20761 Remove stuff dead since r239246.
20763 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
20764 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
20765 (avr_inform_devices): Remove dead stuff.
20767 2017-07-17 Tamar Christina <tamar.christina@arm.com>
20769 * config/arm/arm_neon.h: Fix softp typo.
20771 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20773 PR tree-optimization/81365
20774 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
20775 aggregate moves onto bb predecessor edges, make sure there are no
20776 loads that could alias the lhs in between the start of bb and the
20779 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
20782 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
20783 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
20784 [LSHIFTRT, outer_code = TRUNCATE]: Same.
20786 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20788 PR tree-optimization/81396
20789 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
20790 (init_symbolic_number): Initialize it to 1.
20791 (perform_symbolic_merge): Add n_ops from both operands into the new
20793 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
20794 without base_addr as useless if they need more than one operation.
20795 (bswap_replace): Handle !bswap case for NULL base_addr.
20797 2017-07-17 Tom de Vries <tom@codesourcery.com>
20800 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
20803 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
20805 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
20806 conditional builtin define __FIX_LEON3FT_B2BST.
20808 2017-07-17 Daniel Cederman <cederman@gaisler.com>
20810 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
20811 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
20814 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20816 PR rtl-optimization/81424
20817 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
20818 to remove potential trapping from operands if -fnon-call-exceptions.
20820 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20822 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
20823 profile_proability for scalling.
20824 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
20826 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20828 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
20830 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20832 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
20833 fixpoint arithmetics.
20835 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20837 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
20838 fixpoint arithmetics.
20840 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20842 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
20843 fixpoint arithmetics.
20845 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20847 * profile-count.h (profile_probability::from_reg_br_prob_note,
20848 profile_probability::to_reg_br_prob_note): New functions.
20849 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
20850 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
20851 * predict.c (probability_reliable_p): Update.
20852 (edge_probability_reliable_p): Update.
20853 (br_prob_note_reliable_p): Update.
20854 (invert_br_probabilities): Update.
20855 (add_reg_br_prob_note): New function.
20856 (combine_predictions_for_insn): Update.
20857 * asan.c (asan_clear_shadow): Update.
20858 * cfgbuild.c (compute_outgoing_frequencies): Update.
20859 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
20860 (update_br_prob_note): Update.
20861 (rtl_verify_edges): Update.
20862 (purge_dead_edges): Update.
20863 (fixup_reorder_chain): Update.
20864 * emit-rtl.c (try_split): Update.
20865 * ifcvt.c (cond_exec_process_insns): Update.
20866 (cond_exec_process_if_block): Update.
20867 (dead_or_predicable): Update.
20868 * internal-fn.c (expand_addsub_overflow): Update.
20869 (expand_neg_overflow): Update.
20870 (expand_mul_overflow): Update.
20871 * loop-doloop.c (doloop_modify): Update.
20872 * loop-unroll.c (compare_and_jump_seq): Update.
20873 * optabs.c (emit_cmp_and_jump_insn_1): Update.
20874 * predict.h: Update.
20875 * reorg.c (mostly_true_jump): Update.
20877 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
20878 * config/alpha/alpha.c (emit_unlikely_jump): Update.
20879 * config/arc/arc.c: (emit_unlikely_jump): Update.
20880 * config/arm/arm.c: (emit_unlikely_jump): Update.
20881 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
20882 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
20883 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
20884 (ix86_print_operand): Update.
20885 (ix86_split_fp_branch): Update.
20886 (predict_jump): Update.
20887 * config/ia64/ia64.c (ia64_print_operand): Update.
20888 * config/mmix/mmix.c (mmix_print_operand): Update.
20889 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
20890 (rs6000_expand_split_stack_prologue): Update.
20891 * config/rs6000/rs6000.c: Update.
20892 * config/s390/s390.c (s390_expand_vec_strlen): Update.
20893 (s390_expand_vec_movstr): Update.
20894 (s390_expand_cs_tdsi): Update.
20895 (s390_expand_split_stack_prologue): Update.
20896 * config/sh/sh.c (sh_print_operand): Update.
20897 (expand_cbranchsi4): Update.
20898 (expand_cbranchdi4): Update.
20899 * config/sparc/sparc.c (output_v9branch): Update.
20900 * config/spu/spu.c (get_branch_target): Update.
20901 (ea_load_store_inline): Update.
20902 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
20903 * config/tilepro/tilepro.c: Update.
20905 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20907 * gimplify.c (mostly_copy_tree_r): Revert latest change.
20908 (gimplify_save_expr): Likewise.
20910 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20912 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
20914 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20916 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
20918 * timevar.def (TV_IPA_FNSUMMARY): Define.
20920 2017-07-16 Daniel Cederman <cederman@gaisler.com>
20922 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
20923 to back store errata sensitive sequence from being generated.
20924 (sqrtdf2_fix): Likewise.
20926 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20928 * tree-ssa-threadupdate.c (compute_path_counts,
20929 update_joiner_offpath_counts): Use profile_probability.
20931 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
20934 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20936 * config/arm/arm-c.c (arm_cpu_builtins): Define
20937 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20939 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
20941 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20942 array entries to represent __ieee128 versions of the
20943 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
20944 scalar_extract_sig, and scalar_insert_exp built-in functions.
20945 (altivec_resolve_overloaded_builtin): Add special case handling
20946 for the __builtin_scalar_insert_exp function, as represented by
20947 the P9V_BUILTIN_VEC_VSIEDP constant.
20948 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
20949 exponent support for __ieee128 argument.
20950 (VSESQP): Add scalar extract signature support for __ieee128
20952 (VSTDCNQP): Add scalar test negative support for __ieee128
20954 (VSIEQP): Add scalar insert exponent support for __int128 argument
20955 with __ieee128 result.
20956 (VSIEQPF): Add scalar insert exponent support for __ieee128
20957 argument with __ieee128 result.
20958 (VSTDCQP): Add scalar test data class support for __ieee128
20960 (VSTDCNQP): Add overload support for scalar test negative with
20961 __ieee128 argument.
20962 (VSTDCQP): Add overload support for scalar test data class
20963 __ieee128 argument.
20964 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
20965 UNSPEC_VSX_SXSIGDP.
20966 (UNSPEC_VSX_SIEXPQP): New constant.
20967 (xsxexpqp): New insn for VSX scalar extract exponent quad
20969 (xsxsigqp): New insn for VSX scalar extract significand quad
20971 (xsiexpqpf): New insn for VSX scalar insert exponent quad
20972 precision with floating point argument.
20973 (xststdcqp): New expand for VSX scalar test data class quad
20975 (xststdcnegqp): New expand for VSX scalar test negative quad
20977 (xststdcqp): New insn to match expansions for VSX scalar test data
20978 class quad precision and VSX scalar test negative quad precision.
20979 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
20980 special case operand checking to enforce that second operand of
20981 VSX scalar test data class with quad precision argument is a 7-bit
20983 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20984 prototypes and descriptions of __ieee128 versions of
20985 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
20986 scalar_test_data_class, and scalar_test_neg built-in functions.
20988 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20990 PR tree-optimization/81162
20991 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
20992 replace a negate with an add.
20994 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
20996 * doc/invoke.texi (arm/-mcpu): Document +crypto.
20998 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21000 * config/arm/arm-c.c (arm_cpu_builtins): Define
21001 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
21003 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21005 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
21006 (armv8-r): Set ARM Cortex-R52 as default CPU.
21007 * config/arm/arm-tables.opt: Regenerate.
21008 * config/arm/arm-tune.md: Regenerate.
21009 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
21011 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
21012 extension for -mcpu=cortex-r52.
21014 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21016 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
21017 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
21018 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
21019 (fp-armv8): Define it as FP_ARMv8 only.
21020 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
21021 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
21023 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
21024 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
21025 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
21026 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
21027 than TARGET_FPU_ARMV8.
21028 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
21029 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
21030 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
21032 * config/arm/neon.md (neon_vrint): Likewise.
21033 (neon_vcvt): Likewise.
21034 (neon_<fmaxmin_op><mode>): Likewise.
21035 (<fmaxmin><mode>3): Likewise.
21036 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
21037 * config/arm/predicates.md (arm_cond_move_operator): Check against
21038 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
21040 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
21042 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
21043 to top of function.
21045 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21047 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
21048 loop in comment with memset.
21050 2017-07-14 Martin Liska <mliska@suse.cz>
21052 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
21053 * dwarf2out.c (is_java): Remove the function.
21054 (output_pubname): Remove usage of the function.
21055 (lower_bound_default): Remove usage of DW_LANG_Java.
21056 (gen_compile_unit_die): Likewise.
21057 * gcc.c: Remove compiler defaults for .java and .zip files.
21058 * gimple-expr.c (remove_suffix): Change as there's no longer
21059 extension than 4-letter one.
21060 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
21061 (gimplify_save_expr): Likewise.
21062 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
21063 as it's possible even for other languages than Java.
21064 * langhooks.h (struct lang_hooks): Remove Java from a comment.
21065 * lto-opts.c (lto_write_options): Remove reference to Java.
21066 * opts.c (strip_off_ending): Update file extension handling.
21067 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
21068 * tree-eh.c (lower_resx): Likewise.
21069 * tree.c (free_lang_data_in_type): Remove dead code.
21070 (find_decls_types_r): Likewise.
21071 (build_common_builtin_nodes): Remove Java from a comment.
21072 (verify_type): Remove dead code.
21073 * varasm.c (assemble_external): Remove Java from a comment.
21075 2017-07-14 Martin Liska <mliska@suse.cz>
21077 * opts.c (finish_options): Add quotes.
21078 (common_handle_option): Likewise.
21080 2017-07-14 Martin Liska <mliska@suse.cz>
21082 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
21083 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
21084 Remove N_SO_PASCAL.
21085 * dwarf2out.c (lower_bound_default): Do not handle
21087 (gen_compile_unit_die): Likewise.
21088 * gcc.c: Remove default extension binding for GNU Pascal.
21089 * stmt.c: Remove Pascal language from a comment.
21090 * xcoffout.c: Likewise.
21092 2017-07-13 David Malcolm <dmalcolm@redhat.com>
21095 * diagnostic-show-locus.c (fixit_cmp): New function.
21096 (layout::layout): Sort m_fixit_hints.
21097 (column_range::column_range): Assert that the values are valid.
21098 (struct char_span): New struct.
21099 (correction::overwrite): New method.
21100 (struct source_line): New struct.
21101 (line_corrections::add_hint): Add assertions. Reimplement memcpy
21102 calls in terms of classes source_line and char_span, and
21103 correction::overwrite.
21104 (selftest::test_overlapped_fixit_printing_2): New function.
21105 (selftest::diagnostic_show_locus_c_tests): Call it.
21107 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
21109 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
21110 early if there is no lhs.
21112 2017-07-13 Martin Liska <mliska@suse.cz>
21114 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
21115 (gen_reference_type_die): Likewise.
21116 * stor-layout.c: Remove Pascal-related comment.
21118 2017-07-13 Martin Liska <mliska@suse.cz>
21120 * opts.c (finish_options): Add quotes to error messages.
21121 (parse_sanitizer_options): Likewise.
21123 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
21125 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
21127 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
21129 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
21131 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
21133 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
21135 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
21137 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
21140 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
21141 provides the hardware capability bits, define the macro
21142 __BUILTIN_CPU_SUPPORTS__.
21143 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
21144 if GLIBC does not provide the hardware capability bits. Add a
21145 gcc_unreachable call if the built-in cpu function is neither
21146 __builtin_cpu_is nor __builtin_cpu_supports.
21147 (rs6000_get_function_versions_dispatcher): Change the warning
21148 that an old GLIBC is used which does not export the capability
21149 bits to be an error.
21150 * doc/extend.texi (target_clones attribute): Document the
21151 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
21152 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
21153 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
21154 the macros defined by GCC if the newer GLIBC is available.
21156 2017-07-12 Jeff Law <law@redhat.com>
21158 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
21159 remaining includes slightly.
21160 * config/riscv/riscv-builtins.c: Include profile-count.h.
21162 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21165 * config/avr/avr.c (avr_set_current_function): In diagnostic
21166 messages: Quote keywords and (parts of) identifiers.
21167 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
21170 2017-07-12 Carl Love <cel@us.ibm.com>
21172 * config/rs6000/rs6000-c.c: Add support for built-in functions
21173 vector bool char vec_revb (vector bool char);
21174 vector bool short vec_revb (vector short char);
21175 vector bool int vec_revb (vector bool int);
21176 vector bool long long vec_revb (vector bool long long);
21177 * doc/extend.texi: Update the built-in documentation file for the
21178 new built-in functions.
21180 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21182 * config/s390/s390.md: Remove movcc splitter.
21184 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21186 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
21187 load/store on condition.
21189 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21192 * config/avr/avr.c (avr_encode_section_info)
21193 [progmem && !TREE_READONLY]: Error if progmem object needs
21196 2017-07-11 Michael Collison <michael.collison@arm.com>
21198 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
21201 2017-07-11 Carl Love <cel@us.ibm.com>
21203 * config/rs6000/rs6000-c.c: Add support for builtins
21204 vector unsigned int vec_parity_lsbb (vector signed int);
21205 vector unsigned int vec_parity_lsbb (vector unsigned int);
21206 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
21207 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
21208 vector unsigned long long vec_parity_lsbb (vector signed long long);
21209 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
21210 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
21211 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
21212 * doc/extend.texi: Update the built-in documentation file for the
21213 new built-in functions.
21215 2017-07-11 David Malcolm <dmalcolm@redhat.com>
21217 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
21218 (layout::m_primary_loc): New field.
21219 (layout::layout): Initialize new field. Move location filtering
21220 logic from here to...
21221 (layout::maybe_add_location_range): ...this new method. Add
21222 support for filtering to just the lines already specified by other
21224 (layout::will_show_line_p): New method.
21225 (gcc_rich_location::add_location_if_nearby): New method.
21226 (selftest::test_add_location_if_nearby): New test function.
21227 (selftest::diagnostic_show_locus_c_tests): Call it.
21228 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
21231 2017-07-11 Tom de Vries <tom@codesourcery.com>
21233 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
21234 (bb_first_real_insn): New function.
21235 (nvptx_single): Add extra initialization of broadcasted condition
21238 2017-07-11 Nathan Sidwell <nathan@acm.org>
21240 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
21242 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
21244 * doc/extend.texi (AVR Function Attributes): Remove weblink to
21245 Binutils doc as TEXI will mess them up.
21246 * doc/invoke.texi (AVR Options): Same here.
21248 2017-07-11 Daniel Cederman <cederman@gaisler.com>
21250 * config/sparc/sparc.opt (mfix-ut700): New option.
21251 (mfix-gr712rc): Likewise.
21252 (sparc_fix_b2bst): New variable.
21253 * doc/invoke.texi (SPARC options): Document them.
21254 (ARM options): Fix warnings.
21255 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
21256 instructions to prevent sequences that can trigger the store-store
21257 errata for certain LEON3FT processors.
21258 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
21259 (sparc_option_override): Set sparc_fix_b2bst appropriately.
21260 * config/sparc/sparc.md (fix_b2bst): New attribute.
21261 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
21263 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
21266 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
21268 (*rsqrtsf2_sse): Ditto.
21270 (div<mode>3): Macroize insn from divdf3 and divsf3
21271 using MODEF mode iterator.
21273 2017-07-10 Martin Sebor <msebor@redhat.com>
21275 PR tree-optimization/80397
21276 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
21277 instead of testing for equality to INTEGER_TYPE.
21279 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
21281 * config.gcc: Remove uclibc from arc target spec.
21283 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
21285 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
21287 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21290 * lto-wrapper.c (remove_option): New function.
21291 (merge_and_complain): Merge PIC/PIE options more realistically.
21293 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21295 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
21299 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
21300 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
21301 * config.in: Regenerate.
21302 * configure: Regenerate.
21303 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
21304 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
21305 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
21306 (TARGET_GASISR_PROLOGUES): ...target mask.
21307 * common/config/avr/avr-common.c
21308 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
21309 Set -mgas-isr-prologues.
21310 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
21311 INSERT_PASS_BEFORE for it.
21312 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
21313 * config/avr/avr.c (avr_option_override)
21314 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
21315 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
21316 (avr_attribute_table) <no_gccisr>: Add new function attribute.
21317 (avr_set_current_function) <is_no_gccisr>: Init machine field.
21318 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
21320 (make_avr_pass_pre_proep): New function.
21321 (emit_push_sfr) <treg>: Add argument to function and use it
21322 instead of TMP_REG.
21323 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
21324 and set machine->gasisr.yes.
21325 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
21326 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
21327 __gcc_isr.n_pushed to .L__stack_usage.
21328 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
21329 (avr_asm_final_postscan_insn): ...this new static function.
21330 * config/avr/avr.h (machine_function)
21331 <is_no_gccisr, use_L__stack_usage>: New fields.
21332 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
21333 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
21334 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
21335 (gasisr, *gasisr): New expander and insn.
21336 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21337 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
21338 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
21340 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
21342 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
21345 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21347 Move jump-tables out of .text again.
21350 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
21351 (ASM_OUTPUT_ADDR_VEC): New function.
21352 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
21353 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
21354 INSN_ADDRESSes as asm comment.
21355 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
21356 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
21357 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
21358 * config/avr/avr.md (*tablejump): Adjust comment.
21359 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
21360 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
21362 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
21363 (avr_output_addr_vec): New proto.
21364 (avr_log_t) <insn_addresses>: New field.
21366 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
21369 * config/i386/i386.c (ix86_function_arg_advance): Set
21370 outgoing_args_on_stack to true if there are outgoing arguments
21372 (ix86_function_arg): Likewise.
21373 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
21374 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
21375 * config/i386/i386.h (machine_function): Add
21376 outgoing_args_on_stack.
21378 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
21380 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
21381 supporting pthreds.
21382 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
21384 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
21386 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
21387 (REAL_H): Remove $(MACHMODE_H).
21388 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
21390 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
21391 $(MACHMODE_H) and double-int.h.
21392 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
21394 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
21397 2017-07-07 Andrew Pinski <apinski@cavium.com>
21399 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
21400 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
21402 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21404 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
21405 Add warning if GCC was not configured to link against a GLIBC that
21406 exports the hardware capability bits.
21407 (make_resolver_func): Make resolver function private and not a
21408 COMDAT function. Create the name with clone_function_name instead
21409 of make_unique_name.
21412 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
21413 correct operand in doing the split.
21415 2017-07-07 Carl Love <cel@us.ibm.com>
21417 * config/rs6000/rs6000-c: Add support for built-in function
21418 vector unsigned short vec_pack_to_short_fp32 (vector float,
21420 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
21421 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
21422 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
21423 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
21424 (convert_4f32_8i16): Add define_expand.
21425 * doc/extend.texi: Update the built-in documentation file for the
21426 new built-in function.
21428 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21430 * config/sparc/m8.md: New file.
21431 * config/sparc/sparc.md: Include m8.md.
21433 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21435 * config/sparc/sparc.opt: New option -mvis4b.
21436 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
21437 (sparc_option_override): Handle VIS4B.
21438 (enum sparc_builtins): Define
21439 SPARC_BUILTIN_DICTUNPACK{8,16,32},
21440 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
21441 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
21442 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
21443 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
21444 (check_constant_argument): New function.
21445 (sparc_vis_init_builtins): Define builtins
21446 __builtin_vis_dictunpack{8,16,32},
21447 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
21448 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
21449 __builtin_vis_fpcmpde{8,16,32}shl and
21450 __builtin_vis_fpcmpur{8,16,32}shl.
21451 (sparc_expand_builtin): Check that the constant operands to
21452 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
21453 constant and in range.
21454 * config/sparc/sparc-c.c (sparc_target_macros): Handle
21456 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
21457 (SPARC_IMM5_P): Likewise.
21458 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
21459 (enabled): Handle vis4b.
21460 (UNSPEC_DICTUNPACK): New unspec.
21461 (UNSPEC_FPCMPSHL): Likewise.
21462 (UNSPEC_FPUCMPSHL): Likewise.
21463 (UNSPEC_FPCMPDESHL): Likewise.
21464 (UNSPEC_FPCMPURSHL): Likewise.
21465 (cpu_feature): New CPU feature `vis4b'.
21466 (dictunpack{8,16,32}): New insns.
21467 (FPCSMODE): New mode iterator.
21468 (fpcscond): New code iterator.
21469 (fpcsucond): Likewise.
21470 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
21471 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
21472 (fpcmpde{8,16,32}{si,di}shl): Likewise.
21473 (fpcmpur{8,16,32}{si,di}shl): Likewise.
21474 * config/sparc/constraints.md: Define constraints `q' for unsigned
21475 2-bit integer constants and `t' for unsigned 5-bit integer
21477 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
21479 (imm5_operand_dictunpack16): Likewise.
21480 (imm5_operand_dictunpack32): Likewise.
21481 (imm2_operand): Likewise.
21482 * doc/invoke.texi (SPARC Options): Document -mvis4b.
21483 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
21484 ditunpack* and fpcmp*shl builtins.
21486 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21488 * config.gcc: Handle m8 in --with-{cpu,tune} options.
21489 * config.in: Add HAVE_AS_SPARC6 define.
21490 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
21492 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
21494 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
21495 (CPP_CPU_SPEC): Handle m8.
21496 (ASM_CPU_SPEC): Likewise.
21497 * config/sparc/sparc-opts.h (enum processor_type): Add
21499 * config/sparc/sparc.c (m8_costs): New struct.
21500 (sparc_option_override): Handle TARGET_CPU_m8.
21501 (sparc32_initialize_trampoline): Likewise.
21502 (sparc64_initialize_trampoline): Likewise.
21503 (sparc_issue_rate): Likewise.
21504 (sparc_register_move_cost): Likewise.
21505 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
21506 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
21507 (ASM_CPU64_DEFAULT_SPEC): Likewise.
21508 (CPP_CPU_SPEC): Handle M8.
21509 (ASM_CPU_SPEC): Likewise.
21510 (AS_M8_FLAG): Define.
21511 * config/sparc/sparc.md: Add m8 to the cpu attribute.
21512 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
21513 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
21515 * configure: Regenerate.
21516 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
21519 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21521 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
21523 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
21524 ("*movdi_insn_sp32"): Do not set v3pipe.
21525 ("*movsi_insn"): Likewise.
21526 ("*movdi_insn_sp64"): Likewise.
21527 ("*movsf_insn"): Likewise.
21528 ("*movdf_insn_sp32"): Likewise.
21529 ("*movdf_insn_sp64"): Likewise.
21530 ("*zero_extendsidi2_insn_sp64"): Likewise.
21531 ("*sign_extendsidi2_insn"): Likewise.
21532 ("*mov<VM32:mode>_insn"): Likewise.
21533 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21534 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21535 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21536 ("<vlop:code><VL:mode>3"): Likewise.
21537 ("*not_<vlop:code><VL:mode>3"): Likewise.
21538 ("*nand<VL:mode>_vis"): Likewise.
21539 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
21540 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
21541 ("one_cmpl<VL:mode>2"): Likewise.
21542 ("faligndata<VM64:mode>_vis"): Likewise.
21543 ("alignaddrsi_vis"): Likewise.
21544 ("alignaddrdi_vis"): Likweise.
21545 ("alignaddrlsi_vis"): Likewise.
21546 ("alignaddrldi_vis"): Likewise.
21547 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21548 ("bmaskdi_vis"): Likewise.
21549 ("bmasksi_vis"): Likewise.
21550 ("bshuffle<VM64:mode>_vis"): Likewise.
21551 ("cmask8<P:mode>_vis"): Likewise.
21552 ("cmask16<P:mode>_vis"): Likewise.
21553 ("cmask32<P:mode>_vis"): Likewise.
21554 ("pdistn<P:mode>_vis"): Likewise.
21555 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21557 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21559 * config/sparc/sparc.md ("subtype"): New insn attribute.
21560 ("*wrgsr_sp64"): Set insn subtype.
21561 ("*rdgsr_sp64"): Likewise.
21562 ("alignaddrsi_vis"): Likewise.
21563 ("alignaddrdi_vis"): Likewise.
21564 ("alignaddrlsi_vis"): Likewise.
21565 ("alignaddrldi_vis"): Likewise.
21566 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21567 ("fexpand_vis"): Likewise.
21568 ("fpmerge_vis"): Likewise.
21569 ("faligndata<VM64:mode>_vis"): Likewise.
21570 ("bshuffle<VM64:mode>_vis"): Likewise.
21571 ("cmask8<P:mode>_vis"): Likewise.
21572 ("cmask16<P:mode>_vis"): Likewise.
21573 ("cmask32<P:mode>_vis"): Likewise.
21574 ("fchksm16_vis"): Likewise.
21575 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
21576 ("fmean16_vis"): Likewise.
21577 ("fp<plusminus_insn>64_vis"): Likewise.
21578 ("<plusminus_insn>v8qi3"): Likewise.
21579 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21580 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
21581 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
21582 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
21583 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
21584 ("*movqi_insn"): Likewise.
21585 ("*movhi_insn"): Likewise.
21586 ("*movsi_insn"): Likewise.
21587 ("movsi_pic_gotdata_op"): Likewise.
21588 ("*movdi_insn_sp32"): Likewise.
21589 ("*movdi_insn_sp64"): Likewise.
21590 ("movdi_pic_gotdata_op"): Likewise.
21591 ("*movsf_insn"): Likewise.
21592 ("*movdf_insn_sp32"): Likewise.
21593 ("*movdf_insn_sp64"): Likewise.
21594 ("*zero_extendhisi2_insn"): Likewise.
21595 ("*zero_extendqihi2_insn"): Likewise.
21596 ("*zero_extendqisi2_insn"): Likewise.
21597 ("*zero_extendqidi2_insn"): Likewise.
21598 ("*zero_extendhidi2_insn"): Likewise.
21599 ("*zero_extendsidi2_insn_sp64"): Likewise.
21600 ("ldfsr"): Likewise.
21601 ("prefetch_64"): Likewise.
21602 ("prefetch_32"): Likewise.
21603 ("tie_ld32"): Likewise.
21604 ("tie_ld64"): Likewise.
21605 ("*tldo_ldub_sp32"): Likewise.
21606 ("*tldo_ldub1_sp32"): Likewise.
21607 ("*tldo_ldub2_sp32"): Likewise.
21608 ("*tldo_ldub_sp64"): Likewise.
21609 ("*tldo_ldub1_sp64"): Likewise.
21610 ("*tldo_ldub2_sp64"): Likewise.
21611 ("*tldo_ldub3_sp64"): Likewise.
21612 ("*tldo_lduh_sp32"): Likewise.
21613 ("*tldo_lduh1_sp32"): Likewise.
21614 ("*tldo_lduh_sp64"): Likewise.
21615 ("*tldo_lduh1_sp64"): Likewise.
21616 ("*tldo_lduh2_sp64"): Likewise.
21617 ("*tldo_lduw_sp32"): Likewise.
21618 ("*tldo_lduw_sp64"): Likewise.
21619 ("*tldo_lduw1_sp64"): Likewise.
21620 ("*tldo_ldx_sp64"): Likewise.
21621 ("*mov<VM32:mode>_insn"): Likewise.
21622 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21623 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21625 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21627 * config/sparc/sparc.md ("type"): New insn type viscmp.
21628 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
21630 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
21631 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
21632 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21633 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
21635 ("n7_vis_logical_11cycle"): Likewise.
21636 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
21637 * config/sparc/niagara2.md ("niag3_vis": Likewise.
21638 * config/sparc/niagara.md ("niag_vis"): Likewise.
21639 * config/sparc/ultra3.md ("us3_fga"): Likewise.
21640 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
21642 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21644 * config/sparc/sparc.md: New instruction type `bmask'.
21645 (bmaskdi_vis): Use the `bmask' type.
21646 (bmasksi_vis): Likewise.
21647 * config/sparc/ultra3.md (us3_array): Likewise.
21648 * config/sparc/niagara7.md (n7_array): Likewise.
21649 * config/sparc/niagara4.md (n4_array): Likewise.
21650 * config/sparc/niagara2.md (niag2_vis): Likewise.
21651 (niag3_vis): Likewise.
21652 * config/sparc/niagara.md (niag_vis): Likewise.
21654 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21656 * ipa-comdats.c: Remove optimize check from gate.
21657 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
21658 for functions not optimized.
21659 (ipa_fn_summary_read): Skip optimize check.
21660 (ipa_fn_summary_write): Likewise.
21661 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
21663 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
21665 (can_inline_edge_p): Check flag_pcc_struct_return for match.
21666 (check_callers): Give up on caller which is not optimized.
21667 (inline_small_functions): Likewise.
21668 (ipa_inline): Do not give up when not optimizing.
21669 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
21670 away unoptimizes cdtors.
21671 (whole_program_function_and_variable_visibility): Do
21672 ipa_discover_readonly_nonaddressable_vars in LTO mode.
21673 * ipa.c (process_references): Do not check optimize.
21674 (symbol_table::remove_unreachable_nodes): Update optimize check.
21675 (set_writeonly_bit): Update optimize check.
21676 (pass_ipa_cdtor_merge::gate): Do not check optimize.
21677 (pass_ipa_single_use::gate): Remove.
21679 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21681 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
21682 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
21683 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
21684 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
21685 permute_load, permute_store, adjust_extract, adjust_splat,
21686 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
21687 replace_swap_with_copy, dump_swap_insn_table,
21688 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
21689 recombine_lvx_pattern, recombine_stvx_pattern,
21690 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
21691 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
21692 to file rs6000-p8swap.c.
21693 * config/rs6000/rs6000-p8swap.c: New file.
21694 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
21695 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
21696 and rs6000*-*-* targets.
21698 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21700 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
21702 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21704 * lto-wrapper.c (merge_and_complain): Do not merge
21705 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
21706 fsigned_zeros, ftrapping_math, fwrapv.
21707 (append_compiler_options): Do not track these options.
21708 (append_linker_options): Likewie
21710 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21712 * cgraphunit.c (cgraph_node::finalize_function): When
21713 !flag_toplevel_reorde set no_reorder flag.
21714 (varpool_node::finalize_decl): Likewise.
21715 (symbol_table::compile): Drop no toplevel reorder path.
21717 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21719 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
21720 edges; zero probability is not better than uninitialized.
21722 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
21724 * asan.h (asan_sanitize_allocas_p): Declare.
21725 * asan.c (asan_sanitize_allocas_p): New function.
21726 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
21727 (handle_builtin_alloca): Likewise.
21728 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
21729 if !asan_sanitize_allocas_p.
21730 * params.def (asan-instrument-allocas): Add new option.
21731 * params.h (ASAN_PROTECT_ALLOCAS): Define.
21732 * opts.c (common_handle_option): Disable allocas sanitization for
21735 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
21737 * asan.c: Include gimple-fold.h.
21738 (get_last_alloca_addr): New function.
21739 (handle_builtin_stackrestore): Likewise.
21740 (handle_builtin_alloca): Likewise.
21741 (asan_emit_allocas_unpoison): Likewise.
21742 (get_mem_refs_of_builtin_call): Add new parameter, remove const
21743 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
21744 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
21745 (instrument_builtin_call): Pass gimple iterator to
21746 get_mem_refs_of_builtin_call.
21747 (last_alloca_addr): New global.
21748 * asan.h (asan_emit_allocas_unpoison): Declare.
21749 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
21750 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
21751 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
21752 if function calls alloca.
21753 * gimple-fold.c (replace_call_with_value): Remove static keyword.
21754 * gimple-fold.h (replace_call_with_value): Declare.
21755 * internal-fn.c: Include asan.h.
21756 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
21757 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
21759 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21761 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
21762 (C_SELFTEST_FLAGS): New.
21763 (CPP_SELFTEST_FLAGS): New.
21764 (SELFTEST_DEPS): New, from deps of s-selftest.
21765 (C_SELFTEST_DEPS): New, from deps of s-selftest.
21766 (CPP_SELFTEST_DEPS): New.
21767 (selftest): Add dependency on s-selftest-c++.
21768 (s-selftest): Rename to...
21769 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
21770 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
21771 than SELFTEST_FLAGS.
21772 (selftest-gdb): Rename to...
21773 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
21775 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
21776 (selftest-valgrind): Rename to...
21777 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
21779 (selftest-valgrind): Reintroduce as an alias for
21780 selftest-c-valgrind.
21781 (s-selftest-c++): New.
21782 (selftest-c++-gdb): New.
21783 (selftest-c++-valgrind): New.
21785 2017-07-06 Olivier Hainque <hainque@adacore.com>
21787 * gcc.c (process_command): When deciding if undefined variables
21788 should be ignored when processing specs, accept "gcc -v" as well.
21790 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21792 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
21793 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
21795 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21797 * config/arm/arm-cpus.in (armv8-r): Add new entry.
21798 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
21799 * config/arm/arm-tables.opt: Regenerate.
21800 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
21802 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
21804 2017-07-06 Carl Love <cel@us.ibm.com>
21806 * ChangeLog: Clean up from mid air collision
21808 2017-07-06 Carl Love <cel@us.ibm.com>
21810 * config/rs6000/rs6000-c.c: Add support for built-in functions
21811 vector signed int vec_subc (vector signed int, vector signed int);
21812 vector signed __int128 vec_subc (vector signed __int128,
21813 vector signed __int128);
21814 vector unsigned __int128 vec_subc (vector unsigned __int128,
21815 vector unsigned __int128);
21816 vector signed int vec_sube (vector signed int, vector signed int,
21817 vector signed int);
21818 vector unsigned int vec_sube (vector unsigned int,
21819 vector unsigned int,
21820 vector unsigned int);
21821 vector signed __int128 vec_sube (vector signed __int128,
21822 vector signed __int128,
21823 vector signed__int128);
21824 vector unsigned __int128 vec_sube (vector unsigned __int128,
21825 vector unsigned __int128,
21826 vector unsigned __int128);
21827 vector signed int vec_subec (vector signed int, vector signed int,
21828 vector signed int);
21829 vector unsigned int vec_subec (vector unsigned int,
21830 vector unsigned int,
21831 vector unsigned int);
21832 vector signed __int128 vec_subec (vector signed __int128,
21833 vector signed __int128,
21834 vector signed__int128);
21835 vector unsigned __int128 vec_subec (vector unsigned __int128,
21836 vector unsigned __int128,
21837 vector unsigned __int128);
21838 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
21839 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
21840 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
21841 BU_ALTIVEC_OVERLOAD_X definitions.
21842 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
21843 * doc/extend.texi: Update the built-in documentation file for the new
21844 built-in functions.
21846 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21849 * diagnostic-show-locus.c (layout::layout): Use start and finish
21850 spelling location for the start and finish of each range.
21851 * genmatch.c (linemap_client_expand_location_to_spelling_point):
21852 Add unused aspect param.
21853 * input.c (expand_location_1): Add "aspect" param, and use it
21854 to access the correct part of the location.
21855 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
21857 (expand_location_to_spelling_point): Likewise.
21858 (linemap_client_expand_location_to_spelling_point): Add "aspect"
21859 param, and pass it to expand_location_1.
21861 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
21863 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
21864 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
21865 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
21866 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
21867 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
21868 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
21869 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
21870 _mm_maskz_getmant_ss): New intrinsics.
21871 (__builtin_ia32_getexpss128_mask): Changed to ...
21872 __builtin_ia32_getexpss128_round ... this.
21873 (__builtin_ia32_getexpsd128_mask): Changed to ...
21874 __builtin_ia32_getexpsd128_round ... this.
21875 * config/i386/i386-builtin-types.def
21876 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
21877 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
21878 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
21879 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
21880 __builtin_ia32_getmantss_mask_round): New builtins.
21881 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
21882 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
21883 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
21884 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
21885 * config/i386/sse.md
21886 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
21887 avx512f_sgetexp<mode><mask_scalar_name>
21888 <round_saeonly_scalar_name> ... this.
21889 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
21890 %0, %1, %2<round_saeonly_op3>}): Changed to ...
21891 vgetexp<ssescalarmodesuffix>
21892 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21893 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
21894 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
21895 avx512f_vgetmant<mode><mask_scalar_name>
21896 <round_saeonly_scalar_name> ... this.
21897 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
21898 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
21899 vgetmant<ssescalarmodesuffix>
21900 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
21901 %0<mask_scalar_operand4>, %1, %2
21902 <round_saeonly_scalar_mask_op4>, %3} ... this.
21903 * config/i386/subst.md (mask_scalar_operand4,
21904 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
21905 round_saeonly_scalar_nimm_predicate): New subst attributes.
21907 2017-07-06 Julia Koval <julia.koval@intel.com>
21909 * config/i386/i386.c (ix86_erase_embedded_rounding):
21910 Remove code for old rounding pattern.
21912 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
21914 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
21916 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
21918 * doc/sourcebuild.texi (Test Directives, Variants of
21919 dg-require-support): Add documentation for dg-require-stack-check.
21921 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
21923 * config/i386/subst.md (mask_scalar, round_scalar,
21924 round_saeonly_scalar): New meta-templates.
21925 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
21926 round_scalar_mask_operand3, round_scalar_mask_op3,
21927 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
21928 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
21929 round_saeonly_scalar_constraint,
21930 round_saeonly_scalar_prefix): New subst attribute.
21931 * config/i386/sse.md
21932 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
21933 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
21934 <round_scalar_name> ... this.
21935 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
21936 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
21937 <round_scalar_name> ... this.
21938 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
21939 <sse>_vm<code><mode>3<mask_scalar_name>
21940 <round_saeonly_scalar_name> ... this.
21941 (v<plusminus_mnemonic><ssescalarmodesuffix>
21942 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21943 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21944 v<plusminus_mnemonic><ssescalarmodesuffix>
21945 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21946 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21947 (v<multdiv_mnemonic><ssescalarmodesuffix>
21948 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21949 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21950 v<multdiv_mnemonic><ssescalarmodesuffix>
21951 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21952 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21953 (v<maxmin_float><ssescalarmodesuffix>
21954 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
21955 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
21956 v<maxmin_float><ssescalarmodesuffix>
21957 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21958 %0<mask_scalar_operand3>, %1, %<iptr>2
21959 <round_saeonly_scalar_mask_op3>} ... this.
21961 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
21963 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
21964 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
21966 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
21967 Alan Hayward <alan.hayward@arm.com>
21968 David Sherwood <david.sherwood@arm.com>
21970 * combine.c (simplify_if_then_else): Remove "enum" before
21972 * compare-elim.c (can_eliminate_compare): Likewise.
21973 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
21975 (aarch64_lookup_simd_builtin_type): Likewise.
21976 (aarch64_simd_builtin_type): Likewise.
21977 (aarch64_init_simd_builtin_types): Likewise.
21978 (aarch64_simd_expand_args): Likewise.
21979 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
21981 (aarch64_reverse_mask): Likewise.
21982 (aarch64_simd_emit_reg_reg_move): Likewise.
21983 (aarch64_gen_adjusted_ldpstp): Likewise.
21984 (aarch64_ccmp_mode_to_code): Likewise.
21985 (aarch64_operands_ok_for_ldpstp): Likewise.
21986 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21987 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
21989 (aarch64_min_divisions_for_recip_mul): Likewise.
21990 (aarch64_reassociation_width): Likewise.
21991 (aarch64_get_condition_code_1): Likewise.
21992 (aarch64_simd_emit_reg_reg_move): Likewise.
21993 (aarch64_simd_attr_length_rglist): Likewise.
21994 (aarch64_reverse_mask): Likewise.
21995 (aarch64_operands_ok_for_ldpstp): Likewise.
21996 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21997 (aarch64_gen_adjusted_ldpstp): Likewise.
21998 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
22000 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
22001 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
22002 (arm_lookup_simd_builtin_type): Likewise.
22003 (arm_simd_builtin_type): Likewise.
22004 (arm_init_simd_builtin_types): Likewise.
22005 (arm_expand_builtin_args): Likewise.
22006 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
22007 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
22008 (ft32_setup_incoming_varargs): Likewise.
22009 (ft32_function_arg): Likewise.
22010 (ft32_function_arg_advance): Likewise.
22011 (ft32_pass_by_reference): Likewise.
22012 (ft32_arg_partial_bytes): Likewise.
22013 (ft32_valid_pointer_mode): Likewise.
22014 (ft32_addr_space_pointer_mode): Likewise.
22015 (ft32_addr_space_legitimate_address_p): Likewise.
22016 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
22018 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
22019 (ix86_emit_outlined_ms2sysv_restore): Likewise.
22020 (iamcu_alignment): Likewise.
22021 (canonicalize_vector_int_perm): Likewise.
22022 (ix86_noce_conversion_profitable_p): Likewise.
22023 (ix86_mpx_bound_mode): Likewise.
22024 (ix86_operands_ok_for_move_multiple): Likewise.
22025 * config/microblaze/microblaze-protos.h
22026 (microblaze_expand_conditional_branch_reg): Likewise.
22027 * config/microblaze/microblaze.c
22028 (microblaze_expand_conditional_branch_reg): Likewise.
22029 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
22031 (rs6000_reassociation_width): Likewise.
22032 (rs6000_invalid_binary_op): Likewise.
22033 (fusion_p9_p): Likewise.
22034 (emit_fusion_p9_load): Likewise.
22035 (emit_fusion_p9_store): Likewise.
22036 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
22038 (riscv_hard_regno_mode_ok_p): Likewise.
22039 (riscv_address_insns): Likewise.
22040 (riscv_split_symbol): Likewise.
22041 (riscv_legitimize_move): Likewise.
22042 (riscv_function_value): Likewise.
22043 (riscv_hard_regno_nregs): Likewise.
22044 (riscv_expand_builtin): Likewise.
22045 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
22046 (riscv_build_integer): Likewise.
22047 (riscv_split_integer): Likewise.
22048 (riscv_legitimate_constant_p): Likewise.
22049 (riscv_cannot_force_const_mem): Likewise.
22050 (riscv_regno_mode_ok_for_base_p): Likewise.
22051 (riscv_valid_base_register_p): Likewise.
22052 (riscv_valid_offset_p): Likewise.
22053 (riscv_valid_lo_sum_p): Likewise.
22054 (riscv_classify_address): Likewise.
22055 (riscv_legitimate_address_p): Likewise.
22056 (riscv_address_insns): Likewise.
22057 (riscv_load_store_insns): Likewise.
22058 (riscv_force_binary): Likewise.
22059 (riscv_split_symbol): Likewise.
22060 (riscv_force_address): Likewise.
22061 (riscv_legitimize_address): Likewise.
22062 (riscv_move_integer): Likewise.
22063 (riscv_legitimize_const_move): Likewise.
22064 (riscv_legitimize_move): Likewise.
22065 (riscv_address_cost): Likewise.
22066 (riscv_subword): Likewise.
22067 (riscv_output_move): Likewise.
22068 (riscv_canonicalize_int_order_test): Likewise.
22069 (riscv_emit_int_order_test): Likewise.
22070 (riscv_function_arg_boundary): Likewise.
22071 (riscv_pass_mode_in_fpr_p): Likewise.
22072 (riscv_pass_fpr_single): Likewise.
22073 (riscv_pass_fpr_pair): Likewise.
22074 (riscv_get_arg_info): Likewise.
22075 (riscv_function_arg): Likewise.
22076 (riscv_function_arg_advance): Likewise.
22077 (riscv_arg_partial_bytes): Likewise.
22078 (riscv_function_value): Likewise.
22079 (riscv_pass_by_reference): Likewise.
22080 (riscv_setup_incoming_varargs): Likewise.
22081 (riscv_print_operand): Likewise.
22082 (riscv_elf_select_rtx_section): Likewise.
22083 (riscv_save_restore_reg): Likewise.
22084 (riscv_for_each_saved_reg): Likewise.
22085 (riscv_register_move_cost): Likewise.
22086 (riscv_hard_regno_mode_ok_p): Likewise.
22087 (riscv_hard_regno_nregs): Likewise.
22088 (riscv_class_max_nregs): Likewise.
22089 (riscv_memory_move_cost): Likewise.
22090 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
22091 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
22092 (rl78_addr_space_address_mode): Likewise.
22093 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22095 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
22096 (rs6000_reassociation_width): Likewise.
22097 (rs6000_invalid_binary_op): Likewise.
22098 (fusion_p9_p): Likewise.
22099 (emit_fusion_p9_load): Likewise.
22100 (emit_fusion_p9_store): Likewise.
22101 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
22102 (ok_for_simple_move_operands): Likewise.
22103 (ok_for_simple_move_strict_operands): Likewise.
22104 (ok_for_simple_arith_logic_operands): Likewise.
22105 (visium_legitimize_reload_address): Likewise.
22106 (visium_select_cc_mode): Likewise.
22107 (output_cbranch): Likewise.
22108 (visium_split_double_move): Likewise.
22109 (visium_expand_copysign): Likewise.
22110 (visium_expand_int_cstore): Likewise.
22111 (visium_expand_fp_cstore): Likewise.
22112 * config/visium/visium.c (visium_pass_by_reference): Likewise.
22113 (visium_function_arg): Likewise.
22114 (visium_function_arg_advance): Likewise.
22115 (visium_libcall_value): Likewise.
22116 (visium_setup_incoming_varargs): Likewise.
22117 (visium_legitimate_constant_p): Likewise.
22118 (visium_legitimate_address_p): Likewise.
22119 (visium_legitimize_address): Likewise.
22120 (visium_secondary_reload): Likewise.
22121 (visium_register_move_cost): Likewise.
22122 (visium_memory_move_cost): Likewise.
22123 (prepare_move_operands): Likewise.
22124 (ok_for_simple_move_operands): Likewise.
22125 (ok_for_simple_move_strict_operands): Likewise.
22126 (ok_for_simple_arith_logic_operands): Likewise.
22127 (visium_function_value_1): Likewise.
22128 (rtx_ok_for_offset_p): Likewise.
22129 (visium_legitimize_reload_address): Likewise.
22130 (visium_split_double_move): Likewise.
22131 (visium_expand_copysign): Likewise.
22132 (visium_expand_int_cstore): Likewise.
22133 (visium_expand_fp_cstore): Likewise.
22134 (visium_split_cstore): Likewise.
22135 (visium_select_cc_mode): Likewise.
22136 (visium_split_cbranch): Likewise.
22137 (output_cbranch): Likewise.
22138 (visium_print_operand_address): Likewise.
22139 * expmed.c (flip_storage_order): Likewise.
22140 * expmed.h (emit_cstore): Likewise.
22141 (flip_storage_order): Likewise.
22142 * genrecog.c (validate_pattern): Likewise.
22143 * hsa-gen.c (gen_hsa_addr): Likewise.
22144 * internal-fn.c (expand_arith_overflow): Likewise.
22145 * ira-color.c (allocno_copy_cost_saving): Likewise.
22146 * lra-assigns.c (find_hard_regno_for_1): Likewise.
22147 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
22148 (process_invariant_for_inheritance): Likewise.
22149 * lra-eliminations.c (move_plus_up): Likewise.
22150 * omp-low.c (lower_oacc_reductions): Likewise.
22151 * simplify-rtx.c (simplify_subreg): Likewise.
22152 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
22153 (TARGET_CHKP_BOUND_MODE): Likewise..
22154 * targhooks.c (default_chkp_bound_mode): Likewise.
22155 (default_setup_incoming_vararg_bounds): Likewise.
22156 * targhooks.h (default_chkp_bound_mode): Likewise.
22157 (default_setup_incoming_vararg_bounds): Likewise.
22158 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
22159 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
22160 (have_whole_vector_shift): Likewise.
22161 * tree-vect-stmts.c (vectorizable_load): Likewise.
22162 * doc/tm.texi: Regenerate.
22164 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22166 Graceful degrade if Binutils PR21472 is not available.
22169 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
22170 .rodata in flash test fails.
22171 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
22172 * confgure: Regenerate.
22173 * config.in: Regenerate.
22174 * config/avr/avr.c (avr_asm_named_section)
22175 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
22176 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
22177 (avr_asm_init_sections): Same.
22179 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22181 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
22182 (fma<VH:mode>4_intrinsic): Likewise.
22183 (*fmsub<VCVTF:mode>4): Likewise.
22184 (*fmsub<VH:mode>4_intrinsic): Likewise.
22186 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22189 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
22190 Don't depend on "optimize > 0".
22191 (out_movhi_r_mr, out_movqi_mr_r): Same.
22192 (out_movhi_mr_r, out_movqi_r_mr): Same.
22193 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
22194 io_address_operand on "optimize > 0".
22196 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22198 * tree-loop-distribution.c: Add general explanantion on the pass.
22199 (generate_loops_for_partition): Mark distributed loop.
22200 (pg_add_dependence_edges): New parameter. Handle alias data
22201 dependence specially and record it in the parameter if asked.
22202 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
22203 (init_partition_graph_vertices, add_partition_graph_edge): New.
22204 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
22205 (free_partition_graph_vdata, build_partition_graph): New.
22206 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
22207 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
22208 (data_ref_segment_size, latch_dominated_by_data_ref): New.
22209 (compute_alias_check_pairs, version_loop_by_alias_check): New.
22210 (version_for_distribution_p, finalize_partitions): New.
22211 (distribute_loop): Handle alias data dependence specially. Factor
22212 out loop fusion code as functions and call these functions.
22214 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22216 * tree-loop-distribution.c (classify_partition): New parameter and
22217 better handle reduction statement.
22218 (rdg_build_partitions): Revise comment.
22219 (distribute_loop): Compute statements in all partitions and pass it
22220 to classify_partition.
22222 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22224 * tree-loop-distribution.c (enum partition_type): New.
22225 (struct partition): New field type.
22226 (partition_merge_into): Add parameter. Update partition type.
22227 (data_dep_in_cycle_p, update_type_for_merge): New functions.
22228 (build_rdg_partition_for_vertex): Compute partition type.
22229 (rdg_build_partitions): Dump partition type.
22230 (distribute_loop): Update calls to partition_merge_into.
22232 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22234 * tree-loop-distribution.c (struct ddr_hasher): New.
22235 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
22237 (classify_partition): Call get_data_dependence.
22238 (pg_add_dependence_edges): Ditto.
22239 (distribute_loop): Release data dependence hash table.
22241 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22243 * tree-loop-distribution.c (ref_base_address): Delete.
22244 (similar_memory_accesses): Rename ...
22245 (share_memory_accesses): ... to this. Check if partitions access
22246 the same memory reference.
22247 (distribute_loop): Call share_memory_accesses.
22249 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22251 * tree-loop-distribution.c (struct partition): New field recording
22252 its data reference.
22253 (partition_alloc, partition_free): Init and release data refs.
22254 (partition_merge_into): Merge data refs.
22255 (build_rdg_partition_for_vertex): Collect data refs for partition.
22256 (pg_add_dependence_edges): Change parameters from vector to bitmap.
22258 (distribute_loop): Remve data refs from vertice data of partition
22261 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22263 * tree-loop-distribution.c (params.h): Include header file.
22264 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
22265 (datarefs_vec): New global var.
22266 (create_rdg_vertices): Use datarefs_vec directly.
22267 (free_rdg): Don't free data references.
22268 (build_rdg): Update use. Don't free data references.
22269 (distribute_loop): Compute global variable for data references.
22270 Bail out if there are too many data references.
22272 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22274 * tree-loop-distribution.c (loop_nest): New global var.
22275 (build_rdg): Use loop directly, rather than loop nest.
22276 (pg_add_dependence_edges): Remove loop nest parameter. Use global
22278 (distribute_loop): Compute global variable loop nest. Update use.
22280 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22282 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
22283 (partition_merge_into): New parameter. Dump reason for fusion.
22284 (distribute_loop): Update use of partition_merge_into.
22286 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22288 * tree-loop-distribution.c (bb_top_order_index): New.
22289 (bb_top_order_index_size, bb_top_order_cmp): New.
22290 (stmts_from_loop): Use topological order.
22291 (pass_loop_distribution::execute): Compute and release topological
22292 order for basic blocks.
22294 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22296 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
22299 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22301 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
22302 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
22303 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
22304 * internal-fn.def (LOOP_DIST_ALIAS): New.
22305 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
22306 (fold_loop_internal_call): ... this.
22307 (vect_loop_dist_alias_call): New function.
22308 (set_uid_loop_bbs): Call fold_loop_internal_call.
22309 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
22312 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22315 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
22316 Require dead FLAGS_REG at the beginning of a peephole.
22318 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22321 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
22322 arguments in the call to __builtin_ia32_sbb_u32.
22323 (_subborrow_u64): Swap _X and _Y arguments in the call to
22324 __builtin_ia32_sbb_u64.
22326 2017-07-04 Jakub Jelinek <jakub@redhat.com>
22329 * tree-vrp.c (compare_assert_loc): Turn into a function template
22330 with stable template parameter. Only test if a->e is NULL,
22331 !a->e == !b->e has been verified already. Use e == NULL or
22332 e != NULL instead of e or ! e tests. If stable is true, don't use
22333 iterative_hash_expr, on the other side allow a or b or both NULL
22334 and sort the NULLs last.
22335 (process_assert_insertions): Sort using compare_assert_loc<false>
22336 instead of compare_assert_loc, later sort using
22337 compare_assert_loc<true> before calling process_assert_insertions_for
22338 in a loop. Use break instead of continue once seen NULL pointer.
22340 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22342 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
22343 Cortex-R7 and Cortex-R8 processors.
22345 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22347 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
22348 uninitialized while src is not.
22350 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
22352 * common/config/arm/arm-common.c: Adjust include path for
22354 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
22355 (arm-cpu.h): Create in build directory. Adjust dependency rules.
22356 (arm-cpu-data.h): Likewise.
22357 (arm-cpu-cdata.h): Likewise.
22358 * config/arm/arm-cpu.h: Delete.
22359 * config/arm/arm-cpu-cdata.h: Delete.
22360 * config/arm/arm-cpu-data.h: Delete.
22362 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
22364 * config/arm/arm-cpus.in (cortex-a55): New.
22365 (cortex-a75): Likewise.
22366 (cortex-a75.cortex-a55): Likewise.
22367 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
22369 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
22370 * config/arm/arm-cpu-cdata.h: Regenerate.
22371 * config/arm/arm-cpu-data.h: Regenerate.
22372 * config/arm/arm-cpu.h: Regenerate.
22373 * config/arm/arm-tables.opt: Regenerate.
22374 * config/arm/arm-tune.md: Regenerate.
22376 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22378 * haifa-sched.c (sched_create_recovery_edges): Update profile.
22380 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22382 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
22385 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
22387 PR tree-optimization/81292
22388 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
22389 full_string_p, also call adjust_related_strinfos if the adjustment
22390 is simple, otherwise invalidate related strinfos.
22392 2017-07-04 Martin Liska <mliska@suse.cz>
22395 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
22396 newly created variable as DECL_IGNORED_P.
22398 2017-07-04 Martin Liska <mliska@suse.cz>
22401 * ipa-inline.c (inline_small_functions):
22402 Use xstrdup_for_dump.
22404 2017-07-04 Tom de Vries <tom@codesourcery.com>
22406 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
22408 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22411 * config/darwin.c (darwin_function_switched_text_sections):
22414 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
22416 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
22418 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22420 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
22422 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22424 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
22425 min_profitable_iters, and th as inclusive lower bounds.
22426 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
22427 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
22428 for min_profitable_iters and min_profitable_estimate.
22429 (vect_transform_loop): Treat th as an inclusive lower bound.
22430 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22432 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22435 * config/darwin.c (darwin_function_switched_text_sections):
22436 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
22437 in two pieces, and suppress the use of buf.
22439 2017-07-03 Nathan Sidwell <nathan@acm.org>
22441 * hash-table.h (hash_table_mod1): Fix indentation.
22443 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22445 PR middle-end/81290
22446 * predict.c (force_edge_cold): Be more careful about propagation
22448 * profile-count.h (profile_probability::guessed,
22449 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
22451 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
22453 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
22455 * doc/invoke.texi (rcpc architecture extension): Document it.
22457 2017-07-03 Richard Biener <rguenther@suse.de>
22459 PR tree-optimization/60510
22460 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
22461 the scalar reduction PHI and use it.
22462 (vectorizable_reduction): Properly guard the single_defuse_cycle
22463 path for non-SLP reduction chains where we cannot use it.
22464 Rework reduc_def/index and vector type deduction. Rework
22465 vector operand gathering during reduction op code-gen.
22466 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
22467 chains dissolve the chain and leave it to non-SLP reduction
22470 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22472 * tree-data-ref.h (dr_alignment): Declare.
22473 * tree-data-ref.c (dr_alignment): New function.
22474 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
22475 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
22477 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
22479 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22481 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
22482 and base_misalignment fields.
22483 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
22484 * tree-data-ref.c: Include builtins.h.
22485 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
22486 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
22487 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
22488 * tree-vect-data-refs.c: Include tree-cfg.h.
22489 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
22490 fields instead of calculating an alignment here.
22491 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
22492 innermost_loop_behavior fields.
22494 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22496 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
22498 (DR_STEP_ALIGNMENT): New macro.
22499 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
22500 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
22501 (create_data_ref): Print it.
22502 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
22503 to tell whether the step preserves vector (mis)alignment.
22504 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22505 Move the check for an integer step and generalise to all INTEGER_CST.
22506 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
22507 Print the outer step alignment.
22509 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22511 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
22512 with offset_alignment.
22513 (DR_ALIGNED_TO): Delete.
22514 (DR_OFFSET_ALIGNMENT): New macro.
22515 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
22516 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
22517 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
22518 (create_data_ref): Likewise.
22519 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22520 (vect_analyze_data_refs): Likewise.
22521 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
22522 creating dummy innermost behavior.
22524 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22526 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
22527 with a "innermost_loop_behavior *" and refeence tree.
22528 * tree-data-ref.c (dr_analyze_innermost): Likewise.
22529 (create_data_ref): Update call accordingly.
22530 * tree-predcom.c (find_looparound_phi): Likewise.
22532 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22534 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
22535 fields with dr_wrt_vec_loop.
22536 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
22537 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
22538 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
22539 (vect_dr_behavior): New function.
22540 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22541 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
22542 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
22543 track whether the step preserves the misalignment.
22544 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22545 Use vect_dr_behavior.
22546 (vect_setup_realignment): Update call accordingly.
22547 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
22548 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
22549 call to vect_create_addr_base_for_vector_ref.
22550 (vect_create_cond_for_align_checks): Likewise.
22551 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
22552 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
22553 (vect_recog_mask_conversion_pattern): Likewise.
22554 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
22555 (new_stmt_vec_info): Remove redundant zeroing.
22557 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22559 * common/config/arm/arm-common.c (arm_be8_option): New function.
22560 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
22561 (ISA_ARMv6): Add isa_bit_be8.
22562 * config/arm/arm.h (arm_be8_option): Add prototype.
22563 (BE8_SPEC_FUNCTION): New define.
22564 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
22565 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
22566 (mlittle-endian): Similarly.
22567 (mbe8, mbe32): New options.
22568 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
22569 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
22571 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22573 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
22575 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22577 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
22578 (cleanup_tree_cfg_bb): Use it.
22579 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
22581 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
22583 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22586 * loop-doloop.c (add_test): Update profile.
22588 2017-07-03 Martin Liska <mliska@suse.cz>
22591 * sanopt.c (rewrite_usage_of_param): New function.
22592 (sanitize_rewrite_addressable_params): Likewise.
22593 (pass_sanopt::execute): Call rewrite_usage_of_param.
22595 2017-07-03 Richard Biener <rguenther@suse.de>
22597 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
22598 back to using VIEW_CONVERT_EXPR.
22600 2017-07-03 Martin Liska <mliska@suse.cz>
22603 * doc/extend.texi: Document when a resolver function is
22604 generated for target_clones.
22606 2017-07-03 Martin Liska <mliska@suse.cz>
22608 * asan.c (asan_emit_stack_protection): Unpoison just red zones
22609 and shadow memory of auto variables which are subject of
22610 use-after-scope sanitization.
22611 (asan_expand_mark_ifn): Add do set only when is_poison.
22613 2016-07-03 Richard Biener <rguenther@suse.de>
22615 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
22617 (vect_force_simple_reduction): Record reduction def -> phi mapping.
22618 (vectorizable_reduction): Perform reduction PHI creation when
22619 visiting a reduction PHI and adjust and simplify code generation
22620 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
22621 (vect_transform_loop): Visit reduction PHIs.
22622 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
22623 defs into the SLP tree.
22624 (vect_build_slp_tree): Reduction defs terminate the recursion.
22625 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
22627 (vect_get_vec_defs_for_stmt_copy): Export.
22628 (vect_get_vec_defs): Likewise.
22629 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
22631 (vect_get_vec_defs_for_stmt_copy): Declare.
22632 (vect_get_vec_defs): Likewise.
22634 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22636 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
22637 parameter with a "loop" parameter and use it instead of the
22638 loop containing DR_STMT. Don't check simple_iv when doing
22639 BB analysis. Describe the two analysis modes in the comment.
22641 2017-07-03 Tom de Vries <tom@codesourcery.com>
22643 PR tree-optimization/69468
22644 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
22645 (find_same_succ_bb): Handle ignore_edge_flags.
22647 2017-07-03 Tom de Vries <tom@codesourcery.com>
22649 PR tree-optimization/81192
22650 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
22652 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
22654 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
22656 2017-07-03 Tom de Vries <tom@codesourcery.com>
22658 PR tree-optimization/81192
22659 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
22660 BB_SAME_SUCC (bb) == NULL.
22662 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22664 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
22667 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22669 * dumpfile.c: Include profile-count.h
22670 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
22672 (insert_cond_bb): Update profile.
22673 * tree-cfg.h (insert_cond_bb): Update prototype.
22674 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
22675 * tree-dump.c: Do not include tree-cfg.
22677 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22679 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
22681 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22683 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
22686 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22688 * tree-complex.c (expand_complex_div_wide): update profile.
22690 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22691 Alan Hayward <alan.hayward@arm.com>
22692 David Sherwood <david.sherwood@arm.com>
22694 * Makefile.in (MACHMODE_H): Remove insn-modes.h
22695 (CORETYPES_H): New define.
22696 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
22697 (insn-modes-inline.h, s-modes-inline-h): New rules.
22698 (generated_files): Add insn-modes-inline.h.
22699 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
22700 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
22701 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
22702 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
22703 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
22704 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
22705 (build/gencodes.o, build/genconditions.o): Likewise.
22706 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
22707 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
22708 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
22709 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
22710 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
22711 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
22712 * coretypes.h: Include everything up to real.h for generators.
22713 Include insn-modes.h first. Include wide-int-print.h after
22714 wide-int.h. Include insn-modes-inline.h and then machmode.h.
22715 * machmode.h: Don't include insn-modes.h here.
22716 * function-tests.c: Remove includes of signop.h, machmode.h,
22717 double-int.h and wide-int.h.
22719 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
22721 * optc-save-gen.awk: Likewise.
22722 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
22723 * godump.c: Remove include of wide-int-print.h.
22724 * pretty-print.h: Likewise.
22725 * wide-int-print.cc: Likewise.
22726 * wide-int.cc: Likewise.
22727 * hash-map-tests.c: Remove include of signop.h.
22728 * hash-set-tests.c: Likewise.
22729 * rtl-tests.c: Likewise.
22730 * mkconfig.sh: Remove include of machmode.h.
22731 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
22733 (emit_insn_modes_inline_h): ...this new function. Emit the code
22734 into an insn-modes-inline.h header file, adding appropriate
22735 include guards and end comments.
22736 (emit_insn_modes_c_header): Remove include of machmode.h.
22737 (emit_min_insn_modes_c_header): Include coretypes.h rather than
22739 (main): Handle -i flag and call emit_insn_modes_inline_h when
22742 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22744 * tree-ssa-strlen.c (strinfo): Rename the length field to
22745 nonzero_chars. Add a full_string_p field.
22746 (compare_nonzero_chars, zero_length_string_p): New functions.
22747 (get_addr_stridx): Add an offset_out parameter.
22748 Use compare_nonzero_chars.
22749 (get_stridx): Update accordingly. Use compare_nonzero_chars.
22750 (new_strinfo): Update after above changes to strinfo.
22751 (set_endptr_and_length): Set full_string_p.
22752 (get_string_length): Update after above changes to strinfo.
22753 (unshare_strinfo): Update call to new_strinfo.
22754 (maybe_invalidate): Likewise.
22755 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
22756 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
22757 as a uhwi instead of an shwi. Update after above changes to
22758 strinfo and new_strinfo.
22759 (zero_length_string): Assert that chainsi contains full strings.
22760 Use zero_length_string_p. Update call to new_strinfo.
22761 (adjust_related_strinfos): Update after above changes to strinfo.
22762 Copy full_string_p from origsi.
22763 (adjust_last_stmt): Use zero_length_string_p.
22764 (handle_builtin_strlen): Update after above changes to strinfo and
22765 new_strinfo. Install the lhs as the string length if the previous
22766 entry didn't describe a full string.
22767 (handle_builtin_strchr): Update after above changes to strinfo
22769 (handle_builtin_strcpy): Likewise.
22770 (handle_builtin_strcat): Likewise.
22771 (handle_builtin_malloc): Likewise.
22772 (handle_pointer_plus): Likewise.
22773 (handle_builtin_memcpy): Likewise. Track nonzero characters
22774 that aren't necessarily followed by a nul terminator.
22775 (handle_char_store): Likewise.
22777 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22779 PR tree-optimization/80769
22780 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
22781 for malloc and calloc. Document the new invariant that all related
22782 strinfos have delayed lengths or none do.
22783 (verify_related_strinfos): Move earlier in file.
22784 (set_endptr_and_length): New function, split out from...
22785 (get_string_length): ...here. Also set the lengths of related
22787 (zero_length_string): Assert that chainsi has known (rather than
22789 (adjust_related_strinfos): Likewise.
22791 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22793 PR tree-optimization/81136
22794 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
22795 assert that two references with the same misalignment have the same
22796 compile-time misalignment if those compile-time misalignments
22799 2017-07-01 Andi Kleen <ak@linux.intel.com>
22801 * print-tree.c (print_node): Print all attributes.
22803 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22805 * cfg.c (scale_bbs_frequencies): New function.
22806 * cfg.h (scale_bbs_frequencies): Declare it.
22807 * cfgloopanal.c (single_likely_exit): Cleanup.
22808 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
22810 (scale_loop_profile): Likewise.
22811 (loop_version): Likewise.
22812 (create_empty_loop_on_edge): Update.
22813 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
22814 scale_loop_frequencies, scale_loop_profile, loopify,
22815 loop_version): Update prototypes.
22816 * modulo-sched.c (sms_schedule): Update.
22817 * predict.c (unlikely_executed_edge_p): Also check probability.
22818 (probably_never_executed_edge_p): Fix typo.
22819 * tree-if-conv.c (version_loop_for_if_conversion): Update.
22820 * tree-parloops.c (gen_parallel_loop): Update.
22821 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
22822 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22823 * tree-ssa-loop-split.c (split_loop): Update.
22824 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
22825 * tree-vect-loop-manip.c (vect_do_peeling): Update.
22826 (vect_loop_versioning): Update.
22827 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22829 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22831 * trans-mem.c (split_bb_make_tm_edge): Update profile.
22833 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22835 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
22836 to keep profile consistent.
22838 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22840 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
22841 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
22842 * profile-count.h (max_safe_multiplier): Make unsigned.
22843 (profile_count::guessed_zero): New.
22845 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22847 * bb-reorder.c (fix_up_crossing_landing_pad,
22848 fix_crossing_conditional_branches): Use make_single_succ_edge
22849 to keep profile consistent.
22851 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22853 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
22856 2017-07-01 Jakub Jelinek <jakub@redhat.com>
22859 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
22860 the right scopes, make sure cond_jump isn't preserved between multiple
22861 iterations. Search for fallthru edge whenever there are 3+ edges and
22862 use find_fallthru_edge for it.
22864 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22866 Patch by Alexander Monakov <amonakov@ispras.ru>
22867 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
22868 probabilities consistently.
22870 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22872 * pa.c (pa_expand_compare_and_swap_loop): Update call of
22873 emit_cmp_and_jump_insns.
22875 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22878 * tree-inline.c (expand_call_inline): Combine profile statuses.
22880 2017-06-30 Andrew Pinski <apinski@cavium.com>
22882 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
22883 fold_stmt returned true.
22885 2017-06-30 Nathan Sidwell <nathan@acm.org>
22887 * ggc.h (empty_string): Delete.
22888 * cfgexpand.c (expand_asm_stmt): Use plain "".
22889 * optabs.c (expand_asm_memory_barrier): Likewise.
22890 * stringpool.c (empty_string): Delete.
22891 (digit_vector, digit_string): Delete.
22892 (ggc_alloc_string): Use plain "", don't optimize single digit
22893 strings. Use ggc_alloc_atomic.
22895 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
22897 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
22898 comparison set and one other set, use the cost of the non-comparison
22901 2017-06-30 Nathan Sidwell <nathan@acm.org>
22903 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
22906 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
22908 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
22909 loops. Remove now unneeded calls to gimple_switch_set_label() that
22910 just set removed labels to NULL_TREE.
22912 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
22914 * tree-ssanames.c (set_range_info_raw): Abstract from ...
22915 (set_range_info): ...here. Only call set_range_info_raw if domain
22917 (set_nonzero_bits): Call set_range_info_raw.
22918 * tree-ssanames.h (set_range_info_raw): New.
22920 2017-06-30 Jakub Jelinek <jakub@redhat.com>
22923 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
22924 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
22925 of nonimmediate_operand and <store_mask_constraint> instead of m
22926 for the input operand. For V8FI iterator, always split if input
22927 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
22928 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
22929 <store_mask_predicate> instead of register_operand and
22930 <store_mask_constraint> instead of v for the input operand. Make
22931 sure both operands aren't MEMs for if not <mask_applied>.
22933 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
22935 * lto-wrapper.c (copy_file) Close both file descriptors before
22938 2017-06-30 Martin Liska <mliska@suse.cz>
22941 * multiple_target.c (create_dispatcher_calls): Make ifunc
22942 also for function that don't have calls or are not referenced.
22944 2017-06-30 Richard Biener <rguenther@suse.de>
22946 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
22947 analyze the first scalar stmt. Move vector type computation
22948 for the BB case here from ...
22949 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
22950 live operation processing in the SLP case properly.
22952 2017-06-30 Richard Biener <rguenther@suse.de>
22954 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
22956 2017-06-30 Martin Liska <mliska@suse.cz>
22959 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
22960 before BUILT_IN_UNWIND_RESUME when ASAN is used.
22962 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
22964 * doc/invoke.texi (AArch64): Add missing options and remove redundant
22967 2017-06-30 Richard Biener <rguenther@suse.de>
22969 PR tree-optimization/81249
22970 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
22971 condition reduction result to original scalar type.
22973 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22975 * profile-count.h (enum profile_quality): Fix typos and whitespace
22978 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22980 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
22981 type for branch probabilities.
22983 2017-06-29 Julian Brown <julian@codesourcery.com>
22984 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22986 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
22987 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
22988 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
22989 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
22991 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22993 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
22994 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
22995 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
22996 CC usage from generic code to here.
22997 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
22998 CC usage into the target macros.
23000 2017-06-29 Maya Rashish <coypu@sdf.org>
23002 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
23005 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23007 * arm/arm-builtins.c: Include profile-count.h
23008 * except.c (sjlj_emit_function_enter): Use
23009 profile_probability::unlikely.
23011 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23013 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
23014 and tocrel_offset be pointer args rather than implicitly using
23016 (legitimate_constant_pool_address_p, rs6000_emit_move,
23017 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
23018 tocrel_offset and use in toc_relative_expr_p call.
23019 (print_operand, print_operand_address): Use static tocrel_base_oac
23020 and tocrel_offset_oac.
23021 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
23024 2017-06-29 Maya Rashish <coypu@sdf.org>
23026 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
23028 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
23030 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
23031 objects, take into account only the alignment of 'op0' and 'mode1' if
23034 2017-06-29 Steve Ellcey <sellcey@cavium.com>
23036 * ccmp.c (ccmp_tree_comparison_p): New function.
23037 (ccmp_candidate_p): Update to use above function.
23038 (get_compare_parts): New function.
23039 (expand_ccmp_next): Update to use new functions.
23040 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
23042 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
23043 take mode as argument.
23044 * ccmp.h (expand_ccmp_expr): Add mode as argument.
23045 * expr.c (expand_expr_real_1): Pass mode as argument.
23047 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
23049 * combine.c (combine_instructions): Print insns to dump_file, together
23052 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23054 * asan.c (asan_emit_stack_protection): Update.
23055 (create_cond_insert_point): Update.
23056 * auto-profile.c (afdo_propagate_circuit): Update.
23057 * basic-block.h (struct edge_def): Turn probability to
23058 profile_probability.
23059 (EDGE_FREQUENCY): Update.
23060 * bb-reorder.c (find_traces_1_round): Update.
23061 (better_edge_p): Update.
23062 (sanitize_hot_paths): Update.
23063 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
23064 (make_single_succ_edge): Update.
23065 (check_bb_profile): Update.
23066 (dump_edge_info): Update.
23067 (update_bb_profile_for_threading): Update.
23068 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
23069 probabilitycount to 0.
23070 * cfgbuild.c (compute_outgoing_frequencies): Update.
23071 * cfgcleanup.c (try_forward_edges): Update.
23072 (outgoing_edges_match): Update.
23073 (try_crossjump_to_edge): Update.
23074 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
23075 (expand_gimple_tailcall): Update.
23076 (construct_init_block): Use make_single_succ_edge.
23077 (construct_exit_block): Use make_single_succ_edge.
23078 * cfghooks.c (verify_flow_info): Update.
23079 (redirect_edge_succ_nodup): Update.
23080 (split_edge): Update.
23081 (account_profile_record): Update.
23082 * cfgloopanal.c (single_likely_exit): Update.
23083 * cfgloopmanip.c (scale_loop_profile): Update.
23084 (set_zero_probability): Remove.
23085 (duplicate_loop_to_header_edge): Update.
23086 * cfgloopmanip.h (loop_version): Update prototype.
23087 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
23088 (force_nonfallthru_and_redirect): Update.
23089 (update_br_prob_note): Update.
23090 (rtl_verify_edges): Update.
23091 (purge_dead_edges): Update.
23092 (rtl_lv_add_condition_to_bb): Update.
23093 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
23094 * cgraphunit.c (init_lowered_empty_function): Update.
23095 (cgraph_node::expand_thunk): Update.
23096 * cilk-common.c: Include profile-count.h
23097 * dojump.c (inv): Remove.
23098 (jumpifnot): Update.
23099 (jumpifnot_1): Update.
23100 (do_jump_1): Update.
23102 (do_jump_by_parts_greater_rtx): Update.
23103 (do_compare_rtx_and_jump): Update.
23104 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
23105 do_jump_1. do_compare_rtx_and_jump): Update prototype.
23106 * dwarf2cfi.c: Include profile-count.h
23107 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
23108 (sjlj_emit_dispatch_table): Likewise.
23109 * explow.c: Include profile-count.h
23110 * expmed.c (emit_store_flag_force): Update.
23111 (do_cmp_and_jump): Update.
23112 * expr.c (compare_by_pieces_d::generate): Update.
23113 (compare_by_pieces_d::finish_mode): Update.
23114 (emit_block_move_via_loop): Update.
23115 (store_expr_with_bounds): Update.
23116 (store_constructor): Update.
23117 (expand_expr_real_2): Update.
23118 (expand_expr_real_1): Update.
23119 * expr.h (try_casesi, try_tablejump): Update prototypes.
23120 * gimple-pretty-print.c (dump_probability): Update.
23121 (dump_profile): New.
23122 (dump_gimple_label): Update.
23123 (dump_gimple_bb_header): Update.
23124 * graph.c (draw_cfg_node_succ_edges): Update.
23125 * hsa-gen.c (convert_switch_statements): Update.
23126 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
23127 (find_if_case_1): Update.
23128 (find_if_case_2): Update.
23129 * internal-fn.c (expand_arith_overflow_result_store): Update.
23130 (expand_addsub_overflow): Update.
23131 (expand_neg_overflow): Update.
23132 (expand_mul_overflow): Update.
23133 (expand_vector_ubsan_overflow): Update.
23134 * ipa-cp.c (good_cloning_opportunity_p): Update.
23135 * ipa-split.c (split_function): Use make_single_succ_edge.
23136 * ipa-utils.c (ipa_merge_profiles): Update.
23137 * loop-doloop.c (add_test): Update.
23138 (doloop_modify): Update.
23139 * loop-unroll.c (compare_and_jump_seq): Update.
23140 (unroll_loop_runtime_iterations): Update.
23141 * lra-constraints.c (lra_inheritance): Update.
23142 * lto-streamer-in.c (input_cfg): Update.
23143 * lto-streamer-out.c (output_cfg): Update.
23144 * mcf.c (adjust_cfg_counts): Update.
23145 * modulo-sched.c (sms_schedule): Update.
23146 * omp-expand.c (expand_omp_for_init_counts): Update.
23147 (extract_omp_for_update_vars): Update.
23148 (expand_omp_ordered_sink): Update.
23149 (expand_omp_for_ordered_loops): Update.
23150 (expand_omp_for_generic): Update.
23151 (expand_omp_for_static_nochunk): Update.
23152 (expand_omp_for_static_chunk): Update.
23153 (expand_cilk_for): Update.
23154 (expand_omp_simd): Update.
23155 (expand_omp_taskloop_for_outer): Update.
23156 (expand_omp_taskloop_for_inner): Update.
23157 * omp-simd-clone.c (simd_clone_adjust): Update.
23158 * optabs.c (expand_doubleword_shift): Update.
23159 (expand_abs): Update.
23160 (emit_cmp_and_jump_insn_1): Update.
23161 (expand_compare_and_swap_loop): Update.
23162 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
23163 * predict.c (predictable_edge_p): Update.
23164 (edge_probability_reliable_p): Update.
23165 (set_even_probabilities): Update.
23166 (combine_predictions_for_insn): Update.
23167 (combine_predictions_for_bb): Update.
23168 (propagate_freq): Update.
23169 (estimate_bb_frequencies): Update.
23170 (force_edge_cold): Update.
23171 * profile-count.c (profile_count::dump): Add missing space into dump.
23172 (profile_count::debug): Add newline.
23173 (profile_count::differs_from_p): Explicitly convert to unsigned.
23174 (profile_count::stream_in): Update.
23175 (profile_probability::dump): New member function.
23176 (profile_probability::debug): New member function.
23177 (profile_probability::differs_from_p): New member function.
23178 (profile_probability::differs_lot_from_p): New member function.
23179 (profile_probability::stream_in): New member function.
23180 (profile_probability::stream_out): New member function.
23181 * profile-count.h (profile_count_quality): Rename to ...
23182 (profile_quality): ... this one.
23183 (profile_probability): New.
23184 (profile_count): Update.
23185 * profile.c (compute_branch_probabilities): Update.
23186 * recog.c (peep2_attempt): Update.
23187 * sched-ebb.c (schedule_ebbs): Update.
23188 * sched-rgn.c (find_single_block_region): Update.
23189 (compute_dom_prob_ps): Update.
23190 (schedule_region): Update.
23191 * sel-sched-ir.c (compute_succs_info): Update.
23192 * stmt.c (struct case_node): Update.
23193 (do_jump_if_equal): Update.
23194 (get_outgoing_edge_probs): Update.
23195 (conditional_probability): Update.
23196 (emit_case_dispatch_table): Update.
23197 (expand_case): Update.
23198 (expand_sjlj_dispatch_table): Update.
23199 (emit_case_nodes): Update.
23200 * targhooks.c: Update.
23201 * tracer.c (better_p): Update.
23202 (find_best_successor): Update.
23203 * trans-mem.c (expand_transaction): Update.
23204 * tree-call-cdce.c: Update.
23205 * tree-cfg.c (gimple_split_edge): Upate.
23206 (move_sese_region_to_fn): Upate.
23207 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
23208 * tree-eh.c (lower_resx): Upate.
23209 (cleanup_empty_eh_move_lp): Upate.
23210 * tree-if-conv.c (version_loop_for_if_conversion): Update.
23211 * tree-inline.c (copy_edges_for_bb): Update.
23212 (copy_cfg_body): Update.
23213 * tree-parloops.c (gen_parallel_loop): Update.
23214 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
23215 (gimple_gen_time_profiler): Update.
23216 * tree-ssa-dce.c (remove_dead_stmt): Update.
23217 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
23218 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
23219 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
23220 (unloop_loops): Update.
23221 (try_peel_loop): Update.
23222 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23223 * tree-ssa-loop-split.c (connect_loops): Update.
23224 (split_loop): Update.
23225 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
23226 (hoist_guard): Update.
23227 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
23228 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
23229 (value_replacement): Update.
23230 * tree-ssa-reassoc.c (branch_fixup): Update.
23231 * tree-ssa-tail-merge.c (replace_block_by): Update.
23232 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
23233 (create_edge_and_update_destination_phis): Update.
23234 (compute_path_counts): Update.
23235 (recompute_probabilities): Update.
23236 (update_joiner_offpath_counts): Update.
23237 (freqs_to_counts_path): Update.
23238 (duplicate_thread_path): Update.
23239 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
23240 (struct switch_conv_info): Update.
23241 (gen_inbound_check): Update.
23242 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23243 (vect_do_peeling): Update.
23244 (vect_loop_versioning): Update.
23245 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23246 (optimize_mask_stores): Update.
23247 * ubsan.c (ubsan_expand_null_ifn): Update.
23248 * value-prof.c (gimple_divmod_fixed_value): Update.
23249 (gimple_divmod_fixed_value_transform): Update.
23250 (gimple_mod_pow2): Update.
23251 (gimple_mod_pow2_value_transform): Update.
23252 (gimple_mod_subtract): Update.
23253 (gimple_mod_subtract_transform): Update.
23254 (gimple_ic): Update.
23255 (gimple_stringop_fixed_value): Update.
23256 (gimple_stringops_transform): Update.
23257 * value-prof.h: Update.
23259 2017-06-29 Carl Love <cel@us.ibm.com>
23261 * config/rs6000/rs6000-c.c: Add support for built-in functions
23262 vector signed int vec_signed (vector float);
23263 vector signed long long vec_signed (vector double);
23264 vector signed int vec_signed2 (vector double, vector double);
23265 vector signed int vec_signede (vector double);
23266 vector signed int vec_signedo (vector double);
23267 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
23268 instruction generator.
23269 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23270 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
23271 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
23273 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
23274 vunsignede_v2df): Add define_expands.
23275 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
23276 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
23277 VEC_UNSIGNEDO): Add definitions.
23278 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23279 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
23280 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
23281 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
23282 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
23283 * config/rs6000/altivec.h (vec_signed, vec_signed2,
23284 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
23285 vec_unsignede, vec_unsignedo): Add builtin defines.
23286 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
23288 * doc/extend.texi: Update the built-in documentation file for the
23289 new built-in functions.
23291 2017-06-29 Richard Biener <rguenther@suse.de>
23293 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
23294 reduction chains to LOOP_VINFO_REDUCTIONS.
23295 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
23296 SLP reductions after processing reduction chains.
23298 2017-06-29 Nathan Sidwell <nathan@acm.org>
23300 * builtins.c (fold_builtin_FUNCTION): Use
23301 lang_hooks.decl_printable_name.
23303 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
23305 PR middle-end/81194
23306 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
23307 with only one label.
23308 * stmt.c (expand_case): Assert NCASES is greater than one.
23310 2017-06-29 Richard Biener <rguenther@suse.de>
23312 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
23314 (group_case_labels): Likewise.
23315 (find_taken_edge): Push sanity checking on val to workers...
23316 (find_taken_edge_cond_expr): ... here
23317 (find_taken_edge_switch_expr): ... and here, handle cases
23318 with just a default label.
23319 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
23320 (group_case_labels): Likewise.
23321 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
23322 group_case_labels does anything cleanup the CFG again.
23324 2017-06-29 Bin Cheng <bin.cheng@arm.com>
23326 PR tree-optimization/81196
23327 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
23328 exit condition comparing two IVs.
23330 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
23332 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
23333 profile to the dummy entry at the end of the list of architectures.
23334 * config/arm/arm-cpu-cdata.h: Regenerated.
23336 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23337 Michael Collison <michael.collison@arm.com>
23340 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
23342 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
23343 (*aarch64_reg_<mode>3_minus_mask): New pattern.
23344 (*aarch64_<optab>_reg_di3_mask2): New pattern.
23345 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
23346 of shift when the shift amount is masked with constant equal to
23347 the size of the mode.
23348 * config/aarch64/predicates.md (subreg_lowpart_operator): New
23351 2017-06-29 Martin Liska <mliska@suse.cz>
23353 * config/i386/i386.opt: Change range from [1,5] to [0,5].
23355 2017-06-29 Yury Gribov <tetra2005@gmail.com>
23358 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
23361 (edge_growth_cache_entry::edge_growth_cache_entry): New
23363 (reset_edge_growth_cache): Update to use constructor.
23365 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23367 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
23368 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23369 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23371 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
23373 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
23374 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
23376 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
23378 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
23379 (*-linux-uclibc*): Add t-uclibc tmake_file.
23380 * config/t-musl: New.
23381 * config/t-uclibc: New.
23383 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
23385 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
23387 (gen_comm_data): Emit architectural setting of arch_prof.
23388 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
23390 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
23391 (armv8-m.base, armv8-m.main): Likewise.
23392 * arm-protos.h (arm_build_target): Add profile field.
23393 (arch_option): Likewise.
23394 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
23396 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
23397 arm_active_target.profile.
23399 2017-06-28 Richard Biener <rguenther@suse.de>
23401 PR middle-end/81227
23402 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
23403 TYPE_OVERFLOW_WRAPS.
23404 * match.pd (negate_expr_p): Likewise.
23405 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
23406 fold_build2, not fold_binary.
23408 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23410 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
23411 Convert memory address to Pmode.
23412 (aarch64_print_operand): Assert MEM operands are always Pmode.
23414 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23417 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
23418 Remove redundant if.
23419 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
23420 * config/arm/aarch-common-protos.h
23421 (aarch_forward_to_shift_is_not_shifted_re): Remove.
23422 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
23424 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
23427 * multiple_target.c (create_dispatcher_calls): Set the default
23428 clone to be static, not public.
23430 2017-06-28 Richard Biener <rguenther@suse.de>
23432 * tree-vect-loop.c (vectorizable_reduction): Move special
23433 cond reduction IV var creation ...
23434 (vect_create_epilog_for_reduction): ... here. Remove induction_index
23435 parameter. Use STMT_VINFO_VECTYPE.
23436 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
23439 2017-06-28 Martin Liska <mliska@suse.cz>
23442 * ipa-visibility.c (non_local_p): Handle visibility.
23444 2017-06-28 Martin Liska <mliska@suse.cz>
23447 * common.opt: Add IntegerRange to various options.
23448 * opt-functions.awk (integer_range_info): New function.
23449 * optc-gen.awk: Add integer_range_info to cl_options struct.
23450 * opts-common.c (decode_cmdline_option): Handle
23451 CL_ERR_INT_RANGE_ARG.
23452 (cmdline_handle_error): Likewise.
23453 * opts.c (print_filtered_help): Show valid interval in
23454 when --help is provided.
23455 * opts.h (struct cl_option): Add range_min and range_max fields.
23456 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
23458 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
23460 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
23461 (x * C EQ/NE y * C): New transformation.
23463 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
23465 * genmultilib (combination_space): Accept '+' in option names.
23467 2017-06-28 Martin Liska <mliska@suse.cz>
23470 * asan.c (instrument_derefs): Bail out inner references
23471 that are hard register variables.
23473 2017-06-28 Jakub Jelinek <jakub@redhat.com>
23476 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
23477 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
23479 2017-06-28 Richard Biener <rguenther@suse.de>
23481 * tree-vectorizer.h (vect_get_vec_defs): Remove.
23482 (vect_get_slp_defs): Adjust.
23483 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
23485 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
23487 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
23488 get_initial_defs_for_reduction instead of vect_get_vec_defs.
23489 (vectorizable_reduction): Adjust.
23490 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
23492 (vect_get_slp_defs): Likewise.
23493 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
23494 (vectorizable_bswap): Adjust.
23495 (vectorizable_call): Likewise.
23496 (vectorizable_conversion): Likewise.
23497 (vectorizable_assignment): Likewise.
23498 (vectorizable_shift): Likewise.
23499 (vectorizable_operation): Likewise.
23500 (vectorizable_store): Likewise.
23501 (vectorizable_condition): Likewise.
23502 (vectorizable_comparison): Likewise.
23504 2017-06-28 Michael Collison <michael.collison@arm.com>
23507 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
23509 (arm_gen_movmemqi): Removed unused variable 'i'.
23510 Convert 'for' loop into 'while' loop.
23511 (arm_expand_prologue): Remove last unnecessary set of insn.
23512 (thumb_pop): Remove unused variable 'pushed_words'.
23513 (thumb_exit): Remove last unnecessary set of regs_to_pop.
23515 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23517 * config/s390/predicates.md: Use s390_rel_address_ok_p.
23518 * config/s390/s390-protos.h: Add prototype of
23519 s390_rel_address_ok_p.
23520 * config/s390/s390.c (s390_got_symbol): New function.
23521 (s390_rel_address_ok_p): New function.
23522 (legitimize_pic_address): Use s390_rel_address_ok_p.
23523 (s390_load_got): Use s390_got_symbol.
23524 (s390_option_override): Issue error if
23525 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
23526 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
23528 * config/s390/s390.opt: New option mpic-data-is-text-relative.
23530 2017-06-27 Andrew Pinski <apinski@cavium.com>
23532 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
23533 (X * copysign (1.0, X)): New pattern.
23534 (X * copysign (1.0, -X)): New pattern.
23535 (copysign (-1.0, CST)): New pattern.
23537 2017-06-27 Joseph Myers <joseph@codesourcery.com>
23539 * genmultilib (combination_space): Remove variable.
23540 Validate reuse rules against regular expression for any sequence
23541 of multilib options in any order.
23543 2017-06-27 Michael Collison <michael.collison@arm.com>
23545 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
23546 call aarch64_split_simd_combine.
23547 * (aarch64_combine_internal<mode>): Delete pattern.
23548 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
23549 Allow register and subreg operands.
23551 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23553 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
23554 specific need, just fallback on defaults.
23555 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
23557 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23558 Olivier Hainque <hainque@adacore.com>
23560 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
23562 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
23563 targets. Pick a default if no particular attempt applied.
23564 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
23567 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23569 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
23570 (x86_64-wrs-vxworks7): Likewise.
23572 2017-06-27 Marek Polacek <polacek@redhat.com>
23575 * ubsan.c (instrument_null): Check get_base_address's result for null.
23577 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
23579 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
23581 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
23583 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
23584 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
23585 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
23586 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
23587 New function types.
23588 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
23589 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
23590 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
23591 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
23592 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
23593 BUILT_IN_FEUPDATEENV): New builtins.
23594 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
23595 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
23596 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
23597 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
23599 (builtin_structptr_types): Adjust size.
23600 * tree.c (builtin_structptr_types): Add four entries.
23602 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23603 Olivier Hainque <hainque@adacore.com>
23605 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
23606 (TLS_SYM): New local macro, forcing reference to __tls__ on
23607 link command lines for VxWorks 7 RTPs, triggering initialization
23609 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
23610 OS features TLS support, true for RTPs on VxWorks 7.
23611 * config/vxworks.c (vxworks_override_options): Setup emutls
23614 2017-06-27 Jakub Jelinek <jakub@redhat.com>
23616 * predict.c (test_prediction_value_range): Use -1U instead of -1
23617 to avoid narrowing conversion warning.
23618 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
23619 to avoid narrowing conversion warning.
23620 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
23622 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
23624 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23626 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
23627 64bit configurations.
23628 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
23629 (SIZE_TYPE): Likewise.
23630 * config/vxworks.c (vxworks_emutls_var_fields): Use
23631 long_unsigned_type_node instead of unsigned_type_node as the offset
23632 field type, which is "pointer" mode in emutls.c.
23634 2017-06-27 Jakub Jelinek <jakub@redhat.com>
23637 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
23639 PR middle-end/81207
23640 * gimple-fold.c (replace_call_with_call_and_fold): Handle
23641 gimple_vuse copying separately from gimple_vdef copying.
23643 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23645 * value-prof.c (free_hist): Remove call to memset and the enclosing if
23648 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
23649 Olivier Hainque <hainque@adacore.com>
23651 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
23652 for all vxworks7 targets.
23653 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
23654 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
23655 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
23656 variations for VX6/VX7 and 32/64bits later on in ...
23657 (VXWORKS_LIB_SPEC): Leverage new macros.
23658 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
23659 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
23661 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
23663 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
23664 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
23666 2017-06-26 Carl Love <cel@us.ibm.com>
23668 * config/rs6000/rs6000-c.c: Add support for built-in functions
23669 vector bool char vec_reve (vector bool char);
23670 vector signed char vec_reve (vector signed char);
23671 vector unsigned char vec_reve (vector unsigned char);
23672 vector bool int vec_reve (vector bool int);
23673 vector signed int vec_reve (vector signed int);
23674 vector unsigned int vec_reve (vector unsigned int);
23675 vector bool long long vec_reve (vector bool long long);
23676 vector signed long long vec_reve (vector signed long long);
23677 vector unsigned long long vec_reve (vector unsigned long long);
23678 vector bool short vec_reve (vector bool short);
23679 vector signed short vec_reve (vector signed short);
23680 vector double vec_reve (vector double);
23681 vector float vec_reve (vector float);
23682 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
23683 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
23684 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
23685 (altivec_vreve): New pattern.
23686 * config/rs6000/altivec.h (vec_reve): New define.
23687 * doc/extend.texi (vec_rev): Update the built-in documentation file
23688 for the new built-in functions.
23690 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23692 PR tree-optimization/71815
23693 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
23695 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
23697 (slsr_process_phi): Likewise.
23698 (replace_uncond_cands_and_profitable_phis): Don't replace a
23699 multiply candidate with a stride of 1 (copy or cast).
23700 (phi_incr_cost): Call uses_consumed_by_stmt rather than
23702 (lowest_cost_path): Likewise.
23703 (total_savings): Likewise.
23705 2017-06-26 Richard Biener <rguenther@suse.de>
23708 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
23709 Use def_builtin_pure for all gather builtins.
23711 2017-06-26 Richard Biener <rguenther@suse.de>
23713 PR tree-optimization/81203
23714 * tree-tailcall.c (find_tail_calls): Do not move stmts into
23715 non-dominating BBs.
23717 2017-06-26 Marek Polacek <polacek@redhat.com>
23720 * doc/invoke.texi: Document -Wmultistatement-macros.
23722 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
23724 * doc/sourcebuild.texi (ARM-specific attributes): Document new
23725 arm_neon_ok_no_float_abi effective target.
23727 2017-06-26 Richard Biener <rguenther@suse.de>
23729 PR tree-optimization/80928
23730 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
23731 (copy_bbs): Set BB_DUPLICATED flag early.
23732 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
23734 (execute_on_shrinking_pred): Likewise.
23735 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
23736 BB_DUPLICATED blocks.
23737 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
23738 iterate over all PHIs considering removal of *gsi.
23740 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
23742 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
23745 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23747 * config/rs6000/rs6000-string.c: (expand_block_clear,
23748 do_load_for_compare, select_block_compare_mode,
23749 compute_current_alignment, expand_block_compare,
23750 expand_strncmp_align_check, expand_strn_compare,
23751 expand_block_move, rs6000_output_load_multiple)
23752 Move functions related to string/block move/compare
23753 to a separate file.
23754 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
23755 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
23756 for this function which is now used in two files.
23757 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
23758 * config.gcc: Add rs6000-string.o to extra_objs for
23759 targets powerpc*-*-* and rs6000*-*-*.
23761 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
23764 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
23765 32-bit, since indexed is not valid for DImode.
23766 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
23767 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
23768 (define_peephole2 for Altivec d-form load): Add 32-bit support.
23769 (define_peephole2 for Altivec d-form store): Likewise.
23772 * multiple_target.c (create_dispatcher_calls): Only create the
23773 dispatcher call if the function is the default clone of a
23774 versioned function.
23776 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
23778 PR middle-end/80902
23779 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
23780 a call, force the call to not be a tail call.
23782 2017-06-23 Jeff Law <law@redhat.com>
23784 * doc/contrib.texi: Add entry for Steven Pemberton's work on
23787 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
23789 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
23790 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
23791 handling for early expansion of vector shifts (sl,sr,sra,rl).
23792 (builtin_function_type): Add vector shift right instructions
23793 to the unsigned argument list.
23795 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
23797 rtl-optimizatoin/79286
23798 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
23799 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
23800 trap. PIC register plus a const unspec without offset can never trap.
23802 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
23804 * tree.h (builtin_structptr_type): New type.
23805 (builtin_structptr_types): Declare new array.
23806 * tree.c (builtin_structptr_types): New array.
23807 (free_lang_data, build_common_tree_nodes): Use it.
23809 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
23812 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
23815 2017-06-22 Matt Turner <mattst88@gmail.com>
23817 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
23818 Lake models to skylake case. Assume skylake for unknown
23819 models with clflushopt.
23821 2017-06-22 Jeff Law <law@redhat.com>
23823 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
23824 frame sizes that do not satisfy aarch64_uimm12_shift.
23826 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
23828 * profile-count.h (apply_probability,
23829 apply_scale, probability_in): Fix checks for zero.
23831 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23833 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
23834 * doc/cppdiropts.texi (-I @var{dir}): Document it.
23836 2016-06-22 Richard Biener <rguenther@suse.de>
23838 * tree-vect-loop.c (vect_model_reduction_cost): Handle
23839 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
23840 REDUC_MAX_EXPR support.
23841 (vectorizable_reduction): Likewise.
23842 (vect_create_epilog_for_reduction): Likewise.
23844 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
23846 * match.pd (A / (1 << B) -> A >> B): New.
23847 * generic-match-head.c: Include optabs-tree.h.
23848 * gimple-match-head.c: Likewise.
23849 * optabs-tree.h (target_supports_op_p): New.
23850 * optabs-tree.c (target_supports_op_p): New.
23852 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23854 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
23855 $gcc_cv_ld --help output.
23856 (gcc_cv_ld_demangle): Likewise.
23857 (gcc_cv_ld_eh_frame_hdr): Likewise.
23858 (gcc_cv_ld_pie): Likewise.
23859 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
23860 (gcc_cv_ld_buildid): Likewise.
23861 (gcc_cv_ld_sysroot): Likewise.
23862 (ld_bndplt_support): Likewise.
23863 (ld_pushpopstate_support): Likewise.
23864 * configure: Regenerate.
23865 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
23867 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23870 * config/i386/sse.md (round<mode>2): Renumber match_dup and
23871 operands indexes to avoid gap between operands and match_dups.
23873 2017-06-21 Andrew Pinski <apinski@cavium.com>
23875 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
23876 Increment Arith_shift and Arith_shift_reg by 1.
23877 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
23879 * config/aarch64/aarch64.c (thunderx_tunings): Enable
23880 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
23881 (aarch64_strip_extend): Add new argument and test for it.
23882 (aarch64_cheap_mult_shift_p): New function.
23883 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
23884 add a cost if it is true.
23885 Update calls to aarch64_strip_extend.
23886 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
23888 2017-06-21 Andrew Pinski <apinski@cavium.com>
23890 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
23892 (thunderxt88): Likewise.
23893 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
23894 (thunderx_prefetch_tune): New variable.
23895 (thunderx2t99_prefetch_tune): Update for the correct values.
23896 (thunderxt88_tunings): New variable.
23897 (thunderx_tunings): Use thunderx_prefetch_tune instead of
23898 generic_prefetch_tune.
23899 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
23901 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23903 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
23904 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
23905 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
23906 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
23907 (aarch64_atomic_cas<mode>, GPI): Likewise.
23909 2017-06-21 Martin Liska <mliska@suse.cz>
23911 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
23912 statements on cold and hot labels.
23913 * predict.c (tree_estimate_probability_bb): Remove the
23914 prediction from this place.
23916 2017-06-21 Martin Liska <mliska@suse.cz>
23918 PR tree-optimization/79489
23919 * gimplify.c (maybe_add_early_return_predict_stmt): New
23921 (gimplify_return_expr): Call the function.
23922 * predict.c (tree_estimate_probability_bb): Remove handling
23924 * predict.def: Update comment about early return predictor.
23925 * gimple-predict.h (is_gimple_predict): New function.
23926 * predict.def: Change default value of early return to 66.
23927 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
23929 * passes.def: Put pass_strip_predict_hints to the beginning of
23932 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
23934 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
23935 FUNCTION_DECL declarations.
23936 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
23938 (dwaf2out_decl): Likewise.
23939 * godump.c (go_early_global_decl): Skip call to the real debug hook
23940 for FUNCTION_DECL declarations.
23941 * passes.c (rest_of_decl_compilation): Skip call to the
23942 early_global_decl debug hook for FUNCTION_DECL declarations, unless
23943 -fdump-go-spec is passed.
23945 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23947 * config/i386/i386.c (struct builtin_isa): New field pure_p.
23948 Reorder for compactness.
23949 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
23950 (def_builtin_pure, def_builtin_pure2): New functions.
23951 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
23953 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23955 * match.pd (nop_convert): New predicate.
23956 ((A +- CST1) +- CST2): Allow some NOP conversions.
23958 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23961 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
23962 with ctors/dtors if GOVD_SHARED is set.
23964 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23966 * config/aarch64/aarch64.md (movti_aarch64):
23967 Emit mov rather than orr.
23968 (movtf_aarch64): Likewise.
23969 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
23970 Emit mov rather than orr.
23972 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23974 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
23975 Swap alternatives, make integer dup more expensive.
23977 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23979 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
23980 Return true for non-tls symbols.
23982 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
23984 * config/aarch64/aarch64-cores.def (cortex-a55): New.
23985 (cortex-a75): Likewise.
23986 (cortex-a75.cortex-a55): Likewise.
23987 * config/aarch64/aarch64-tune.md: Regenerate.
23988 * doc/invoke.texi (-mtune): Document new values for -mtune.
23990 2017-06-21 Tom de Vries <tom@codesourcery.com>
23992 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
23993 stack_size feature.
23994 (Effective-Target Keywords, Other attributes): Suggest using
23995 dg-add-options stack_size feature to get stack limit in stack_size
23996 effective target documentation.
23998 2017-06-21 Julian Brown <julian@codesourcery.com>
23999 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24001 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
24002 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
24003 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
24005 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
24006 attribute type list for neon_multiply.
24007 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
24008 attribute type list for neon_multiply.
24009 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
24010 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
24011 attribute type list for neon_multiply.
24012 * config/arm/types.md (crypto_pmull): Add.
24013 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
24014 attribute type list.
24016 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
24018 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
24021 2017-06-20 Jakub Jelinek <jakub@redhat.com>
24023 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
24024 to make sure not to dereference a NULL cost_classes_ptr pointer.
24026 2017-06-20 Carl Love <cel@us.ibm.com>
24028 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24029 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
24030 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
24031 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
24032 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
24033 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
24034 VMULOSW): New enum "unspec" values.
24035 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
24036 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
24037 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
24038 altivec_vmulosw): New patterns.
24039 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
24040 VMULOSW): Add definitions.
24042 2017-06-20 Julia Koval <julia.koval@intel.com>
24044 * config/i386/i386.c: Fix rounding expand for new pattern.
24045 * config/i386/subst.md: Fix pattern (parallel -> unspec).
24047 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24049 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
24050 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
24052 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24054 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
24057 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24059 * config/aarch64/aarch64-cores.def: Rearrange to sort by
24060 architecture, then by implementer ID.
24061 * config/aarch64/aarch64-tune.md: Regenerate.
24063 2017-06-20 Richard Biener <rguenther@suse.de>
24065 PR middle-end/81097
24066 * fold-const.c (split_tree): Fold to type before negating.
24068 2017-06-20 David Malcolm <dmalcolm@redhat.com>
24070 * diagnostic-show-locus.c
24071 (selftest::test_fixit_deletion_affecting_newline): New function.
24072 (selftest::diagnostic_show_locus_c_tests): Call it.
24074 2017-06-20 Andreas Schwab <schwab@suse.de>
24077 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
24080 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24082 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
24083 __ARM_FEATURE_COPROC according to support.
24085 2017-06-20 Jakub Jelinek <jakub@redhat.com>
24087 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
24088 Rewritten to avoid overflow for > 32-bit pointers.
24091 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
24092 by removing enum keyword.
24093 (ubsan_type_descriptor): Likewise. Formatting fix.
24096 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
24097 splitter): Require TARGET_SSE2 in the condition.
24099 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
24102 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
24103 for doing vector set of SFmode on ISA 3.0.
24104 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
24105 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
24107 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
24108 SFmode value into a V4SF variable that was extracted from another
24109 V4SF variable without converting the element to double precision
24110 and back to single precision vector format.
24111 (vsx_insert_extract_v4sf_p9_2): Likewise.
24113 2017-06-19 Jakub Jelinek <jakub@redhat.com>
24115 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
24116 in UWHI to avoid undefined overflow.
24119 * ubsan.h (enum ubsan_encode_value_phase): New.
24120 (ubsan_encode_value): Change second argument to
24121 enum ubsan_encode_value_phase with default value of
24122 UBSAN_ENCODE_VALUE_GENERIC.
24123 * ubsan.c (ubsan_encode_value): Change second argument to
24124 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
24125 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
24126 create_tmp_var_raw instead of create_tmp_var and use a
24128 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
24129 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
24130 ubsan_encode_value callers.
24133 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
24134 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
24135 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
24137 2017-06-19 Richard Biener <rguenther@suse.de>
24139 PR middle-end/81118
24140 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
24141 estimates if we changed anything.
24143 2017-06-19 Richard Biener <rguenther@suse.de>
24145 PR tree-optimization/80887
24146 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
24147 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
24148 simplified lookups, then reset mprts_hook.
24149 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
24151 (try_to_simplify): Likewise.
24153 2017-06-19 Martin Liska <mliska@suse.cz>
24156 * gimplify.c (gimplify_switch_expr):
24157 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
24159 2017-06-19 Martin Liska <mliska@suse.cz>
24161 * doc/install.texi: Document that PGO runs in 4 stages.
24163 2017-06-19 Martin Liska <mliska@suse.cz>
24166 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
24167 to dispatcher function name.
24168 * multiple_target.c (replace_function_decl): New function.
24169 (create_dispatcher_calls): Redirect both edges and references.
24171 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
24173 * profile-count.c (profile_count::dump): Dump quality.
24174 (profile_count::differs_from_p): Update for unsigned val.
24175 * profile-count.h (profile_count_quality): New enum.
24176 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
24178 2017-06-19 Richard Biener <rguenther@suse.de>
24180 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
24181 struct function as arg.
24182 (estimate_numbers_of_iterations): Export overload with loop arg.
24183 (free_numbers_of_iterations_estimates_loop): Use an overload of
24184 free_numbers_of_iterations_estimates instead.
24185 * tree-cfg.c (remove_bb): Adjust.
24186 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
24187 * tree-parloops.c (gen_parallel_loop): Likewise.
24188 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
24190 (tree_unroll_loops_completely): Likewise.
24191 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
24192 Use an overload instead and export.
24193 (estimated_loop_iterations): Adjust.
24194 (max_loop_iterations): Likewise.
24195 (likely_max_loop_iterations): Likewise.
24196 (estimate_numbers_of_iterations): Take struct function as arg
24198 (loop_exits_before_overflow): Adjust.
24199 (free_numbers_of_iterations_estimates_loop): Use an overload.
24200 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
24201 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
24203 2017-06-19 Richard Biener <rguenther@suse.de>
24206 * ipa-prop.c (find_constructor_constant_at_offset): Handle
24207 RANGE_EXPR conservatively.
24209 2017-06-16 Carl Love <cel@us.ibm.com>
24211 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24212 definitions for vec_float, vec_float2, vec_floato,
24213 vec_floate built-ins.
24214 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
24215 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
24217 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
24218 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
24219 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
24220 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
24221 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
24222 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
24223 vec_floato): Add builtin defines.
24224 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
24225 Update the built-in documentation file for the new built-in
24228 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24230 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
24231 (mthumb): Mark as the negative of -marm.
24233 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24235 * doc/invoke.texi (ARM Options, -mcpu): Document supported
24237 (ARM Options, -mtune): Document that this accepts the same
24238 extension options as -mcpu.
24239 (ARM Options, -mfpu): Document addition of -mfpu=auto.
24241 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24243 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
24244 permitted extensions.
24246 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24248 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
24249 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
24250 (armv8-m.main): Add option +nodsp.
24251 * config/arm/arm-cpu-cdata.h: Regenerated.
24253 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24255 * config/arm/t-fuchsia: New file.
24256 * config.gcc (arm*-*-fuchsia*): Use it.
24258 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24260 * config/arm/t-symbian: Rewrite for new option infrastructure.
24262 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24264 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
24265 (MULTILIB_REQUIRED): Likewise.
24267 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24269 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
24270 (MULTILIB_RESUE): Likewise.
24271 (MULTILIB_MATCHES): Likewise.
24272 (MULTLIB_REQUIRED): Likewise.
24274 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24276 * config/arm/t-rtems: Rewrite for new option framework.
24278 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24280 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
24281 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
24282 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
24283 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
24284 * config/arm/t-multilib: ... here.
24285 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
24286 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
24287 armv7-a and armv8*-a when A-profile libraries have not been built.
24288 * config/arm/t-rmprofile: Rewrite.
24290 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24292 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
24293 with a backslash. Remove the backslash after substituting unescaped
24295 * doc/fragments.texi (MULTILIB_REUSE): Document it.
24297 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24299 * config.gcc: (arm*-*-*): When building a-profile libraries, force
24300 the driver to pass through the default setting of -mfloat-abi.
24301 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
24303 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
24304 (all_feat_combs): New rule.
24305 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
24307 * config/arm/t-aprofile: Rewrite.
24309 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24311 * config/arm/arm.h (FPUTYPE_AUTO): Define.
24312 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
24313 fpu is not specified by the user/command-line.
24314 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
24315 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
24316 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
24317 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
24318 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
24319 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
24321 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24323 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
24324 * config/arm/t-arm-elf: Rewritten.
24326 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24328 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
24329 have some floating-point instructions.
24330 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
24331 (TARGET_MAYBE_HARD_FLOAT): New macro.
24332 * config/arm/arm-builtins.c (arm_init_builtins): Use
24333 TARGET_MAYBE_HARD_FLOAT.
24334 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
24336 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24338 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
24339 (configargs.h): Include it.
24340 (arm_print_hint_for_fpu_option): New function.
24341 (arm_parse_fpu_option): New function.
24342 (candidate_extension): New class.
24343 (arm_canon_for_multilib): New function.
24344 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
24345 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
24346 (ARCH_CANONICAL_SPECS): New macro.
24347 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
24349 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24351 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
24352 are set after handling multilib fragments. Set target_cpu_default2
24355 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24357 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
24359 (arm*-*-*): Set target_cpu_default2 to a quoted string.
24360 * config/arm/parsecpu.awk (check_cpu): Validate any extension
24362 (check_arch): Likewise.
24363 * config/arm/arm.c (arm_configure_build_target): Handle
24364 TARGET_CPU_DEFAULT being a string constant. Scan any feature
24365 options in the default.
24367 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24369 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
24370 when an option is an alias of another.
24371 * config/arm/parsecpu.awk (optalias): New parser token.
24372 (gen_comm_data): Mark non-alias options as such. Emit entries
24373 for extension aliases.
24374 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
24375 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
24376 (armv6kz, armv6zk, armv6t2): Likewise.
24377 (armv7): Make vfpv3-d16 an alias.
24378 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
24380 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
24381 Sort in canonical order.
24382 (armv8-a): Sort in canonical order.
24383 (armv8.1-a, armv8.2-a): Likewise.
24384 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
24386 (cortex-a9): Sort in canonical order.
24387 * config/arm/arm.c (selftests.h): Include it.
24388 (arm_test_cpu_arch_data): New function.
24389 (arm_run_self_tests): New function.
24390 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
24391 (targetm): Move declaration to the end of the file.
24392 * arm-cpu-cdata.h: Regenerated.
24394 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24396 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
24397 call to target_mode_check describing the type of option passed.
24398 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
24399 (arm_target_thumb_only): Use arm_parse_arch_option_name or
24400 arm_parse_cpu_option_name to match parameters against list of
24402 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
24403 arm_arch_core_flags data structure.
24404 * config/arm/arm-cpu_cdata.h: Regenerated.
24406 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24408 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
24410 (arm_print_hint_for_cpu_option): Likewise.
24411 (arm_print_hint_for_arch_option): Likewise.
24412 (arm_parse_cpu_option_name): Likewise.
24413 (arm_parse_arch_option_name): Likewise.
24414 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
24415 of entries in the all_fpus list.
24416 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
24417 (arm_parse_cpu_option_name): Declare.
24418 (arm_parse_arch_option_name): Declare.
24419 (arm_parse_option_features): Declare.
24420 (arm_intialize_isa): Declare.
24421 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
24423 (gen_comm_data): ... here. Make definitions non-static.
24424 * config/arm/arm-cpu-data.h: Regenerated.
24425 * config/arm/arm-cpu-cdata.h: Regenerated.
24427 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24429 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
24430 (cpu_arch_extension): New structure.
24431 (cpu_arch_option, arch_option, cpu_option): New structures.
24432 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
24433 architecture types.
24434 (gen_data): Generate new format data tables.
24435 * config/arm/arm.c (cpu_tune): New structure.
24436 (cpu_option, processors): Delete.
24437 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
24438 (arm_print_hint_for_cpu_option): ... this and ...
24439 (arm_print_hint_for_arch_option): ... this.
24440 (arm_parse_arch_cpu_name): Delete. Replace with ...
24441 (arm_parse_cpu_option_name): ... this and ...
24442 (arm_parse_arch_option_name): ... this.
24443 (arm_unrecognized_feature): Change type of target parameter to
24445 (arm_parse_arch_cpu_features): Delete. Replace with ...
24446 (arm_parse_option_features): ... this.
24447 (arm_configure_build_target): Rework to use new configuration data
24449 (arm_print_tune_info): Rework for new configuration data tables.
24450 * config/arm/arm-cpu-data.h: Regenerated.
24451 * config/arm/arm-cpu.h: Regenerated.
24453 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24455 * Makefile.in (OBJS): Move sbitmap.o from here ...
24456 (OBJS-libcommon): ... to here.
24458 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24460 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
24461 (ISA_ALL_CRYPTO): New macro.
24462 (ISA_ALL_SIMD): New macro
24463 (ISA_ALL_FP): New macro.
24464 * config/arm/arm.c (fpu_bitlist): Update initializer.
24465 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
24467 (arm9e): Add fpu. Add option for nofp
24468 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
24469 (arm926ej-s, arm1026ej-s): Likewise.
24470 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
24471 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
24472 neon-fp16, neon-vfpv4, nofp and nosimd.
24473 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
24474 (cortex-a8): Add fpu. Add option for nofp.
24475 (cortex-a9): Add fpu. Add options for nosimd and nofp.
24476 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
24477 (cortex-r4f): Add fpu.
24478 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
24479 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
24481 (cortex-r8): Likewise.
24482 (cortex-m4): Add fpu. Add option for nofp.
24483 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
24484 (cortex-a17.cortex-a7): Likewise.
24485 (cortex-a32): Add fpu. Add options for crypto and nofp.
24486 (cortex-a35, cortex-a53): Likewise.
24487 (cortex-a57): Add fpu. Add option for crypto.
24488 (cortex-a72, cortex-a73): Likewise.
24489 (exynos-m1): Likewise.
24490 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
24491 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
24492 (cortex-m33): Add fpu. Add option for nofp.
24493 * config/arm/arm-cpu-cdata.h: Regenerated
24494 * config/arm/arm-cpu-data.h: Regenerated.
24496 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24498 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
24499 (armv5te, armv5tej): Likewise.
24500 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
24501 (armv7): Add options fp and vfpv3-d16.
24502 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
24503 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
24505 (armv7ve): Likewise.
24506 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
24507 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
24508 (armv8-a): Add nocrypto option.
24509 (armv8.1-a, armv8.2-a): Likewise.
24510 (armv8-m.main): add options fp, fp.dp and nofp.
24512 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24514 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
24516 (armv8-a+crc): Delete.
24517 (armv8.1-a): Add options simd, crypto and nofp.
24518 (armv8.2-a): Add options fp16, simd, crypto and nofp.
24519 (armv8.2-a+fp16): Delete.
24520 (armv8-m.main): Add option dsp.
24521 (armv8-m.main+dsp): Delete.
24522 (cortex-a8): Add fpu. Add nofp option.
24523 (cortex-a9): Add fpu. Add nofp and nosimd options.
24524 * config/arm/parsecpu.awk (gen_data): Generate option tables and
24525 link to main cpu and architecture data structures.
24526 (gen_comm_data): Only put isa attributes from the main architecture
24528 (option): New statement for architecture and CPU entries.
24529 * arm.c (struct cpu_option): New structure.
24530 (struct processors): Add entry for options.
24531 (arm_unrecognized_feature): New function.
24532 (arm_parse_arch_cpu_name): Ignore any characters after the first
24534 (arm_parse_arch_cpu_feature): New function.
24535 (arm_configure_build_target): Separate out any CPU and architecture
24536 features and parse separately. Don't error out if -mfpu=auto is
24537 used with only an architecture string.
24538 (arm_print_asm_arch_directives): New function.
24539 (arm_file_start): Call it.
24540 * config/arm/arm-cpu-cdata.h: Regenerated.
24541 * config/arm/arm-cpu-data.h: Likewise.
24542 * config/arm/arm-tables.opt: Likewise.
24544 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24546 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
24547 assembler when it is not -mfpu=auto.
24549 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24551 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
24552 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
24553 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
24554 (ASM_CPU_SPEC): Rewrite.
24555 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
24556 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
24557 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
24558 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
24559 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
24560 copied string is NUL-terminated. Also strip any characters prefixed
24562 (arm_rewrite_selected_arch): New function.
24563 (arm_rewrite_march): New function.
24565 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24567 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
24568 (x_arm_cpu_string, x_arm_tune_string): Likewise.
24569 (march, mcpu, mtune): Convert to string-based options.
24570 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
24571 (arm_parse_arch_cpu_name): New function.
24572 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
24573 identify selected architecture or CPU.
24574 (arm_option_save): New function.
24575 (TARGET_OPTION_SAVE): Redefine.
24576 (arm_option_restore): Restore string options.
24577 (arm_option_print): Print string options.
24579 2017-06-16 Martin Sebor <msebor@redhat.com>
24581 PR tree-optimization/80933
24582 PR tree-optimization/80934
24583 * builtins.c (fold_builtin_3): Do not handle bcmp here.
24584 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
24585 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
24586 (gimple_fold_builtin): Call them.
24588 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24590 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
24591 as unlikely; update profile.
24593 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24595 * predict.c (force_edge_cold): Handle declaring edges impossible
24598 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24600 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
24602 (try_unroll_loop_completely): Fix reporting.
24604 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24606 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
24608 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
24611 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
24612 if given a non-constant argument for an intrinsic which requires a
24615 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24617 * profile.c (compare_freqs): New function.
24618 (branch_prob): Sort edge list.
24619 (find_spanning_tree): Assume that the list is priority sorted.
24621 2017-06-16 Richard Biener <rguenther@suse.de>
24623 PR tree-optimization/81090
24624 * passes.def (pass_record_bounds): Remove.
24625 * tree-pass.h (make_pass_record_bounds): Likewise.
24626 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
24627 make_pass_record_bounds): Likewise.
24628 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
24629 not free niter estimates at the beginning but at the end.
24630 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
24632 2017-06-16 Richard Biener <rguenther@suse.de>
24634 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
24635 initializer to workaround ICE in host GCC 4.8.
24637 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24639 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
24641 (clone_inlined_nodes): Update.
24643 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24645 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
24646 prefetch settings, and enable prefetching by default at -O3.
24648 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24650 * config/aarch64/aarch64.c (aarch64_override_options_internal):
24651 Set flag_prefetch_loop_arrays according to tuning data.
24653 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24655 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
24656 New tune structure.
24657 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
24658 [Unrelated to main purpose of the patch] Place the pointer field last
24659 to enable type checking errors when tune structure are wrongly merged.
24660 * config/aarch64/aarch64.c (generic_prefetch_tune,)
24661 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
24662 (thunderx2t99_prefetch_tune): New tune constants.
24663 (tune_params *_tunings): Update all tunings (no functional change).
24664 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
24665 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
24666 from tunings structures.
24668 2017-06-16 Jakub Jelinek <jakub@redhat.com>
24671 * ubsan.c (instrument_null): Add T argument, use it instead
24672 of computing it based on IS_LHS.
24673 (instrument_object_size): Likewise.
24674 (pass_ubsan::execute): Adjust instrument_null and
24675 instrument_object_size callers to pass gimple_get_lhs or
24676 gimple_assign_rhs1 result to it. Use instrument_null instead of
24677 calling get_base_address and instrument_mem_ref. Handle
24678 aggregate call arguments for object-size sanitization.
24680 2017-06-16 Yury Gribov <tetra2005@gmail.com>
24682 PR tree-optimization/81089
24683 * tree-vrp.c (is_masked_range_test): Validate operands of
24686 2017-06-15 Martin Sebor <msebor@redhat.com>
24689 * dumpfile.c (dump_register): Avoid calling memset to initialize
24690 a class with a default ctor.
24691 * gcc.c (struct compiler): Remove const qualification.
24692 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
24693 * hash-table.h: Ditto.
24694 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
24696 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
24697 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
24699 * params.h (struct param_info): Make struct members non-const.
24700 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
24701 with default initialization.
24702 * vec.h (vec_copy_construct, vec_default_construct): New helper
24704 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
24705 with vec_copy_construct.
24706 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
24707 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
24708 * doc/invoke.texi (-Wclass-memaccess): Document.
24710 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24712 * emit-rtl.h (is_leaf): Update comment about local
24713 register allocator.
24715 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
24718 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
24719 for a variable to have a section before checking if the section has a
24721 Set section to.persistent if persistent attribute is set.
24722 Warn if .persistent attribute is used on an automatic variable.
24724 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
24726 PR rtl-optimization/80474
24727 * reorg.c (update_block): Do not ignore instructions in a delay slot.
24729 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
24731 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
24734 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
24736 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
24737 (casesi): Emit bounds checking as RTL.
24738 (casesi_internal_mips16_<mode>): Remove bounds checking.
24740 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
24742 * config/xtensa/xtensa.c (xtensa_option_override): Append
24743 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
24744 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
24745 xtensa_doloop_hooks): Define unconditionally.
24746 (xtensa_reorg_loops): Only call reorg_loops in the presence of
24748 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
24749 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
24750 for it in xtensa_option_override.
24751 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
24752 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
24754 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
24756 * doc/cppopts.texi: Document '-' special value to -MF.
24758 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
24760 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
24761 (cortex_a53_fconst): Likewise.
24762 (cortex_a53_fpmul): Likewise.
24763 (cortex_a53_f_load_64): Likewise.
24764 (cortex_a53_f_load_many): Likewise.
24765 (cortex_a53_advsimd_alu): Likewise.
24766 (cortex_a53_advsimd_alu_q): Likewise.
24767 (cortex_a53_advsimd_mul): Likewise.
24768 (cortex_a53_advsimd_mul_q): Likewise.
24769 (fpmac bypass): Add new bypass for fpmac-fpmac case.
24770 Add missing fmul, r2f_cvt and fconst cases.
24772 2017-06-14 Richard Biener <rguenther@suse.de>
24774 PR middle-end/81088
24775 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
24777 (fold_binary_loc): When associating do not treat pre-existing
24778 TREE_OVERFLOW on literal constants as a reason to allow
24779 TREE_OVERFLOW on associated literal constants.
24781 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24783 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
24784 (MASK_FEATURES): New macro.
24785 * config/sparc/sparc.c (sparc_option_override): Remove the special
24786 handling of -mfpu and generalize it to all MASK_FEATURES switches.
24788 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24790 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
24791 a division of 0 if non-call exceptions are enabled.
24793 2017-06-14 Andrew Pinski <apinski@cavium.com>
24794 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24797 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
24798 Improve vector initialization code gen for only variable case.
24800 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24802 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
24804 2017-06-14 Richard Biener <rguenther@suse.de>
24806 PR tree-optimization/81083
24807 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
24810 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24812 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
24813 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
24814 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
24815 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
24816 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
24817 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
24819 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24821 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
24822 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
24824 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24826 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
24828 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24830 * config/rs6000/t-rtems: Don't handle SPE.
24832 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24834 * config/rs6000/t-linux: Don't handle SPE.
24836 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24838 * config/rs6000/eabispe.h: Delete file.
24840 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24842 * config/rs6000/t-spe: Delete file.
24844 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24846 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
24847 (rs6000_legitimate_offset_address_p): Return false for anything in
24848 V2SImode or V2SFmode.
24850 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24852 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
24853 except V2SF and V2SI. Rearrange the vector modes, and add comments.
24854 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
24856 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
24857 (rs6000_legitimate_offset_address_p): Ditto.
24858 (rs6000_emit_move): Ditto.
24859 (rs6000_init_builtins): Remove V4HI_type_node.
24861 2017-06-13 Martin Liska <mliska@suse.cz>
24864 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
24865 (gate_asan): Likewise.
24866 * asan.h (asan_no_sanitize_address_p): Remove the function.
24867 (sanitize_flags_p): New function.
24868 * builtins.def: Fix coding style.
24869 * common.opt: Use renamed enum value.
24870 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
24871 * doc/extend.texi: Document no_sanitize attribute.
24872 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
24873 to SANITIZE_UNDEFINED_NONDEFAULT.
24874 * gcc.c (sanitize_spec_function): Use the renamed enum value.
24875 * gimple-fold.c (optimize_atomic_compare_exchange_p):
24876 Use sanitize_flags_p.
24877 * gimplify.c (gimplify_function_tree): Likewise.
24878 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
24879 * opts.c (parse_no_sanitize_attribute): New function.
24880 (common_handle_option): Use renamed enum value.
24881 * opts.h (parse_no_sanitize_attribute): Declare.
24882 * tree.c (sanitize_flags_p): New function.
24883 * tree.h: Declared here.
24884 * tsan.c: Use sanitize_flags_p.
24885 * ubsan.c (ubsan_expand_null_ifn): Likewise.
24886 (instrument_mem_ref): Likewise.
24887 (instrument_bool_enum_load): Likewise.
24888 (do_ubsan_in_current_function): Remove the function.
24889 (pass_ubsan::execute): Use sanitize_flags_p.
24890 * ubsan.h: Remove do_ubsan_in_current_function
24891 * tree-cfg.c (print_no_sanitize_attr_value): New function.
24892 (dump_function_to_file): Use it here.
24894 2017-06-13 Martin Jambor <mjambor@suse.cz>
24896 PR tree-optimization/80803
24897 PR tree-optimization/81063
24898 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
24899 (propagate_subaccesses_across_link): Enqueue subtree whenever
24900 necessary instead of relying on the caller.
24902 2017-06-13 Martin Jambor <mjambor@suse.cz>
24904 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
24905 that have a first_link.
24906 (sort_and_splice_var_accesses): Do not check first_link before
24908 (subtree_mark_written_and_enqueue): Likewise.
24909 (propagate_all_subaccesses): Likewise and do not stop at first
24910 parent with a first_link.
24912 2017-06-13 Martin Jambor <mjambor@suse.cz>
24914 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
24917 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24919 * match.pd: New pattern.
24921 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24923 * tree-vrp.c (is_masked_range_test): New function.
24924 (register_edge_assert_for): Determine ranges for
24927 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24929 PR tree-optimization/67328
24930 * fold-const.c (maskable_range_p): New function.
24931 (build_range_check): Generate bittests if possible.
24933 2017-06-13 Martin Liska <mliska@suse.cz>
24935 * gimple-pretty-print.c (dump_probability): Add new argument.
24936 (dump_edge_probability): Dump both probability and count.
24937 (dump_gimple_label): Likewise.
24938 (dump_gimple_bb_header): Likewise.
24940 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
24943 * config/avr/avr-devices.c: Fix indentation.
24944 * config/avr/gen-avr-mmcu-specs.c: Dito.
24946 2017-06-13 Richard Biener <rguenther@suse.de>
24948 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
24949 instead get vector type from stmt_info.
24950 (vectorizable_reduction): Adjust. Remove dead code.
24952 2017-06-13 Richard Biener <rguenther@suse.de>
24954 PR middle-end/81065
24955 * fold-const.c (extract_muldiv_1): Remove bogus distribution
24956 case of C * (x * C2 + C3).
24957 (fold_addr_of_array_ref_difference): Properly fold index difference.
24959 2017-06-12 David S. Miller <davem@davemloft.net>
24962 * config/sparc/sparc.md (return expander): Emit frame blockage if
24963 function uses alloca.
24965 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
24967 * combine.c (make_field_assignment): Check len rather than the mode
24968 precision when calling force_to_mode.
24970 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
24972 Support multilibs and devices that see flash in RAM address range.
24975 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
24976 (avr_mcu_t) <flash_pm_offset>: New field.
24977 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
24978 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
24979 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
24980 (AVR_TINY_PM_OFFSET): Remove macro.
24981 * config/avr/avr.opt (-mshort-calls): New option.
24982 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24983 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
24984 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
24985 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
24986 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
24987 instead of avr_arch->have_jmp_call.
24988 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
24989 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
24990 avr_arch->flash_pm_offset to define.
24991 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
24992 new field flash_pm_offset. Add entry for avrxmega3.
24993 (avr_texinfo): Add entry for avrxmega3.
24994 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
24995 attiny212, attiny214,
24996 attiny412, attiny414, attiny416, attiny417,
24997 attiny814, attiny816, attiny817,
24998 attiny1614, attiny1616, attiny1617,
24999 attiny3214, attiny3216, attiny3217.
25000 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
25001 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
25002 (avr_print_operand_address) [AVR_TINY]: Same.
25003 (avr_asm_init_sections) <readonly_data_section>: Only patch
25004 callback if avr_arch->flash_pm_offset = 0.
25005 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
25006 for rodata if avr_arch->flash_pm_offset != 0.
25007 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
25008 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
25009 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
25010 (m_options): Append opt_rcall.
25011 (m_dirnames): Append dir_rcall.
25012 * config/avr/t-multilib: Regenerate.
25014 * configure.ac [target=avr]: Check whether avrxmega3 default
25015 linker description file works as needed.
25016 * configure: Regenerate.
25017 * doc/avr-mmcu.texi: Regenerate.
25018 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
25019 <__AVR_ARCH__>: Document avrxmega3 and 103.
25020 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
25021 <__AVR_SHORT_CALLS__>: Document it.
25022 <__AVR_PM_BASE_ADDRESS__>: Document it.
25023 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
25024 (AVR Variable Attributes) <progmem>: Document this is
25025 not needed for avrxmega3.
25026 (AVR Named Address Spaces) <__flash>: Dito.
25028 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
25030 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
25032 2017-06-12 Doug Rupp <rupp@adacore.com>
25034 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
25035 Append vxworks-stdint.h to the tm_file list.
25036 * config/vxworks-stdint.h: New file.
25038 2017-06-12 Martin Liska <mliska@suse.cz>
25040 PR tree-optimization/81041
25041 * tree-profile.c (gimple_gen_ic_func_profiler):
25042 Create an extra BB in profile-generate
25043 (gimple_gen_time_profiler): Likewise.
25045 2017-06-12 Jakub Jelinek <jakub@redhat.com>
25047 PR tree-optimization/81003
25048 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
25049 (update_range_test): Use it instead of force_gimple_operand_gsi.
25051 2017-06-12 Richard Biener <rguenther@suse.de>
25053 PR tree-optimization/81053
25054 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
25055 with backedge value not defined in loop. Simplify def stmt
25058 2017-06-11 Tom de Vries <tom@codesourcery.com>
25061 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
25063 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
25064 nvptx_cannot_force_const_mem.
25066 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25068 * opts.c (finish_options): Move test for flag_split_stack after
25069 it has been initialized.
25071 2017-06-11 Jason Merrill <jason@redhat.com>
25073 * tree.h (id_equal): New.
25074 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
25075 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
25076 instead of strcmp of IDENTIFIER_POINTER.
25078 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25080 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
25081 (mark_all_inlined_calls_cdtor): Fix formating.
25082 (inline_transform): Rescale profile before inlining.
25084 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25086 * cgraph.h (cgraph_edge::clone): Update prototype.
25087 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
25088 (cgraph_node::create_clone): Update.
25089 (cgraph_node::create_version_clone): Update.
25090 * tree-inline.c (copy_bb): Update.
25091 (expand_call_inline): Update.
25093 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25095 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
25096 factored out from ...
25097 (rs6000_emit_prologue): ... here.
25099 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25101 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
25102 factored out from ...
25103 (rs6000_emit_prologue): ... here.
25105 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25107 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
25109 (handle_missing_profiles): Fix computation of tp_first_run.
25110 (counts_to_freqs): Do not touch freqs when count is 0.
25112 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25114 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
25117 2017-06-10 Tom de Vries <tom@codesourcery.com>
25119 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
25120 attributes): Document signal effective target.
25122 2017-06-10 Tom de Vries <tom@codesourcery.com>
25124 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
25125 Document effective target stack_size.
25127 2017-06-09 David Malcolm <dmalcolm@redhat.com>
25129 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
25130 to the edit_context if they can be auto-applied.
25132 2017-06-9 Ian Lance Taylor <iant@golang.org>
25134 * opts.c (finish_options): If -fsplit-stack, disable implicit
25135 -forder-blocks-and-partition.
25136 * doc/invoke.texi (Optimize Options): Document that when using
25137 -fsplit-stack -forder-blocks-and-partition is not implicitly
25140 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25142 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
25143 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
25144 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
25145 * builtins.def (abort, trap, unreachable): Declare cold.
25146 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
25147 * tree-core.h (ECF_COLD): New.
25148 * tree.c (set_call_expr_flags): Handle ECF_COLD.
25149 (build_common_builtin_nodes): Mark unreachable and abort as cold.
25151 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25153 * predict.c (unlikely_executed_stmt_p): Cleanup.
25155 2017-06-09 Richard Biener <rguenther@suse.de>
25157 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
25158 model if the ref is always written to.
25160 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25162 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
25164 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25166 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
25169 2017-06-09 Tom de Vries <tom@codesourcery.com>
25172 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
25173 "target cannot support label values" when encountering LABEL_REF.
25175 2017-06-09 Martin Liska <mliska@suse.cz>
25177 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
25178 (gimple_gen_ic_func_profiler): Emit direct comparison
25179 of __gcov_indirect_call_callee with NULL.
25180 (gimple_gen_time_profiler): Change probability from
25181 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
25183 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25185 * profile.c (edge_gcov_counts): Turn to pointer.
25186 (compute_branch_probabilities, compute_branch_probabilities): Update.
25187 (branch_prob): Do not clear edge_gcov_count.
25188 * profile.h (edge_gcov_counts): Turn to pointer.
25189 (edge_gcov_count): Update.
25191 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25193 * gimple.h (gimple_check_failed): Mark cold.
25195 2017-06-09 Richard Biener <rguenther@suse.de>
25197 PR tree-optimization/66623
25198 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
25199 refactor check_reduction into two parts, properly computing
25200 whether we have to check reduction validity for outer loop
25203 2017-06-09 Richard Biener <rguenther@suse.de>
25205 PR tree-optimization/79483
25206 * graphite-scop-detection.c (order): New global.
25207 (get_order): Compute bb to order mapping that satisfies code
25208 generation constraints.
25209 (cmp_pbbs): New helper.
25210 (build_scops): Start domwalk at entry block, sort generated
25213 2017-06-09 Richard Biener <rguenther@suse.de>
25215 PR middle-end/81007
25216 * ipa-polymorphic-call.c
25217 (ipa_polymorphic_call_context::restrict_to_inner_class):
25218 Skip FIELD_DECLs with error_mark_node type.
25219 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
25222 2017-06-09 Martin Liska <mliska@suse.cz>
25224 * predict.c (struct branch_predictor): New struct.
25225 (test_prediction_value_range): New test.
25226 (predict_c_tests): New function.
25227 * selftest-run-tests.c (selftest::run_tests): Run the function.
25228 * selftest.h: Declare new tests.
25230 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
25233 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
25234 gen_add3_insn did not fail.
25235 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
25236 r0, construct that number in a temporary reg and add that reg to r0.
25237 If asked to put the result in r0 as well, fail.
25239 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
25241 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25242 for early expansion of vec_eqv.
25244 2017-06-08 Jakub Jelinek <jakub@redhat.com>
25246 PR middle-end/81005
25247 * ubsan.c (instrument_null): Avoid pointless code temporary.
25248 (pass_ubsan::execute): Instrument aggregate arguments of calls.
25250 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
25254 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
25257 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
25260 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25262 * predict.c (unlikely_executed_edge_p): Move ahead.
25263 (probably_never_executed_edge_p): Use it.
25265 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25267 PR middle-end/79988
25268 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
25269 gimple_call_builtin_p call.
25271 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25273 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
25274 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
25275 rtl_check_failed_type2, rtl_check_failed_code1,
25276 rtl_check_failed_code2, rtl_check_failed_code_mode,
25277 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
25278 rtvec_check_failed_bounds, rtl_check_failed_flag,
25279 _fatal_insn_not_found, _fatal_insn): Likewise.
25280 * tree.h (tree_contains_struct_check_failed,
25281 tree_check_failed, tree_not_check_failed,
25282 tree_class_check_failed, tree_range_check_failed,
25283 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
25284 tree_vec_elt_check_failed, phi_node_elt_check_failed,
25285 tree_operand_check_failed, omp_clause_check_failed,
25286 omp_clause_operand_check_failed, omp_clause_range_check_failed):
25289 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25291 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
25292 flag_branch_probabilities.
25293 * ipa-inline.c (edge_badness): Likewise.
25294 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
25295 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
25296 * predict.c (maybe_hot_frequency_p): Likewise.
25297 (probably_never_executed): Likewise.
25298 * sched-ebb.c (schedule_ebbs): Likewise.
25299 * sched-rgn.c (find_single_block_region): Likewise.
25300 * tracer.c (tail_duplicate): Likewise.
25302 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25304 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
25305 longer requires x_flag_profile_use.
25307 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25309 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
25310 instead of flag_reorder_blocks_and_partition.
25311 * dbxout.c (dbxout_function_end): Likewise.
25312 * dwarf2out.c (gen_subprogram_die): Likewise.
25313 * haifa-sched.c (sched_create_recovery_edges): Likewise.
25314 * hw-doloop.c (reorg_loops): Likewise.
25315 * varasm.c (assemble_start_function,
25316 assemble_end_function): Likewise.
25317 (decide_function_section): Do not check for
25318 flag_reorder_blocks_and_partition.
25320 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25322 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
25324 (chkp_get_hard_register_fake_addr_expr): Ditto.
25325 (chkp_build_addr_expr): Add check for hard reg case.
25326 (chkp_parse_array_and_component_ref): Ditto.
25327 (chkp_find_bounds_1): Ditto.
25328 (chkp_process_stmt): Don't generate bounds store for
25331 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25333 * predict.c (maybe_hot_bb_p): Do not check profile status.
25334 (maybe_hot_edge_p): Likewise.
25335 (probably_never_executed): Check for zero counts even if profile
25337 (unlikely_executed_edge_p): New function.
25338 (unlikely_executed_stmt_p): New function.
25339 (unlikely_executed_bb_p): New function.
25340 (set_even_probabilities): Use unlikely predicates.
25341 (combine_predictions_for_bb): Likewise.
25342 (predict_paths_for_bb): Likewise.
25343 (predict_paths_leading_to_edge): Likewise.
25344 (determine_unlikely_bbs): New function.
25345 (estimate_bb_frequencies): Use it.
25346 (compute_function_frequency): Use zero counts even if profile is
25348 * profile-count.h: Fix typo.
25350 2017-08-08 Julia Koval <julia.koval@intel.com>
25352 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
25353 _mm512_mask_cvtsepi16_storeu_epi8,
25354 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
25355 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
25356 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
25357 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
25358 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
25359 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
25360 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
25361 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
25362 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
25363 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
25364 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
25365 __builtin_ia32_pmovuswb256mem_mask,
25366 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
25367 __builtin_ia32_pmovwb512mem_mask): New builtins.
25369 2017-08-08 Julia Koval <julia.koval@intel.com>
25371 PR target/73350,80862
25372 * config/i386/subst.md (round): Fix round pattern.
25373 * config/i386/i386.c (ix86_erase_embedded_rounding):
25374 Fix erasing rounding for the fixed pattern.
25376 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25378 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
25380 2017-06-08 Martin Liska <mliska@suse.cz>
25382 PR gcov-profile/80911
25383 * gcov.c (block_info::block_info): New constructor.
25385 2017-06-07 Carl Love <cel@us.ibm.com>
25387 * config/rs6000/rs6000-c: The return type of the following
25388 built-in functions was implemented as int not long long. Fix sign
25389 of return value for the unsigned version of vec_mulo and vec_mule.
25390 vector unsigned long long vec_bperm (vector unsigned long long,
25391 vector unsigned char)
25392 vector signed long long vec_mule (vector signed int,
25394 vector unsigned long long vec_mule (vector unsigned int,
25395 vector unsigned int)
25396 vector signed long long vec_mulo (vector signed int,
25398 vector unsigned long long vec_mulo (vector unsigned int,
25399 vector unsigned int)
25400 * doc/extend.texi: Fix the documentation for the built-in
25403 2017-06-07 Carl Love <cel@us.ibm.com>
25406 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
25409 2017-06-07 Carl Love <cel@us.ibm.com>
25411 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
25412 support, Generate doublehv for signed int/float for BE case only.
25414 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
25416 * doc/invoke.texi (mcx16): Rewrite.
25418 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25420 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
25421 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
25422 *mov<mode>_softfloat, and an anonymous splitter): Use
25423 nonimmediate_operand instead of rs6000_nonimmediate_operand.
25425 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25427 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
25429 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25430 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
25431 (rs6000_debug_reg_global): Adjust.
25432 (rs6000_init_hard_regno_mode_ok): Adjust.
25433 (rs6000_dbx_register_number): Adjust.
25434 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
25435 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25436 Remove SPE_ACC and SPEFSCR.
25437 (REG_ALLOC_ORDER): Ditto.
25438 (FRAME_POINTER_REGNUM): Change to 111.
25439 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
25440 (REG_CLASS_NAMES): Ditto.
25441 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
25442 (REGISTER_NAMES): Ditto.
25443 (ADDITIONAL_REG_NAMES): Ditto.
25444 (rs6000_reg_names): Ditto.
25445 * config/rs6000/rs6000.md: Renumber some register number
25448 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25450 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
25452 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25453 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
25455 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
25456 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25457 Delete the SPE high registers.
25458 (REG_ALLOC_ORDER): Ditto.
25459 (enum reg_class): Remove SPE_HIGH_REGS.
25460 (REG_CLASS_NAMES): Ditto.
25461 (REG_CLASS_CONTENTS): Delete the SPE high registers.
25462 (REGISTER_NAMES): Ditto.
25463 (rs6000_reg_names): Ditto.
25464 * doc/tm.texi.in: Remove SPE as example.
25465 * doc/tm.texi: Regenerate.
25467 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25469 * config/rs6000/8540.md (ppc8540_brinc): Delete.
25470 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
25471 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
25472 * config/rs6000/rs6000.md (type): Remove "brinc".
25474 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25476 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
25477 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
25478 * config/rs6000/linuxspe.h: Delete file.
25479 * config/rs6000/rs6000.md: Don't include spe.md.
25480 * config/rs6000/spe.h: Delete file.
25481 * config/rs6000/spe.md: Delete file.
25482 * config/rs6000/t-rs6000: Remove spe.md.
25484 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25486 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
25487 (reg_or_none500mem_operand): Delete.
25488 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
25489 instead of reg_or_none500mem_operand.
25491 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25493 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
25494 handling of SPE flags.
25495 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
25497 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25499 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
25501 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
25502 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
25503 paired_divv2sf3): Similar.
25504 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
25505 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
25506 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
25508 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
25509 Rename the paired_* instruction patterns.
25510 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
25512 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
25513 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
25514 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
25515 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
25516 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
25517 PAIRED_VECTOR_MODE.
25518 (struct machine_function): Delete field spe_insn_chain_scanned_p.
25519 (spe_func_has_64bit_regs_p): Delete.
25520 (spe_expand_predicate_builtin): Delete.
25521 (spe_expand_evsel_builtin): Delete.
25522 (TARGET_DWARF_REGISTER_SPAN): Do not define.
25523 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
25524 (invalid_e500_subreg): Delete.
25525 (rs6000_legitimize_address): Always force_reg op2 as well, for
25526 paired single memory accesses.
25527 (rs6000_member_type_forces_blk): Delete.
25528 (rs6000_spe_function_arg): Delete.
25529 (rs6000_expand_unop_builtin): Delete SPE handling.
25530 (rs6000_expand_binop_builtin): Ditto.
25531 (spe_expand_stv_builtin): Delete.
25532 (bdesc_2arg_spe): Delete.
25533 (spe_expand_builtin): Delete.
25534 (spe_expand_predicate_builtin): Delete.
25535 (spe_expand_evsel_builtin): Delete.
25536 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
25537 (spe_init_builtins): Delete.
25538 (spe_func_has_64bit_regs_p): Delete.
25539 (savres_routine_name): Delete "info" parameter. Adjust callers.
25540 (rs6000_emit_stack_reset): Ditto.
25541 (rs6000_dwarf_register_span): Delete.
25542 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
25543 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
25544 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
25546 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
25548 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
25549 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
25550 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
25551 mulv2sf3, divv2sf3): Delete expanders.
25553 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25555 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
25557 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25559 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
25560 * config/rs6000/rs6000.c: Ditto.
25562 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25564 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
25565 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
25566 comparison_operator.
25568 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25570 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
25571 * config/rs6000/rs6000.opt: Ditto.
25572 * config/rs6000/t-rtems: Ditto.
25574 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25576 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
25577 TARGET_E500_SINGLE by 0, simplify.
25578 * config/rs6000/rs6000.c: Ditto.
25579 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
25580 (spe_build_register_parallel): Delete.
25581 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
25582 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
25583 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
25584 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
25585 (E500_CONVERT): Delete.
25586 * config/rs6000/spe.md: Remove many patterns and all define_constants.
25588 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25590 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
25591 * config/rs6000/dfp.md: Ditto.
25592 (negdd2, *negdd2_fpr): Merge.
25593 (absdd2, *absdd2_fpr): Merge.
25594 (negtd2, *negtd2_fpr): Merge.
25595 (abstd2, *abstd2_fpr): Merge.
25596 * config/rs6000/e500.h: Delete file.
25597 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
25598 TARGET_FPRS by 1 and simplify.
25599 * config/rs6000/rs6000-c.c: Ditto.
25600 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
25601 TARGET_DF_SPE by 0.
25602 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
25604 * config/rs6000/rs6000.md: Ditto.
25605 (floatdidf2, *floatdidf2_fpr): Merge.
25606 (move_from_CR_gt_bit): Delete.
25607 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
25608 (E500_CR_IOR_COMPARE): Delete.
25609 (All patterns that require !TARGET_FPRS): Delete.
25610 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
25612 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25614 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
25616 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25618 * graphds.c (add_edge): Intitialize edge's attached data.
25619 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
25620 pointer parameter. Call pointed function on each edge during
25621 graph traversing. Skip traversing the edge when the function
25623 (graphds_dfs, graphds_scc): Ditto.
25624 (for_each_edge): New parameter. Pass the new parameter to callback
25626 * graphds.h (skip_edge_callback): New function pointer type.
25627 (graphds_dfs, graphds_scc): New function pointer parameter.
25628 (graphds_edge_callback, for_each_edge): New parameter.
25630 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25632 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
25633 out code checking if runtime alias check is possible to below ...
25634 Call the new function.
25635 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
25636 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
25638 2017-06-07 Marek Polacek <polacek@redhat.com>
25641 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
25642 TYPE_OVERFLOW_WRAPS check.
25644 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25646 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
25647 if versioning is required.
25648 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
25649 peeling with the check for versioning.
25651 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25653 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
25654 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
25655 Set true to new parameter if new ssa variable is defined.
25656 (vect_gen_vector_loop_niters): Refactor. Set range information
25657 for the new vector loop bound variable.
25658 (vect_do_peeling): Ditto.
25660 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25662 * tree-affine.c (ssa.h): Include header file.
25663 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
25664 has wrapping overflow behavior.
25666 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25668 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
25670 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25672 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
25673 (tree_to_aff_combination): ... here.
25675 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25677 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
25678 reg_pressure model function.
25679 (ivopts_global_cost_for_size): Delete.
25680 (determine_set_costs, iv_ca_recount_cost): Call new model function
25681 ivopts_estimate_reg_pressure.
25683 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25685 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
25686 expensive than udiv. Remove floating point cases from mod.
25688 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25690 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
25691 Increase idiv cost.
25693 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25695 * config/aarch64/aarch64.md
25696 (copysignsf3): Fix mask generation.
25698 2017-06-07 Jakub Jelinek <jakub@redhat.com>
25700 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
25702 (class dump_manager): Add register_dumps method.
25703 * dumpfile.c: Include langhooks.h.
25704 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
25705 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
25706 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
25707 (dump_manager::dump_register): Start with 512 entries instead of 32.
25708 (dump_manager::register_dumps): New method.
25709 * toplev.c (general_init): Instead of invoking register_dumps
25710 langhook, invoke register_dumps method on the dump manager.
25711 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
25714 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
25716 * doc/md.texi: Clarify the restrictions on a define_insn condition.
25717 Say that # requires an associated define_split to exist, and that
25718 the define_split must be suitable for use after register allocation.
25720 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25722 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
25723 (compute_outgoing_frequencies): Also initialize zero counts.
25724 (find_many_sub_basic_blocks): Do not produce uninitialized profile
25725 around loops; preserve more of profile when nothing changes.
25727 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
25729 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
25731 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
25732 * config/arm/arm-cpu-cdata.h: Regenerate.
25733 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
25734 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
25735 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
25736 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
25737 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
25739 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
25741 * config/arm/t-rmprofile: Likewise.
25742 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
25744 2017-06-06 David S. Miller <davem@davemloft.net>
25747 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
25748 blockage if function uses alloca.
25750 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25752 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
25753 New "uid" fields to hold pretty-print IDs of group and ref.
25754 Memory references are now identified as <group_id>:<ref_id>
25755 instead of using [random] addresses.
25756 (dump_mem_details): Simplify, no functional change.
25757 (dump_mem_ref): Simplify and make output more concise.
25758 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
25759 (find_or_create_group): Initialize group uid.
25760 (record_ref): Initialize ref uid. Improve debug output.
25761 (prune_group_by_reuse, should_issue_prefetch_p,)
25762 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
25763 (mark_nontemporal_store, determine_loop_nest_reuse):
25764 Improve debug output.
25766 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25768 * dbgcnt.def (prefetch): New debug counter.
25769 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
25770 (schedule_prefetches): Stop issueing prefetches if debug counter
25773 2017-06-06 Tom de Vries <tom@codesourcery.com>
25775 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
25776 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
25778 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25780 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
25781 Use aarch64_reg_or_zero predicate for operand 4.
25782 (aarch64_compare_and_swap<mode> define_insn_and_split):
25783 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
25784 (aarch64_store_exclusive<mode>): Likewise for operand 2.
25786 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25788 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
25789 (arm_compute_save_core_reg_mask): This.
25790 (thumb1_compute_save_reg_mask): Rename into ...
25791 (thumb1_compute_save_core_reg_mask): This.
25792 (arm_compute_save_reg0_reg12_mask): Adapt comment.
25793 (arm_compute_frame_layout): Likewise.
25795 2017-06-06 Richard Biener <rguenther@suse.de>
25797 PR tree-optimization/80974
25798 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
25799 keep or clear leaders SSA info.
25801 2017-06-06 Tom de Vries <tom@codesourcery.com>
25803 * config/nvptx/nvptx.c (split_mode_p): New function.
25804 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
25806 2017-06-06 Tom de Vries <tom@codesourcery.com>
25808 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
25810 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25813 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
25816 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25818 * shrink-wrap.c (handle_simple_exit): Update profile.
25819 (try_shrink_wrapping): Upate profile.
25821 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25823 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
25824 (tree_guess_outgoing_edge_probabilities): New.
25825 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
25826 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
25828 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25830 * ipa-split.c (split_function): Initialize return bb profile.
25832 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25834 * profile.c (compute_branch_probabilities): Also initialize
25835 EXIT_BLOCK profile.
25837 2017-06-06 Richard Biener <rguenther@suse.de>
25839 PR tree-optimization/80928
25840 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
25841 (vect_analyze_loop_operations): Properly guard analysis for
25843 (vect_transform_loop): Likewise.
25844 (vect_analyze_loop_2): Also reset SLP type on PHIs.
25845 (vect_model_induction_cost): Do not cost for pure SLP.
25846 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
25847 of induction in inner loop vectorization.
25848 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
25849 (vect_get_and_check_slp_defs): Handle vect_induction_def.
25850 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
25852 (vect_analyze_slp_cost_1): Cost induction.
25853 (vect_detect_hybrid_slp_stmts): Handle PHIs.
25854 (vect_get_slp_vect_defs): Likewise.
25855 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
25856 (vect_transform_stmt): Handle SLP reductions.
25857 * tree-vectorizer.h (vectorizable_induction): Adjust.
25859 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25861 * config/rs6000/rs6000.c (make_resolver_func): Update
25862 init_lowered_empty_function call.
25864 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
25866 * doc/invoke.texi: Document the -fprofile-abs-path option.
25867 * common.opt (fprofile-abs-path): New option.
25868 * gcov-io.h (gcov_write_filename): Declare.
25869 * gcov-io.c (gcov_write_filename): New function.
25870 * coverage.c (coverage_begin_function): Use gcov_write_filename.
25871 * profile.c (output_location): Likewise.
25873 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25875 * shring-wrap.c: Revert accidental commit.
25877 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
25879 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
25881 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25883 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
25885 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
25886 profile in callgraph edge.
25887 * profile-count.h (apply_probability): If THIS is 0, then result is 0
25888 (apply_scale): Likewise.
25889 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
25890 Also scale profile when inlining function with zero profile.
25891 (initialize_cfun): Update exit block profile even when it is zero.
25892 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
25893 when profile is read.
25895 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25897 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
25898 (CLONE_*): New constants to define the processors we can generate
25899 code for with the target_clone attribute.
25900 (rs6000_clone_map): New array to identify which clone processors
25901 the current program is running on.
25902 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
25903 target_clone attribute.
25904 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
25905 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
25906 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
25907 (cpu_expand_builtin): Add support for target_clone attribute.
25908 (rs6000_valid_attribute_p): Allow "default" attribute.
25909 (get_decl_name): New debug function to simplify printing the
25910 current function name in debugging statements.
25911 (rs6000_clone_priority): New functions to support the target_clone
25912 attribute, and be able to generate code to switch between ISA 2.05
25913 through ISA 3.0 (power6 through power9).
25914 (rs6000_compare_version_priority): Likewise.
25915 (rs6000_get_function_versions_dispatcher): Likewise.
25916 (make_resolver_func): Likewise.
25917 (add_condition_to_bb): Likewise.
25918 (dispatch_function_versions): Likewise.
25919 (rs6000_generate_version_dispatcher_body): Likewise.
25920 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
25921 (fusion_gpr_load_p): Fix a spacing issue.
25922 * doc/extend.texi (Common Function Attributes): Document that the
25923 PowerPC supports the target_clone attribute.
25925 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25927 * config/arm/arm.h: explain F symbol found in description of ARM
25928 register allocation in its legend.
25930 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25932 * config/mips/frame-header-opt.c: Include profile-count.h.
25933 * config/riscv/riscv.c: Include profile-count.h
25935 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25937 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
25939 (sm_set_flag_if_changed): Add bbs field.
25940 (execute_sm_if_changed_flag_set): Pass BBS.
25941 (execute_sm): Update.
25943 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25945 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
25948 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25950 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
25951 (peephole2): New peephole2 to emit the above.
25952 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
25954 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25956 * config/aarch64/aarch64.c (define_peephole2 above
25957 *sub_<shift>_<mode>): New peephole.
25959 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25961 * config/i386/i386.c (make_resolver_func): Update.
25962 * Makefile.in: Add profile-count.h and profile-count.o
25963 * auto-profile.c (afdo_indirect_call): Update to new API.
25964 (afdo_set_bb_count): Update.
25965 (afdo_propagate_edge): Update.
25966 (afdo_propagate_circuit): Update.
25967 (afdo_calculate_branch_prob): Update.
25968 (afdo_annotate_cfg): Update.
25969 * basic-block.h: Include profile-count.h
25970 (struct edge_def): Turn count to profile_count.
25971 (struct basic_block_def): Likewie.
25972 (REG_BR_PROB_BASE): Move to profile-count.h
25973 (RDIV): Move to profile-count.h
25974 * bb-reorder.c (max_entry_count): Turn to profile_count.
25975 (find_traces): Update.
25976 (rotate_loop):Update.
25977 (connect_traces):Update.
25978 (sanitize_hot_paths):Update.
25979 * bt-load.c (migrate_btr_defs): Update.
25980 * cfg.c (RDIV): Remove.
25981 (init_flow): Use alloc_block.
25982 (alloc_block): Uninitialize count.
25983 (unchecked_make_edge): Uninitialize count.
25984 (check_bb_profile): Update.
25985 (dump_edge_info): Update.
25986 (dump_bb_info): Update.
25987 (update_bb_profile_for_threading): Update.
25988 (scale_bbs_frequencies_int): Update.
25989 (scale_bbs_frequencies_gcov_type): Update.
25990 (scale_bbs_frequencies_profile_count): New.
25991 * cfg.h (update_bb_profile_for_threading): Update.
25992 (scale_bbs_frequencies_profile_count): Declare.
25993 * cfgbuild.c (compute_outgoing_frequencies): Update.
25994 (find_many_sub_basic_blocks): Update.
25995 * cfgcleanup.c (try_forward_edges): Update.
25996 (try_crossjump_to_edge): Update.
25997 * cfgexpand.c (expand_gimple_tailcall): Update.
25998 (construct_exit_block): Update.
25999 * cfghooks.c (verify_flow_info): Update.
26000 (dump_bb_for_graph): Update.
26001 (split_edge): Update.
26002 (make_forwarder_block): Update.
26003 (duplicate_block): Update.
26004 (account_profile_record): Update.
26005 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
26006 (get_estimated_loop_iterations): Update.
26007 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
26008 (single_likely_exit): Update.
26009 * cfgloopmanip.c (scale_loop_profile): Update.
26011 (set_zero_probability): Update.
26012 (lv_adjust_loop_entry_edge): Update.
26013 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
26014 (purge_dead_edges): Update.
26015 (rtl_account_profile_record): Update.
26016 * cgraph.c (cgraph_node::create): Uninitialize count.
26017 (symbol_table::create_edge): Uninitialize count.
26018 (cgraph_update_edges_for_call_stmt_node): Update.
26019 (cgraph_edge::dump_edge_flags): Update.
26020 (cgraph_node::dump): Update.
26021 (cgraph_edge::maybe_hot_p): Update.
26022 * cgraph.h: Include profile-count.h
26023 (create_clone), create_edge, create_indirect_edge): Update.
26024 (cgraph_node): Turn count to profile_count.
26025 (cgraph_edge0: Likewise.
26026 (make_speculative, clone): Update.
26027 (create_edge): Update.
26028 (init_lowered_empty_function): Update.
26029 * cgraphclones.c (cgraph_edge::clone): Update.
26030 (duplicate_thunk_for_node): Update.
26031 (cgraph_node::create_clone): Update.
26032 * cgraphunit.c (cgraph_node::analyze): Update.
26033 (cgraph_node::expand_thunk): Update.
26034 * final.c (dump_basic_block_info): Update.
26035 * gimple-streamer-in.c (input_bb): Update.
26036 * gimple-streamer-out.c (output_bb): Update.
26037 * graphite.c (print_global_statistics): Update.
26038 (print_graphite_scop_statistics): Update.
26039 * hsa-brig.c: Include basic-block.h.
26040 * hsa-dump.c: Include basic-block.h.
26041 * hsa-gen.c (T sum_slice): Update.
26042 (convert_switch_statements):Update.
26043 * hsa-regalloc.c: Include basic-block.h.
26044 * ipa-chkp.c (chkp_produce_thunks): Update.
26045 * ipa-cp.c (struct caller_statistics): Update.
26046 (init_caller_stats): Update.
26047 (gather_caller_stats): Update.
26048 (ipcp_cloning_candidate_p): Update.
26049 (good_cloning_opportunity_p): Update.
26050 (get_info_about_necessary_edges): Update.
26051 (dump_profile_updates): Update.
26052 (update_profiling_info): Update.
26053 (update_specialized_profile): Update.
26054 (perhaps_add_new_callers): Update.
26055 (decide_about_value): Update.
26056 (ipa_cp_c_finalize): Update.
26057 * ipa-devirt.c (struct odr_type_warn_count): Update.
26058 (struct decl_warn_count): Update.
26059 (struct final_warning_record): Update.
26060 (possible_polymorphic_call_targets): Update.
26061 (ipa_devirt): Update.
26062 * ipa-fnsummary.c (redirect_to_unreachable): Update.
26063 * ipa-icf.c (sem_function::merge): Update.
26064 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
26065 * ipa-inline.c (compute_uninlined_call_time): Update.
26066 (compute_inlined_call_time): Update.
26067 (want_inline_small_function_p): Update.
26068 (want_inline_self_recursive_call_p): Update.
26069 (edge_badness): Update.
26070 (lookup_recursive_calls): Update.
26071 (recursive_inlining): Update.
26072 (inline_small_functions): Update.
26073 (dump_overall_stats): Update.
26074 (dump_inline_stats): Update.
26075 * ipa-profile.c (ipa_profile_generate_summary): Update.
26076 (ipa_propagate_frequency): Update.
26077 (ipa_profile): Update.
26078 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26079 * ipa-utils.c (ipa_merge_profiles): Update.
26080 * loop-doloop.c (doloop_modify): Update.
26081 * loop-unroll.c (report_unroll): Update.
26082 (unroll_loop_runtime_iterations): Update.
26083 * lto-cgraph.c (lto_output_edge): Update.
26084 (lto_output_node): Update.
26085 (input_node): Update.
26086 (input_edge): Update.
26087 (merge_profile_summaries): Update.
26088 * lto-streamer-in.c (input_cfg): Update.
26089 * lto-streamer-out.c (output_cfg): Update.
26090 * mcf.c (create_fixup_graph): Update.
26091 (adjust_cfg_counts): Update.
26092 (sum_edge_counts): Update.
26093 * modulo-sched.c (sms_schedule): Update.
26094 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
26095 * predict.c (maybe_hot_count_p): Update.
26096 (probably_never_executed): Update.
26097 (dump_prediction): Update.
26098 (combine_predictions_for_bb): Update.
26099 (propagate_freq): Update.
26100 (handle_missing_profiles): Update.
26101 (counts_to_freqs): Update.
26102 (rebuild_frequencies): Update.
26103 (force_edge_cold): Update.
26104 * predict.h: Include profile-count.h
26105 (maybe_hot_count_p, counts_to_freqs): UPdate.
26106 * print-rtl-function.c: Do not include cfg.h
26107 * print-rtl.c: Include basic-block.h
26108 * profile-count.c: New file.
26109 * profile-count.h: New file.
26110 * profile.c (is_edge_inconsistent): Update.
26111 (correct_negative_edge_counts): Update.
26112 (is_inconsistent): Update.
26113 (set_bb_counts): Update.
26114 (read_profile_edge_counts): Update.
26115 (compute_frequency_overlap): Update.
26116 (compute_branch_probabilities): Update; Initialize and deinitialize
26118 (branch_prob): Update.
26119 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
26120 (edge_gcov_count): New.
26121 (bb_gcov_count): New.
26122 * shrink-wrap.c (try_shrink_wrapping): Update.
26123 * tracer.c (better_p): Update.
26124 * trans-mem.c (expand_transaction): Update.
26125 (ipa_tm_insert_irr_call): Update.
26126 (ipa_tm_insert_gettmclone_call): Update.
26127 * tree-call-cdce.c: Update.
26128 * tree-cfg.c (gimple_duplicate_sese_region): Update.
26129 (gimple_duplicate_sese_tail): Update.
26130 (gimple_account_profile_record): Update.
26131 (execute_fixup_cfg): Update.
26132 * tree-inline.c (copy_bb): Update.
26133 (copy_edges_for_bb): Update.
26134 (initialize_cfun): Update.
26135 (freqs_to_counts): Update.
26136 (copy_cfg_body): Update.
26137 (expand_call_inline): Update.
26138 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26139 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
26140 (try_unroll_loop_completely): Update.
26141 (try_peel_loop): Update.
26142 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26143 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
26144 * tree-ssa-loop-split.c (connect_loops): Update.
26145 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
26146 * tree-ssa-reassoc.c (branch_fixup): Update.
26147 * tree-ssa-tail-merge.c (replace_block_by): Update.
26148 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
26149 (compute_path_counts): Update.
26150 (update_profile): Update.
26151 (recompute_probabilities): Update.
26152 (update_joiner_offpath_counts): Update.
26153 (estimated_freqs_path): Update.
26154 (freqs_to_counts_path): Update.
26155 (clear_counts_path): Update.
26156 (ssa_fix_duplicate_block_edges): Update.
26157 (duplicate_thread_path): Update.
26158 * tree-switch-conversion.c (case_bit_test_cmp): Update.
26159 (struct switch_conv_info): Update.
26160 * tree-tailcall.c (decrease_profile): Update.
26161 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26162 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26163 * value-prof.c (check_counter): Update.
26164 (gimple_divmod_fixed_value): Update.
26165 (gimple_mod_pow2): Update.
26166 (gimple_mod_subtract): Update.
26167 (gimple_ic_transform): Update.
26168 (gimple_stringop_fixed_value): Update.
26169 * value-prof.h (gimple_ic): Update.
26171 2017-06-02 Carl Love <cel@us.ibm.com>
26173 * config/rs6000/rs6000-c: Add support for built-in functions
26174 vector double vec_doublee (vector signed int);
26175 vector double vec_doublee (vector unsigned int);
26176 vector double vec_doublee (vector float);
26177 vector double vec_doubleh (vector signed int);
26178 vector double vec_doubleh (vector unsigned int);
26179 vector double vec_doubleh (vector float);
26180 vector double vec_doublel (vector signed int);
26181 vector double vec_doublel (vector unsigned int);
26182 vector double vec_doublel (vector float);
26183 vector double vec_doubleo (vector signed int);
26184 vector double vec_doubleo (vector unsigned int);
26185 vector double vec_doubleo (vector float);.
26186 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
26187 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
26189 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
26190 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
26191 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
26193 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
26194 vec_doublel, vec_doubleh.
26195 * doc/extend.texi: Update the built-in documentation file for the
26196 new built-in functions.
26198 2017-06-02 David Malcolm <dmalcolm@redhat.com>
26201 * ipa-inline-analysis.c (free_growth_caches): Set
26202 edge_removal_hook_holder to NULL after removing it.
26204 2017-06-02 Sudakshina Das <sudi.das@arm.com>
26206 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
26207 comparision with zero.
26209 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26210 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26211 for early expansion of vec_min and vec_max builtins.
26212 (builtin_function_type): Add min/max unsigned variants to those
26213 identified as having unsigned arguments.
26215 2017-06-02 Olivier Hainque <hainque@adacore.com>
26217 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
26219 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26221 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
26222 Use VALL_F16 iterator rather than VALL.
26224 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26226 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
26227 Emit CBNZ inside loop when doing a strong exchange and comparing
26228 against zero. Generate the CC flags after the loop.
26230 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
26232 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
26233 (dl_section_ref): New.
26234 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
26235 On AIX, append an expression to subtract the size of the
26236 section length to dl_section_ref.
26238 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26240 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26241 for early expansion of vector absolute builtins.
26243 2017-06-02 Richard Biener <rguenther@suse.de>
26245 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
26246 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
26248 2017-06-02 Richard Biener <rguenther@suse.de>
26250 PR tree-optimization/80948
26251 * tree-tailcall.c (find_tail_calls): Track stmts to move in
26252 stmt order as well.
26254 2017-06-02 Richard Biener <rguenther@suse.de>
26256 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
26258 * tree-vect-stmts.c (process_use): Do not mark backedge defs
26259 for inductions as relevant.
26261 2017-06-02 Richard Biener <rguenther@suse.de>
26263 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
26264 (vectorizable_induction): ... this. Remove dead code.
26266 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
26268 * builtins. (expand_builtin_alloca): Remove second parameter and
26269 infer its value from the first parameter instead.
26270 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
26272 2017-06-02 Jakub Jelinek <jakub@redhat.com>
26274 PR rtl-optimization/80903
26275 * loop-doloop.c (add_test): Unshare sequence.
26277 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
26279 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
26281 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
26283 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
26285 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
26286 xlogue_layout::get_instance, logue_layout::xlogue_layout,
26287 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
26288 (xlogue_layout::get_stub_rtx): Make static.
26289 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
26290 (xlogue_layout::compute_stub_managed_regs): Rename to...
26291 (xlogue_layout::count_stub_managed_regs): ...this.
26292 (xlogue_layout::is_stub_managed_reg): New function.
26293 (xlogue_layout::m_stub_names): Rename to...
26294 (xlogue_layout::s_stub_names): ...this, make static.
26295 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
26296 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
26297 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
26298 xlogue_layout::s_stub_names): Instantiate statics.
26299 (stub_managed_regs): Remove.
26300 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
26301 (disable_call_ms2sysv_xlogues): Rename to...
26302 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
26303 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
26305 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
26306 change after reload_completed.
26307 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
26309 (ix86_expand_prologue): Likewise.
26310 (ix86_expand_epilogue): Likewise.
26311 (ix86_expand_split_stack_prologue): Likewise.
26312 (ix86_compute_frame_layout): Remove frame parameter ...
26313 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
26314 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
26316 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
26317 (ix86_frame): Move from here ...
26318 * config/i386/i386.h (ix86_frame): ... to here.
26319 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
26320 complete ix86_frame data structure instead. Remove some_ld_name.
26322 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26324 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
26325 symbols that hold a DECL_VALUE_EXPR.
26327 2017-06-01 Martin Jambor <mjambor@suse.cz>
26329 PR tree-optimization/80898
26330 * tree-sra.c (process_subtree_disqualification): Removed.
26331 (disqualify_candidate): Do not acll
26332 process_subtree_disqualification.
26333 (subtree_mark_written_and_enqueue): New function.
26334 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
26335 RHS has been disqualified and re-queue LHS if necessary. Apart
26336 from that, ignore disqualified RHS.
26338 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26340 * config/s390/s390.c (s390_emit_epilogue): Disable early return
26341 address fetch for z10 or later.
26343 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26345 * config/arc/arc.md (tst_movb): Add guard when splitting.
26347 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26349 * config/arc/arc.c (arc_can_eliminate): Test against
26350 arc_frame_pointer_needed.
26352 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26354 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
26355 to prevent store reordering.
26356 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
26357 (type): Add block type.
26358 (stack_tie): Define special instruction to be used in
26361 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26363 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
26364 constraint. It is not valid for the pattern.
26365 (noncommutative_binary_comparison): Likewise.
26367 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26369 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
26372 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26374 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
26375 be used by the reg-alloc.
26377 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26379 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
26380 reg-alloc when having mul64 or mul32x16 instructions.
26381 (mulsidi3): Likewise.
26382 (umulsidi3): Likewise.
26383 (mulsi32x16): New pattern.
26384 (mulsi64): Likewise.
26385 (mulsidi64): Likewise.
26386 (umulsidi64): Likewise.
26387 (MUL32x16_REG): Define.
26388 (mul64_600): Use MUL32x16_REG.
26389 (mac64_600): Likewise.
26390 (umul64_600): Likewise.
26391 (umac64_600): Likewise.
26393 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26395 * config/arc/arc.md (mulsi3_700): Make it commutative.
26397 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
26399 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
26401 (*sign_extendsidi2_insn): Likewise for movstosw.
26403 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26405 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
26406 the type of the input discriminant value. Convert the
26407 discriminant value of signedness vary.
26409 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
26411 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
26412 Add to -Wall section.
26414 2017-06-01 Richard Biener <rguenther@suse.de>
26416 PR middle-end/66313
26417 * fold-const.c (fold_plusminus_mult_expr): If the factored
26418 factor may be zero use a wrapping type for the inner operation.
26419 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
26420 and handle moved defs.
26421 (process_assignment): Properly guard the unary op case. Return a
26422 tri-state indicating that moving the stmt before the call may allow
26423 to continue. Pass through to_move.
26424 (find_tail_calls): Handle moving unrelated defs before
26427 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
26430 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
26431 splitter result in the canonical way.
26433 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
26435 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
26436 also for 32bit target. Update insn attributes.
26437 (zero-extendsidi2 splitter): Allow all registers for operand 1.
26439 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
26441 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
26442 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
26443 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
26444 (_mm_maskz_min_ss): New intrinsics.
26446 2017-05-31 Martin Liska <mliska@suse.cz>
26448 * tree-vect-loop.c (vect_create_epilog_for_reduction):
26449 Change comment style to one we normally use.
26450 (vectorizable_reduction): Likewise.
26451 (vectorizable_induction): Likewise.
26452 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
26453 (vectorizable_call): Likewise.
26454 (vectorizable_simd_clone_call): Likewise.
26455 (vectorizable_conversion): Likewise.
26456 (vectorizable_assignment): Likewise.
26457 (vectorizable_shift): Likewise.
26458 (vectorizable_operation): Likewise.
26459 (vectorizable_store): Likewise.
26460 (vectorizable_load): Likewise.
26461 * tree-vectorizer.h: Likewise.
26463 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
26465 * passes.c (emergency_dump_function): New.
26466 * tree-pass.h (emergency_dump_function): Declare.
26467 * plugin.c (plugins_internal_error_function): Remove.
26468 * plugin.h (plugins_internal_error_function): Remove declaration.
26469 * toplev.c (internal_error_function): New static function. Use it...
26470 (general_init): ...here.
26472 2017-05-31 Graham Markall <graham.markall@embecosm.com>
26474 * config/arc/arc.c (arc_print_operand): Handle constant operands.
26475 (arc_rtx_costs): Add costs for new patterns.
26476 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
26477 * config/arc/predicates.md: Add _1_2_3_operand predicate.
26479 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26481 * tree-ssa-strlen.c (get_next_strinfo): New function.
26482 (get_stridx_plus_constant): Use it.
26483 (zero_length_string): Likewise.
26484 (adjust_related_strinfos): Likewise.
26485 (adjust_last_stmt): Likewise.
26487 2017-05-31 Richard Biener <rguenther@suse.de>
26490 * config/i386/i386.c (ix86_expand_builtin): Remove assert
26491 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
26493 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26495 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
26496 loop_vinfo argument and use of dependence distance vectors.
26497 Check instead whether the two references differ only in their
26498 initial value and assume that they have the same alignment if the
26499 difference is a multiple of the vector alignment.
26500 (vect_analyze_data_refs_alignment): Update call accordingly.
26502 2017-05-31 Martin Liska <mliska@suse.cz>
26505 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
26507 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26509 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
26510 (create_intersect_range_checks): Move from ...
26511 * tree-data-ref.c (create_intersect_range_checks_index)
26512 (create_intersect_range_checks): ... to here.
26513 (create_runtime_alias_checks): New function factored from ...
26514 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
26515 here. Call above function.
26516 * tree-data-ref.h (create_runtime_alias_checks): New function.
26518 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26520 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
26521 segment length for dr_b and compute it in wide_int.
26523 2017-05-31 Richard Biener <rguenther@suse.de>
26525 PR tree-optimization/80906
26526 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
26527 and pass through iv_map.
26528 (copy_bb_and_scalar_dependences): Adjust.
26529 (translate_pending_phi_nodes): Likewise.
26530 (copy_loop_close_phi_args): Handle code-generating IVs instead
26533 2017-05-30 David Malcolm <dmalcolm@redhat.com>
26535 * diagnostic-color.c (color_dict): Add "type-diff".
26536 (parse_gcc_colors): Update comment.
26537 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
26538 -fdiagnostics-show-template-tree and -fno-elide-type.
26539 (GCC_COLORS): Add type-diff to example.
26541 (-fdiagnostics-show-template-tree): New.
26542 (-fno-elide-type): New.
26543 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
26544 the pp_format_decoder callback. Call any m_format_postprocessor's
26546 (pretty_printer::pretty_printer): Initialize
26547 m_format_postprocessor.
26548 (pretty_printer::~pretty_printer): Delete any
26549 m_format_postprocessor.
26550 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
26551 (class format_postprocessor): New class.
26552 (struct pretty_printer::format_decoder): Document the new parameters.
26553 (struct pretty_printer::m_format_postprocessor): New field.
26554 * tree-diagnostic.c (default_tree_printer): Update for new
26555 bool and const char ** params.
26556 * tree-diagnostic.h (default_tree_printer): Likewise.
26558 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
26560 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
26561 (lwa_operand): Delete rs6000_gen_cell_microcode test.
26562 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26563 rs6000_gen_cell_microcode code.
26564 (rs6000_final_prescan_insn): Delete.
26565 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
26566 "warn-cell-microcode" entries.
26567 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
26568 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
26569 throughout. Change cc_reg_not_micro_cr0_operand to
26570 cc_reg_not_cr0_operand throughout.
26571 (*extendhi<mode>2_noload): Delete.
26572 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
26573 (mwarn-cell-microcode): Delete.
26574 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
26575 -mgen-cell-microcode and -mwarn-cell-microcode.
26577 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
26580 * config/i386/constraints.md (Yd): New constraint.
26582 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
26583 and (?Yd, r) alternatives. Update insn attributes.
26584 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
26585 and (?*Yd, r) alternatives. Update insn attributes.
26586 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
26588 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
26590 * gimplify.c (gimplify_modify_expr): Don't create a
26591 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
26594 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
26596 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
26598 2017-05-30 Richard Biener <rguenther@suse.de>
26600 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
26601 and reduc_def fields.
26602 (STMT_VINFO_REDUC_TYPE): New define.
26603 (STMT_VINFO_REDUC_DEF): Likewise.
26604 (vect_force_simple_reduction): Adjust prototype.
26605 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
26606 (vect_is_simple_reduction): Remove check_reduction argument.
26607 (vect_force_simple_reduction): Adjust and set
26608 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
26609 (vectorizable_reduction): Do not re-do reduction analysis
26610 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
26611 * tree-parloops.c (gather_scalar_reductions): Adjust.
26613 2017-05-30 Richard Biener <rguenther@suse.de>
26615 PR middle-end/80901
26616 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
26619 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
26621 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
26622 Introduce unknown_misalignment parameter and remove vf.
26623 (vect_peeling_hash_get_lowest_cost):
26624 Pass unknown_misalignment parameter.
26625 (vect_enhance_data_refs_alignment):
26626 Fix unsupportable data ref treatment.
26628 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26630 * tree-vect-data-refs.c (vect_get_data_access_cost):
26631 Workaround for SLP handling.
26632 (vect_enhance_data_refs_alignment):
26633 Compute costs for doing no peeling at all, compare to the best
26634 peeling costs so far and avoid peeling if cheaper.
26636 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26638 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
26639 Return peeling info and set costs to zero for unlimited cost
26641 (vect_enhance_data_refs_alignment): Also inspect all datarefs
26642 with unknown misalignment. Compute and costs for unknown
26643 misalignment, compare them to the costs for known misalignment
26644 and choose the cheapest for peeling.
26646 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26648 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
26649 (vect_get_peeling_costs_all_drs): Create function.
26650 (vect_peeling_hash_get_lowest_cost):
26651 Use vect_get_peeling_costs_all_drs.
26652 (vect_peeling_supportable): Create function.
26653 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
26655 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26657 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
26658 DR_HAS_NEGATIVE_STEP.
26659 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
26660 (vect_enhance_data_refs_alignment): Use.
26661 (vect_duplicate_ssa_name_ptr_info): Use.
26662 * tree-vectorizer.h (dr_misalignment): Use.
26663 (known_alignment_for_access_p): Use.
26665 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
26668 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
26670 (has_section_name): New function.
26672 2017-05-30 Martin Liska <mliska@suse.cz>
26675 * auto-profile.c (get_function_decl_from_block): Fix
26678 2017-05-30 Richard Biener <rguenther@suse.de>
26680 PR middle-end/80876
26681 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
26683 2017-05-30 Martin Liska <mliska@suse.cz>
26685 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
26686 * dumpfile.h (struct dump_file_info): Remove ctors.
26688 2017-05-30 Martin Liska <mliska@suse.cz>
26690 * predict.def: Fix GNU coding style.
26692 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
26694 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
26695 Mark 'to' argument with ATTRIBUTE_UNUSED.
26697 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
26699 * config/xtensa/xtensa.c (xtensa_emit_call): Use
26700 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
26701 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
26704 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
26706 * doc/install.texi (Options specification): Restore entry of
26707 --enable-sjlj-exceptions.
26709 2017-05-27 Michael Eager <eager@eagercon.com>
26712 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
26714 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
26716 * config/microblaze/microblaze.h
26717 (FIXED_REGISTERS): Update in macro.
26718 (CALL_USED_REGISTERS): Update in macro.
26720 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
26722 * doc/install.texi: Add links to macOS binary distributions.
26724 2017-05-27 Jakub Jelinek <jakub@redhat.com>
26728 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26730 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
26732 2017-05-26 Martin Liska <mliska@suse.cz>
26734 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
26736 2017-05-26 Martin Liska <mliska@suse.cz>
26738 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
26739 always leading ';; '.
26740 (dump_bb_info): Likewise.
26741 (brief_dump_cfg): Likewise.
26742 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
26743 * dumpfile.c: Remove usage of TDF_VERBOSE.
26744 * dumpfile.h (enum dump_kind): Likewise.
26745 (dump_gimple_bb_header): Do not use TDF_COMMENT.
26746 * print-tree.c (debug_verbose): Remove.
26747 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
26748 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
26749 * tree-diagnostic.c (default_tree_printer): Replace
26750 TDF_DIAGNOSTIC with TDF_SLIM.
26752 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26754 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
26755 in parameter loop, rather than loop_vinfo.
26756 (create_intersect_range_checks): Ditto.
26757 (vect_create_cond_for_alias_checks): Update call to above functions.
26759 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26761 PR tree-optimization/80815
26762 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
26763 for merging runtime alias checks. Handle negative DR_STEPs.
26765 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26767 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
26769 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
26770 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
26771 out code pruning runtime alias checks.
26772 * tree-data-ref.c (prune_runtime_alias_test_list): New function
26773 factored out from above.
26774 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
26776 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
26778 (prune_runtime_alias_test_list): New decalaration.
26780 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26782 * tree-vect-data-refs.c (compare_tree): Rename and move ...
26783 * tree-data-ref.c (data_ref_compare_tree): ... to here.
26784 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
26785 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
26786 (operator==, comp_dr_with_seg_len_pair): Ditto.
26787 (vect_prune_runtime_alias_test_list): Ditto.
26789 2017-05-26 Martin Liska <mliska@suse.cz>
26792 * params.def: Bound partial-inlining-entry-probability param.
26794 2017-05-26 Marek Polacek <polacek@redhat.com>
26797 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
26800 2017-05-26 Richard Biener <rguenther@suse.de>
26802 PR tree-optimization/80842
26803 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
26806 2017-05-26 Richard Biener <rguenther@suse.de>
26808 PR tree-optimization/80844
26809 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
26811 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
26813 * doc/md.texi (Machine Constraints): Update x86 family
26814 machine constraints section to match 'config/i386/constraints.md'.
26816 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
26818 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
26820 2017-05-25 Nathan Sidwell <nathan@acm.org>
26822 * doc/invoke.texi (--enable-languages): Update documentation.
26824 2017-05-25 Martin Liska <mliska@suse.cz>
26826 * dumpfile.c: Add TDF_FOLDING.
26827 * dumpfile.h (enum dump_kind): Likewise.
26828 * genmatch.c (dt_simplify::gen_1): Use it.
26830 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26832 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
26834 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26836 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
26837 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
26839 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26841 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
26842 * match.pd (X == C): Rewrite it here.
26843 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
26844 with_certain_nonzero_bits2): New predicates.
26845 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
26847 2017-05-24 Nathan Sidwell <nathan@acm.org>
26849 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
26852 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
26855 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26857 * config/powerpcspe: New port. Files are copied from the rs6000
26858 port, with "rs6000" in filenames replaced by "powerpcspe".
26860 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
26862 PR rtl-optimization/80754
26863 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
26865 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
26867 * config/sparc/sparc.md (length): Return the correct value for -mflat
26868 sibcalls to match output_sibcall.
26870 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26874 * config/rs6000/rs6000.c (struct machine_function): Add new field
26876 (rs6000_get_separate_components): Init that field, use it.
26877 (rs6000_components_for_bb): Use the field.
26879 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
26881 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
26883 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
26885 PR middle-end/80823
26886 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
26888 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26891 * config/s390/s390.c (s390_check_qrst_address): Check incoming
26892 address against address_operand predicate.
26893 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
26895 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
26897 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
26898 parameters passed indirectly.
26900 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
26902 * config/i386/i386.md (*movdi_internal): Remove SSE4
26903 alternative 18 (?r, *v). Update insn attributes.
26904 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
26905 Update insn attributes.
26906 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
26907 Update insn attributes.
26908 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
26909 alternative 1 (r, v). Remove isa attribute.
26910 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
26911 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
26912 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
26914 2017-05-23 Tom de Vries <tom@codesourcery.com>
26916 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
26919 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26921 * cgraphunit.c (symbol_table::process_new_functions): Update.
26922 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
26923 (inline_generate_summary): Rename to ...
26924 (ipa_fn_summary_generate): ... this one.
26925 (inline_read_summary): Rename to ...
26926 (ipa_fn_summary_read): ... this one.
26927 (inline_write_summary): Rename to ...
26928 (ipa_fn_summary_write): ... this one.
26929 (inline_free_summary): Rename to ...
26930 (ipa_free_fn_summary): ... this one.
26931 (pass_data_local_fn_summary, pass_local_fn_summary,
26932 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
26933 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
26934 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
26935 make_pass_ipa_fn_summary): New.
26936 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
26937 inline_write_summary, inline_free_summary): Remove.
26938 (ipa_free_fn_summary) : New.
26939 * ipa-inline.c (ipa_inline): Update.
26940 (pass_ipa_inline): Do not generate summaries.
26941 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
26943 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
26944 and add pass_ipa_fn_summary.
26945 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
26947 (make_pass_inline_parameters): Remove.
26949 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
26951 * omp-low.c (struct omp_context): Remove "default_kind" member.
26954 * omp-offload.c (execute_oacc_device_lower): Remove the
26955 parallelism dimensions function attributes for unparallelized
26956 OpenACC kernels constructs.
26958 2017-05-23 Martin Liska <mliska@suse.cz>
26960 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
26962 (cgraph_edge::make_speculative): Likewise.
26963 (cgraph_edge::resolve_speculation): Likewise.
26964 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
26965 (cgraph_node::dump): Likewise.
26966 * cgraph.h: Likewise.
26967 * cgraphunit.c (analyze_functions): Likewise.
26968 (symbol_table::compile): Likewise.
26969 * ipa-cp.c (print_all_lattices): Likewise.
26970 (determine_versionability): Likewise.
26971 (initialize_node_lattices): Likewise.
26972 (ipcp_verify_propagated_values): Likewise.
26973 (estimate_local_effects): Likewise.
26974 (update_profiling_info): Likewise.
26975 (create_specialized_node): Likewise.
26976 (perhaps_add_new_callers): Likewise.
26977 (decide_about_value): Likewise.
26978 (decide_whether_version_node): Likewise.
26979 (identify_dead_nodes): Likewise.
26980 (ipcp_store_bits_results): Likewise.
26981 * ipa-devirt.c (dump_targets): Likewise.
26982 (ipa_devirt): Likewise.
26983 * ipa-icf.c (sem_item::dump): Likewise.
26984 (sem_function::equals): Likewise.
26985 (sem_variable::equals): Likewise.
26986 (sem_item_optimizer::read_section): Likewise.
26987 (sem_item_optimizer::execute): Likewise.
26988 (congruence_class::dump): Likewise.
26989 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
26990 (dump_inline_summary): Likewise.
26991 (estimate_node_size_and_time): Likewise.
26992 (inline_analyze_function): Likewise.
26993 * ipa-inline-transform.c (inline_call): Likewise.
26994 * ipa-inline.c (report_inline_failed_reason): Likewise.
26995 (want_early_inline_function_p): Likewise.
26996 (edge_badness): Likewise.
26997 (update_edge_key): Likewise.
26998 (inline_small_functions): Likewise.
26999 * ipa-profile.c (ipa_profile): Likewise.
27000 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
27001 (ipa_make_edge_direct_to_target): Likewise.
27002 (remove_described_reference): Likewise.
27003 (ipa_impossible_devirt_target): Likewise.
27004 (propagate_controlled_uses): Likewise.
27005 (ipa_print_node_params): Likewise.
27006 (ipcp_transform_function): Likewise.
27007 * ipa-pure-const.c (pure_const_read_summary): Likewise.
27008 (propagate_pure_const): Likewise.
27009 * ipa-reference.c (generate_summary): Likewise.
27010 (read_write_all_from_decl): Likewise.
27011 (propagate): Likewise.
27012 (ipa_reference_read_optimization_summary): Likewise.
27013 * ipa-utils.c (ipa_merge_profiles): Likewise.
27014 * ipa.c (walk_polymorphic_call_targets): Likewise.
27015 (symbol_table::remove_unreachable_nodes): Likewise.
27016 (ipa_single_use): Likewise.
27017 * passes.c (execute_todo): Likewise.
27018 * predict.c (drop_profile): Likewise.
27019 * symtab.c (symtab_node::get_dump_name): New function.
27020 (symtab_node::dump_name): Likewise.
27021 (symtab_node::dump_asm_name): Likewise.
27022 (symtab_node::dump_references): Likewise.
27023 (symtab_node::dump_referring): Likewise.
27024 (symtab_node::dump_base): Likewise.
27025 (symtab_node::debug_symtab): Likewise.
27026 * tree-sra.c (convert_callers_for_node): Likewise.
27027 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
27028 * value-prof.c (init_node_map): Likewise.
27030 2017-05-23 Martin Liska <mliska@suse.cz>
27032 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
27033 and symtab_node::debug_symtab to symbol_table::debug.
27034 * cgraphunit.c (analyze_functions): Use the renamed function.
27035 (symbol_table::compile): Likewise.
27036 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
27037 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
27038 * passes.c (execute_todo): Likewise.
27039 * symtab.c (symbol_table::dump): New function.
27040 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
27042 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27044 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
27045 that nonconst implies exec.
27047 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27049 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
27050 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27051 (inline_edge_summary_vec): Turn into ...
27052 (ipa_call_summaries): ... this one.
27053 (redirect_to_unreachable, edge_set_predicate,
27054 evaluate_properties_for_edge, inline_summary_alloc,
27055 reset_ipa_call_summary, reset_inline_summary,
27056 inline_summary_t::duplicate): Update.
27057 (inline_edge_duplication_hook): Turn to ...
27058 (ipa_call_summary_t::duplicate): ... this one.
27059 (inline_edge_removal_hook): Turn to ...
27060 (ipa_call_summary_t::remove): ... this one.
27061 (dump_inline_edge_summary): Turn to ...
27062 (dump_ipa_call_summary): ... this one.
27063 (estimate_function_body_sizes): Update.
27064 (inline_update_callee_summaries): Update.
27065 (remap_edge_change_prob): Update.
27066 (remap_edge_summaries): Update.
27067 (inline_merge_summary): Update.
27068 (do_estimate_edge_time): Update.
27069 (inline_generate_summary): Update.
27070 (inline_read_section): Update.
27071 (inline_read_summary): Update.
27072 (inline_free_summary): Update.
27073 * ipa-inline.c (can_inline_edge_p): Update.
27074 (compute_inlined_call_time): Update.
27075 (want_inline_small_function_p): Update.
27076 (edge_badness): Update.
27077 (early_inliner): Update.
27078 * ipa-inline.h (inline_edge_summary): Turn to ...
27079 (ipa_call_summary): ... this one.
27080 (ipa_call_summary_t): New class.
27081 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27082 (ipa_call_summaries): New.
27083 (inline_edge_summary): Remove.
27084 (estimate_edge_growth): Update.
27085 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27086 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27087 * ipa-split.c (execute_split_functions): Update.
27088 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27090 2017-05-23 Tom de Vries <tom@codesourcery.com>
27092 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
27093 attributes): Document rdrand effective target.
27095 2017-05-23 Tom de Vries <tom@codesourcery.com>
27097 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
27098 attributes): Sort alphabetically.
27100 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
27102 * config/avr/genmultilib.awk: Use gsub instead of gensub.
27104 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
27107 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
27108 V2DF/V2DI splat into two separate patterns, one that handles
27109 registers, and the other that only handles memory. Drop support
27110 for splatting from a GPR on ISA 2.07 and then splitting the
27111 splat into direct move and splat.
27112 (vsx_splat_<mode>_reg): Likewise.
27113 (vsx_splat_<mode>_mem): Likewise.
27115 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
27117 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
27119 2017-05-22 Jakub Jelinek <jakub@redhat.com>
27121 PR middle-end/80809
27122 * omp-low.c (finish_taskreg_remap): New function.
27123 (finish_taskreg_scan): If unit size of ctx->record_type
27124 is non-constant, unshare the size expression and replace
27125 decls in it with possible outer var refs.
27127 PR middle-end/80809
27128 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
27129 GOVD_SHARED rather than GOVD_PRIVATE with it.
27130 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
27131 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
27133 PR middle-end/80853
27134 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
27135 as last argument to build_outer_var_ref for pointer bases of array
27136 section reductions.
27138 2017-05-19 Martin Sebor <msebor@redhat.com>
27140 * print-tree.c (print_node): Print DECL_READ_P flag.
27142 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27144 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
27145 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
27146 * cgraph.c: Likewise.
27147 * cgraphunit.c: Likewise.
27148 * gengtype.c: Likewise.
27149 * ipa-cp.c: Likewise.
27150 * ipa-devirt.c: Likewise.
27151 * ipa-icf.c: Likewise.
27152 * ipa-predicate.c: Likewise.
27153 * ipa-profile.c: Likewise.
27154 * ipa-prop.c: Likewise.
27155 * ipa-split.c: Likewise.
27157 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
27158 edge_predicate_pool, dump_inline_hints,
27159 inline_summary::account_size_time, redirect_to_unreachable,
27160 edge_set_predicate, set_hint_predicate,
27161 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
27162 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
27163 inline_summary_t::remove, remap_hint_predicate_after_duplication,
27164 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
27165 ipa_call_summary_t::remove, initialize_growth_caches,
27166 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
27167 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
27168 mark_modified, unmodified_parm_1, unmodified_parm,
27169 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
27170 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
27171 compute_bb_predicates, will_be_nonconstant_expr_predicate,
27172 will_be_nonconstant_predicate, record_modified_bb_info,
27173 get_minimal_bb, record_modified, param_change_prob,
27174 phi_result_unknown_predicate, predicate_for_phi_result,
27175 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
27176 estimate_function_body_sizes, compute_inline_parameters,
27177 compute_inline_parameters_for_curren, pass_data_inline_parameters,
27178 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
27179 inline_update_callee_summaries, remap_edge_change_prob,
27180 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
27181 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
27182 inline_analyze_function, inline_summary_t::insert,
27183 inline_generate_summary, read_ipa_call_summary, inline_read_section,
27184 inline_read_summary, write_ipa_call_summary, inline_write_summary,
27185 inline_free_summary): Move to ipa-fnsummary.h
27186 (predicate_t): Remove.
27187 * ipa-fnsummary.c: New file.
27188 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
27189 (enum inline_hints_vals, inline_hints, agg_position_info,
27190 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
27191 inline_summaries, ipa_call_summary, ipa_call_summary_t,
27192 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
27193 dump_inline_summary, dump_inline_hints, inline_generate_summary,
27194 inline_read_summary, inline_write_summary, inline_free_summary,
27195 inline_analyze_function, initialize_inline_failed,
27196 inline_merge_summary, inline_update_overall_summary,
27197 compute_inline_parameters): Move to ipa-fnsummary.h
27198 * ipa-fnsummary.h: New file.
27199 * ipa-inline-transform.h: Include ipa-inline.h.
27200 * ipa-inline.c: LIkewise.
27202 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27204 * ipa-inline.c (edge_badness): Use inlined_time instead of
27205 inline_summaries->get.
27207 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27209 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
27211 2017-05-22 Nathan Sidwell <nathan@acm.org>
27213 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
27214 (fdump-lang): Document 'raw' option.
27215 * dumpfile.h (TDI_tu): Delete.
27216 * dumpfile.c (dump_files): Remove translation-unit.
27217 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
27219 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
27221 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
27222 command option from $(AWK) call.
27223 * config/avr/genmultilib.awk: Simplify and rewrite so that it
27224 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
27225 [FORMAT]: Remove handling of variable.
27226 * config/avr/t-multilib: Regenerate.
27228 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27230 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
27232 (dump_inline_summary): Do not print self_time.
27233 (estimate_function_body_sizes): Do not set self_time.
27234 (compute_inline_parameters): Likewise.
27235 (inline_read_section, inline_write_summary): Do not stream self_time.
27236 * ipa-inline.h (inline_summary): Drop self_time.
27238 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27240 * ipa-inline-analysis.c (account_size_time): Rename to ...
27241 (inline_summary::account_size_time): ... this one.
27242 (reset_ipa_call_summary): Turn to ...
27243 (ipa_call_summary::reset): ... this one.
27244 (reset_inline_summary): Turn to ...
27245 (inline_summary::reset): ... this one.
27246 (inline_summary_t::remove): Update.
27247 (inline_summary_t::duplicate): Update.
27248 (ipa_call_summary_t::remove): Update.
27249 (dump_inline_summary): Update.
27250 (estimate_function_body_sizes): Update.
27251 (compute_inline_parameters): Update.
27252 (estimate_node_size_and_time): Update.
27253 (inline_merge_summary): Update.
27254 (inline_update_overall_summary): Update.
27255 (inline_read_section): Update.
27256 (inline_write_summary): Update.
27257 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
27258 add account_size_time and reset member functions.
27259 (ipa_call_summary): Add reset function.
27260 * ipa-predicate.h (predicate::operator &): Constify.
27262 2017-05-22 Richard Biener <rguenther@suse.de>
27264 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
27266 2017-05-19 Jason Merrill <jason@redhat.com>
27268 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
27270 2017-05-19 Marek Polacek <polacek@redhat.com>
27273 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
27274 TYPE_OVERFLOW_WRAPS checks.
27276 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
27278 * tree-core.h (enum omp_clause_default_kind): Add
27279 "OMP_CLAUSE_DEFAULT_PRESENT".
27280 * tree-pretty-print.c (dump_omp_clause): Handle it.
27281 * gimplify.c (enum gimplify_omp_var_data): Add
27282 "GOVD_MAP_FORCE_PRESENT".
27283 (gimplify_adjust_omp_clauses_1): Map it to
27284 "GOMP_MAP_FORCE_PRESENT".
27285 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
27287 * gimplify.c (oacc_default_clause): Clarify.
27289 2017-05-19 Nathan Sidwell <nathan@acm.org>
27291 LANG_HOOK_REGISTER_DUMPS
27292 * toplev.c (general_init): Call register dump lang hook.
27293 * doc/invoke.texi: Document -fdump-lang option family.
27294 * dumpfile.c (dump_files): Remove class dump here.
27295 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
27296 * dumpfile.h (tree_dump_index): Remove TDI_class.
27297 * langhooks-def.h (lhd_register_dumps): Declare.
27298 (LANG_HOOKS_REGISTER_DUMPS): Define.
27299 (LANG_HOOKS_INITIALIZER): Add it.
27300 * langhooks.c (lhd_register_dumps): Define.
27301 * langhooks.h (struct lang_hooks): Add register_dumps.
27303 2017-05-19 Nathan Sidwell <nathan@acm.org>
27305 * context.h (context::set_passes): New.
27306 * context.c (context::context): Do not create pass manager.
27307 * toplev.c (general_init): Create pass manager here.
27309 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
27311 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
27312 use this splitter if two add or or instructions would also work for
27313 the constant we want to generate.
27315 2017-05-19 Richard Biener <rguenther@suse.de>
27318 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
27319 predicate evaluation.
27321 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27323 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
27325 * ipa-inline.c (want_inline_small_function_p): Do not cast to
27328 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27330 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
27331 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27332 (inline_edge_summary_vec): Turn into ...
27333 (ipa_call_summaries): ... this one.
27334 (redirect_to_unreachable, edge_set_predicate,
27335 evaluate_properties_for_edge, inline_summary_alloc,
27336 reset_ipa_call_summary, reset_inline_summary,
27337 inline_summary_t::duplicate): Update.
27338 (inline_edge_duplication_hook): Turn to ...
27339 (ipa_call_summary_t::duplicate): ... this one.
27340 (inline_edge_removal_hook): Turn to ...
27341 (ipa_call_summary_t::remove): ... this one.
27342 (dump_inline_edge_summary): Turn to ...
27343 (dump_ipa_call_summary): ... this one.
27344 (estimate_function_body_sizes): Update.
27345 (inline_update_callee_summaries): Update.
27346 (remap_edge_change_prob): Update.
27347 (remap_edge_summaries): Update.
27348 (inline_merge_summary): Update.
27349 (do_estimate_edge_time): Update.
27350 (inline_generate_summary): Update.
27351 (inline_read_section): Update.
27352 (inline_read_summary): Update.
27353 (inline_free_summary): Update.
27354 * ipa-inline.c (can_inline_edge_p): Update.
27355 (compute_inlined_call_time): Update.
27356 (want_inline_small_function_p): Update.
27357 (edge_badness): Update.
27358 (early_inliner): Update.
27359 * ipa-inline.h (inline_edge_summary): Turn to ...
27360 (ipa_call_summary): ... this one.
27361 (ipa_call_summary_t): New class.
27362 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27363 (ipa_call_summaries): New.
27364 (inline_edge_summary): Remove.
27365 (estimate_edge_growth): Update.
27366 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27367 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27368 * ipa-split.c (execute_split_functions): Update.
27369 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27371 2017-05-19 Richard Biener <rguenther@suse.de>
27373 PR middle-end/80764
27374 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
27376 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
27378 * config/rs6000/rs6000.c (struct machine_function): Add field
27379 fpr_is_wrapped_separately.
27380 (rs6000_get_separate_components): Use 64 components. Handle the
27381 new FPR components.
27382 (rs6000_components_for_bb): Handle the FPR components.
27383 (rs6000_emit_prologue_components): Handle the FPR components.
27384 (rs6000_emit_epilogue_components): Handle the FPR components.
27385 (rs6000_set_handled_components): Handle the FPR components.
27386 (rs6000_emit_prologue): Don't output prologue code for those FPRs
27387 that are already separately shrink-wrapped.
27388 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
27389 that are already separately shrink-wrapped.
27391 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27394 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
27397 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
27398 (define_peephole2 for Altivec d-form load): Add peepholes to catch
27399 cases where the register allocator uses a move and an offsettable
27400 memory operation to/from a FPR register on ISA 2.06/2.07.
27401 (define_peephole2 for Altivec d-form store): Likewise.
27403 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
27406 * config/i386/mmx.md (*mov<mode>_internal): Enable
27407 alternatives 11, 12, 13 and 14 also for 32bit targets.
27408 Remove alternatives 15, 16, 17 and 18.
27409 * config/i386/sse.md (vec_concatv2di): Change
27410 alternative (!x, *y) to (x, ?!*Yn).
27412 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
27414 * dumpfile.h (enum dump_kind): Remove stray comma.
27416 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27418 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
27419 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
27420 predicate::num_conditions
27421 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
27422 (CHANGED): turn into predicate::changed.
27423 (agg_position_info): Move to ipa-predicate.h
27424 (add_condition, predicate::add_clause, predicate::operator &=,
27425 predicate::or_with, predicate::evaluate, predicate::probability,
27426 dump_condition, dump_clause, predicate::dump,
27427 predicate::remap_after_duplication, predicate::remap_after_inlining,
27428 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
27429 (evaluate_conditions_for_known_args): Update.
27430 (set_cond_stmt_execution_predicate): Update.
27431 * ipa-inline.h: Include ipa-predicate.h
27432 (condition, inline_param_summary, conditions, agg_position_info,
27433 predicate): Move to ipa-predicate.h
27434 * ipa-predicate.c: New file.
27435 * ipa-predicate.h: New file.
27437 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
27439 * final.c (leaf_function_p): Check we are not in a sequence.
27441 2017-05-18 Martin Liska <mliska@suse.cz>
27443 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
27444 * dumpfile.c (dump_register): Use new enum dump_kind.
27445 (get_dump_file_name): Likewise.
27446 (dump_enable_all): Likewise.
27447 (dump_switch_p_1): Likewise.
27448 (enable_rtl_dump_file): Remove usage of TDF_RTL.
27449 * dumpfile.h (enum dump_kind): New enum type.
27450 (struct dump_file_info): Create constructor and
27451 format fields and comments.
27452 * passes.c (pass_manager::register_one_dump_file):
27454 * statistics.c (statistics_early_init): Likewise.
27455 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
27456 TDF_TREE with TDF_SLIM.
27457 (gather_memory_references_ref): Likewise.
27459 2017-05-18 Martin Liska <mliska@suse.cz>
27461 * vec.h (struct vnull): Use it.
27463 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27465 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
27466 (true_predicate, false_predicate, true_predicate_p,
27467 false_predicate_p): Remove.
27468 (single_cond_predicate, not_inlined_predicate): Turn to member function
27470 (add_condition): Update.
27471 (add_clause): Turn to...
27472 (predicate::add_clause): ... this one; update; allow passing NULL
27474 (and_predicates): Turn to ...
27475 (predicate::operator &=): ... this one.
27476 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
27477 (or_predicates): Turn to ...
27478 (predicate::or_with): ... this one.
27479 (evaluate_predicate): Turn to ...
27480 (predicate::evaluate): ... this one.
27481 (predicate_probability): Turn to ...
27482 (predicate::probability): ... this one.
27483 (dump_condition): Update.
27484 (dump_predicate): Turn to ...
27485 (predicate::dump): ... this one.
27486 (account_size_time): Update.
27487 (edge_set_predicate): Update.
27488 (set_hint_predicate): UPdate.
27489 (evaluate_conditions_for_known_args): Update.
27490 (evaluate_properties_for_edge): Update.
27491 (remap_predicate_after_duplication): Turn to...
27492 (predicate::remap_after_duplication): ... this one.
27493 (remap_hint_predicate_after_duplication): Update.
27494 (inline_summary_t::duplicate): UPdate.
27495 (dump_inline_edge_summary): Update.
27496 (dump_inline_summary): Update.
27497 (set_cond_stmt_execution_predicate): Update.
27498 (set_switch_stmt_execution_predicate): Update.
27499 (compute_bb_predicates): Update.
27500 (will_be_nonconstant_expr_predicate): Update.
27501 (will_be_nonconstant_predicate): Update.
27502 (phi_result_unknown_predicate): Update.
27503 (predicate_for_phi_result): Update.
27504 (array_index_predicate): Update.
27505 (estimate_function_body_sizes): Update.
27506 (estimate_node_size_and_time): Update.
27507 (estimate_ipcp_clone_size_and_time): Update.
27508 (remap_predicate): Rename to ...
27509 (predicate::remap_after_inlining): ... this one.
27510 (remap_hint_predicate): Update.
27511 (inline_merge_summary): Update.
27512 (inline_update_overall_summary): Update.
27513 (estimate_size_after_inlining): Update.
27514 (read_predicate): Rename to ...
27515 (predicate::stream_in): ... this one.
27516 (read_inline_edge_summary): Update.
27517 (write_predicate): Rename to ...
27518 (predicate::stream_out): ... this one.
27519 (write_inline_edge_summary): Update.
27520 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
27521 (clause_t): Turn to uint32_t
27522 (predicate): Turn to class; implement constructor and operators
27524 (size_time_entry): Update.
27525 (inline_summary): Update.
27526 (inline_edge_summary): Update.
27528 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
27530 * fold-const.c (fold_binary_loc): Move transformation...
27531 * match.pd (C - X CMP X): ... here.
27533 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
27535 * config/sparc/sparc.c (sparc_option_override): Set function
27536 alignment for -mcpu=niagara7 to 64 to match the I$ line.
27537 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
27539 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
27541 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
27543 2017-05-18 Marek Polacek <polacek@redhat.com>
27546 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
27547 (pass_ubsan::execute): Call gimple_assign_single_p instead of
27548 gimple_assign_load_p.
27550 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
27552 PR middle-end/80692
27553 * real.c (do_compare): Give decimal_do_compare preference over
27554 comparing just the signs.
27556 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
27558 * doc/md.texi (Canonicalization of Instructions): Describe the
27559 canonical form of instructions that inherently set a condition
27562 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
27564 PR middle-end/80775
27565 * tree-cfg.c: Move deletion of unreachable case statements to after
27566 the merging of consecutive case labels.
27568 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
27570 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
27571 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
27572 restoring of callee-saved registers.
27574 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
27576 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
27577 * config/visium/visium.c (single_set_and_flags): Likewise.
27578 * config/visium/visium.md (Substitutions): Likewise.
27580 2017-05-17 Martin Liska <mliska@suse.cz>
27582 * cfg.c: Introduce dump_flags_t type and
27583 use it instead of int type.
27585 * cfghooks.c: Likewise.
27586 * cfghooks.h (struct cfg_hooks): Likewise.
27587 * cfgrtl.c: Likewise.
27588 * cfgrtl.h: Likewise.
27589 * cgraph.c (cgraph_node::get_body): Likewise.
27590 * coretypes.h: Likewise.
27591 * domwalk.c: Likewise.
27592 * domwalk.h: Likewise.
27593 * dumpfile.c (struct dump_option_value_info): Likewise.
27594 (dump_enable_all): Likewise.
27595 (dump_switch_p_1): Likewise.
27596 (opt_info_switch_p): Likewise.
27597 * dumpfile.h (enum tree_dump_index): Likewise.
27598 (struct dump_file_info): Likewise.
27599 * genemit.c: Likewise.
27600 * generic-match-head.c: Likewise.
27601 * gengtype.c (open_base_files): Likewise.
27602 * gimple-pretty-print.c: Likewise.
27603 * gimple-pretty-print.h: Likewise.
27604 * graph.c (print_graph_cfg): Likewise.
27605 * graphite-scop-detection.c (dot_all_sese): Likewise.
27606 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
27607 * loop-unroll.c (report_unroll): Likewise.
27608 * passes.c (pass_manager::register_one_dump_file): Likewise.
27609 * print-tree.c: Likewise.
27610 * statistics.c: Likewise.
27611 * tree-cfg.c: Likewise.
27612 * tree-cfg.h: Likewise.
27613 * tree-dfa.c: Likewise.
27614 * tree-dfa.h: Likewise.
27615 * tree-dump.c (dump_function): Likewise.
27616 * tree-dump.h (struct dump_info): Likewise.
27617 * tree-pretty-print.c: Likewise.
27618 * tree-pretty-print.h: Likewise.
27619 * tree-ssa-live.c: Likewise.
27620 * tree-ssa-live.h: Likewise.
27621 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
27622 * tree-vect-loop.c: Likewise.
27623 * tree-vect-slp.c: Likewise.
27625 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
27626 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27628 PR tree-optimization/80457
27629 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
27630 of all arguments to a statement as scalar_to_vec operations.
27631 (vectorizable_call): Adjust call to vect_model_simple_cost for
27633 (vectorizable_conversion): Likewise.
27634 (vectorizable_assignment): Likewise.
27635 (vectorizable_shift): Likewise.
27636 (vectorizable_operation): Likewise.
27637 (vectorizable_comparison): Likewise.
27638 (vect_is_simple_cond): Record the def types for operands.
27639 (vectorizable_condition): Likewise, call vect_model_simple_cost.
27640 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
27641 for statement argument count.
27643 2017-05-16 Carl Love <cel@us.ibm.com>
27645 * config/rs6000/rs6000-c: Add support for built-in functions
27646 vector unsigned long long vec_bperm (vector unsigned long long,
27647 vector unsigned char)
27648 vector signed long long vec_mule (vector signed int,
27650 vector unsigned long long vec_mule (vector unsigned int,
27651 vector unsigned int)
27652 vector signed long long vec_mulo (vector signed int,
27654 vector unsigned long long vec_mulo (vector unsigned int,
27655 vector unsigned int)
27656 vector signed char vec_sldw (vector signed char,
27657 vector signed char,
27659 vector unsigned char vec_sldw (vector unsigned char,
27660 vector unsigned char,
27662 vector signed short vec_sldw (vector signed short,
27663 vector signed short,
27665 vector unsigned short vec_sldw (vector unsigned short,
27666 vector unsigned short,
27668 vector signed int vec_sldw (vector signed int,
27671 vector unsigned int vec_sldw (vector unsigned int,
27672 vector unsigned int,
27674 vector signed long long vec_sldw (vector signed long long,
27675 vector signed long long,
27677 vector unsigned long long vec_sldw (vector unsigned long long,
27678 vector unsigned long long,
27680 * config/rs6000/rs6000-c: Add support for built-in functions
27681 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
27682 * config/rs6000/altivec.h: Add defintion for vec_sldw.
27683 * doc/extend.texi: Update the built-in documentation for the
27684 new built-in functions.
27686 2017-05-16 Marek Polacek <polacek@redhat.com>
27690 * tree.c (save_expr): Don't fold the expression.
27692 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
27694 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
27695 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
27696 and (?*y,m). Update insn attributes.
27698 2017-05-16 Martin Liska <mliska@suse.cz>
27700 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
27701 flags argument of print_gimple_stmt, print_gimple_expr,
27702 print_generic_stmt and print_generic_expr.
27703 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
27704 * coretypes.h: Likewise.
27705 * except.c (dump_eh_tree): Likewise.
27706 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
27707 * gimple-pretty-print.h: Likewise.
27708 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
27709 (backprop::push_to_worklist): Likewise.
27710 (backprop::pop_from_worklist): Likewise.
27711 (backprop::process_use): Likewise.
27712 (backprop::intersect_uses): Likewise.
27713 (note_replacement): Likewise.
27714 * gimple-ssa-store-merging.c
27715 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
27716 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
27717 (pass_store_merging::execute): Likewise.
27718 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
27719 (ssa_base_cand_dump_callback): Likewise.
27720 (dump_incr_vec): Likewise.
27721 (replace_refs): Likewise.
27722 (replace_mult_candidate): Likewise.
27723 (create_add_on_incoming_edge): Likewise.
27724 (create_phi_basis): Likewise.
27725 (insert_initializers): Likewise.
27726 (all_phi_incrs_profitable): Likewise.
27727 (introduce_cast_before_cand): Likewise.
27728 (replace_one_candidate): Likewise.
27729 * gimplify.c (gimplify_expr): Likewise.
27730 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
27731 (set_rename): Likewise.
27732 (rename_uses): Likewise.
27733 (copy_loop_phi_nodes): Likewise.
27734 (add_close_phis_to_merge_points): Likewise.
27735 (copy_loop_close_phi_args): Likewise.
27736 (copy_cond_phi_args): Likewise.
27737 (graphite_copy_stmts_from_block): Likewise.
27738 (translate_pending_phi_nodes): Likewise.
27739 * graphite-poly.c (print_pdr): Likewise.
27740 (dump_gbb_cases): Likewise.
27741 (dump_gbb_conditions): Likewise.
27742 (print_scop_params): Likewise.
27743 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
27744 (build_cross_bb_scalars_use): Likewise.
27745 (gather_bbs::before_dom_children): Likewise.
27746 * hsa-dump.c (dump_hsa_immed): Likewise.
27747 * ipa-cp.c (print_ipcp_constant_value): Likewise.
27748 (get_replacement_map): Likewise.
27749 * ipa-inline-analysis.c (dump_condition): Likewise.
27750 (estimate_function_body_sizes): Likewise.
27751 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
27752 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
27753 * ipa-prop.c (ipa_dump_param): Likewise.
27754 (ipa_print_node_jump_functions_for_edge): Likewise.
27755 (ipa_modify_call_arguments): Likewise.
27756 (ipa_modify_expr): Likewise.
27757 (ipa_dump_param_adjustments): Likewise.
27758 (ipa_dump_agg_replacement_values): Likewise.
27759 (ipcp_modif_dom_walker::before_dom_children): Likewise.
27760 * ipa-pure-const.c (check_stmt): Likewise.
27761 (pass_nothrow::execute): Likewise.
27762 * ipa-split.c (execute_split_functions): Likewise.
27763 * omp-offload.c (dump_oacc_loop_part): Likewise.
27764 (dump_oacc_loop): Likewise.
27765 * trans-mem.c (tm_log_emit): Likewise.
27766 (tm_memopt_accumulate_memops): Likewise.
27767 (dump_tm_memopt_set): Likewise.
27768 (dump_tm_memopt_transform): Likewise.
27769 * tree-cfg.c (gimple_verify_flow_info): Likewise.
27770 (print_loop): Likewise.
27771 * tree-chkp-opt.c (chkp_print_addr): Likewise.
27772 (chkp_gather_checks_info): Likewise.
27773 (chkp_get_check_result): Likewise.
27774 (chkp_remove_check_if_pass): Likewise.
27775 (chkp_use_outer_bounds_if_possible): Likewise.
27776 (chkp_reduce_bounds_lifetime): Likewise.
27777 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
27778 (chkp_mark_completed_bounds): Likewise.
27779 (chkp_register_incomplete_bounds): Likewise.
27780 (chkp_mark_invalid_bounds): Likewise.
27781 (chkp_maybe_copy_and_register_bounds): Likewise.
27782 (chkp_build_returned_bound): Likewise.
27783 (chkp_get_bound_for_parm): Likewise.
27784 (chkp_build_bndldx): Likewise.
27785 (chkp_get_bounds_by_definition): Likewise.
27786 (chkp_generate_extern_var_bounds): Likewise.
27787 (chkp_get_bounds_for_decl_addr): Likewise.
27788 * tree-chrec.c (chrec_apply): Likewise.
27789 * tree-data-ref.c (dump_data_reference): Likewise.
27790 (dump_subscript): Likewise.
27791 (dump_data_dependence_relation): Likewise.
27792 (analyze_overlapping_iterations): Likewise.
27793 * tree-inline.c (expand_call_inline): Likewise.
27794 (tree_function_versioning): Likewise.
27795 * tree-into-ssa.c (dump_defs_stack): Likewise.
27796 (dump_currdefs): Likewise.
27797 (dump_names_replaced_by): Likewise.
27798 (dump_update_ssa): Likewise.
27799 (update_ssa): Likewise.
27800 * tree-object-size.c (pass_object_sizes::execute): Likewise.
27801 * tree-parloops.c (build_new_reduction): Likewise.
27802 (try_create_reduction_list): Likewise.
27803 (ref_conflicts_with_region): Likewise.
27804 (oacc_entry_exit_ok_1): Likewise.
27805 (oacc_entry_exit_single_gang): Likewise.
27806 * tree-pretty-print.h: Likewise.
27807 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
27808 (get_scalar_evolution): Likewise.
27809 (add_to_evolution): Likewise.
27810 (get_loop_exit_condition): Likewise.
27811 (analyze_evolution_in_loop): Likewise.
27812 (analyze_initial_condition): Likewise.
27813 (analyze_scalar_evolution): Likewise.
27814 (instantiate_scev): Likewise.
27815 (number_of_latch_executions): Likewise.
27816 (gather_chrec_stats): Likewise.
27817 (final_value_replacement_loop): Likewise.
27818 (scev_const_prop): Likewise.
27819 * tree-sra.c (dump_access): Likewise.
27820 (disqualify_candidate): Likewise.
27821 (create_access): Likewise.
27822 (reject): Likewise.
27823 (maybe_add_sra_candidate): Likewise.
27824 (create_access_replacement): Likewise.
27825 (analyze_access_subtree): Likewise.
27826 (analyze_all_variable_accesses): Likewise.
27827 (sra_modify_assign): Likewise.
27828 (initialize_constant_pool_replacements): Likewise.
27829 (find_param_candidates): Likewise.
27830 (decide_one_param_reduction): Likewise.
27831 (replace_removed_params_ssa_names): Likewise.
27832 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
27833 * tree-ssa-copy.c (dump_copy_of): Likewise.
27834 (copy_prop_visit_cond_stmt): Likewise.
27835 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
27836 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
27837 (record_equivalences_from_stmt): Likewise.
27838 * tree-ssa-dse.c (compute_trims): Likewise.
27839 (delete_dead_call): Likewise.
27840 (delete_dead_assignment): Likewise.
27841 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
27842 (forward_propagate_into_cond): Likewise.
27843 (pass_forwprop::execute): Likewise.
27844 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
27845 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
27847 (move_computations_worker): Likewise.
27848 (execute_sm): Likewise.
27849 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
27850 (remove_exits_and_undefined_stmts): Likewise.
27851 (remove_redundant_iv_tests): Likewise.
27852 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
27853 (adjust_iv_update_pos): Likewise.
27854 * tree-ssa-math-opts.c (bswap_replace): Likewise.
27855 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
27856 (value_replacement): Likewise.
27857 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
27858 * tree-ssa-pre.c (print_pre_expr): Likewise.
27859 (get_representative_for): Likewise.
27860 (create_expression_by_pieces): Likewise.
27861 (insert_into_preds_of_block): Likewise.
27862 (eliminate_insert): Likewise.
27863 (eliminate_dom_walker::before_dom_children): Likewise.
27864 (eliminate): Likewise.
27865 (remove_dead_inserted_code): Likewise.
27866 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
27867 * tree-ssa-reassoc.c (get_rank): Likewise.
27868 (eliminate_duplicate_pair): Likewise.
27869 (eliminate_plus_minus_pair): Likewise.
27870 (eliminate_not_pairs): Likewise.
27871 (undistribute_ops_list): Likewise.
27872 (eliminate_redundant_comparison): Likewise.
27873 (update_range_test): Likewise.
27874 (optimize_range_tests_var_bound): Likewise.
27875 (optimize_vec_cond_expr): Likewise.
27876 (rewrite_expr_tree): Likewise.
27877 (rewrite_expr_tree_parallel): Likewise.
27878 (linearize_expr): Likewise.
27879 (break_up_subtract): Likewise.
27880 (linearize_expr_tree): Likewise.
27881 (attempt_builtin_powi): Likewise.
27882 (attempt_builtin_copysign): Likewise.
27883 (transform_stmt_to_copy): Likewise.
27884 (transform_stmt_to_multiply): Likewise.
27885 (dump_ops_vector): Likewise.
27886 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
27887 (print_scc): Likewise.
27888 (set_ssa_val_to): Likewise.
27889 (visit_reference_op_store): Likewise.
27890 (visit_use): Likewise.
27891 (sccvn_dom_walker::before_dom_children): Likewise.
27892 (run_scc_vn): Likewise.
27893 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
27895 (expr_hash_elt::print): Likewise.
27896 (const_and_copies::pop_to_marker): Likewise.
27897 (const_and_copies::record_const_or_copy_raw): Likewise.
27898 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
27899 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
27900 (dump_predicates): Likewise.
27901 (find_uninit_use): Likewise.
27902 (warn_uninitialized_phi): Likewise.
27903 (pass_late_warn_uninitialized::execute): Likewise.
27904 * tree-ssa.c (verify_vssa): Likewise.
27905 (verify_ssa): Likewise.
27906 (maybe_optimize_var): Likewise.
27907 * tree-vrp.c (dump_value_range): Likewise.
27908 (dump_all_value_ranges): Likewise.
27909 (dump_asserts_for): Likewise.
27910 (register_edge_assert_for_2): Likewise.
27911 (vrp_visit_cond_stmt): Likewise.
27912 (vrp_visit_switch_stmt): Likewise.
27913 (vrp_visit_stmt): Likewise.
27914 (vrp_visit_phi_node): Likewise.
27915 (simplify_cond_using_ranges_1): Likewise.
27916 (fold_predicate_in): Likewise.
27917 (evrp_dom_walker::before_dom_children): Likewise.
27918 (evrp_dom_walker::push_value_range): Likewise.
27919 (evrp_dom_walker::pop_value_range): Likewise.
27920 (execute_early_vrp): Likewise.
27922 2017-05-16 Richard Biener <rguenther@suse.de>
27924 * dwarf2out.c (loc_list_from_tree_1): Do not create
27925 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
27927 2017-05-16 Richard Biener <rguenther@suse.de>
27929 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
27931 (note_variable_value_in_expr): If we resolved the decl ref
27932 do not push to the stack.
27934 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
27936 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
27937 operations in fast-math mode.
27938 (vaddq_f16): Likewise.
27939 (vmul_f16): Likewise.
27940 (vmulq_f16): Likewise.
27941 (vsub_f16): Likewise.
27942 (vsubq_f16): Likewise.
27943 * config/arm/neon.md (add<mode>3): New.
27945 (fma:<VH:mode>3): New. Also remove outdated comment.
27948 2017-05-16 Martin Liska <mliska@suse.cz>
27952 * ipa-devirt.c (warn_types_mismatch): Fix typo.
27953 (odr_types_equivalent_p): Likewise.
27955 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
27957 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
27959 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
27962 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
27963 non-interunit SSE move alternatives with '?'.
27964 (zero-extendsidi peephole2): New peephole to skip intermediate
27965 general register in SSE zero-extend sequence.
27967 2017-05-15 Jeff Law <law@redhat.com>
27969 * reorg.c (relax_delay_slots): Create a new variable to hold
27970 the temporary target rather than clobbering TARGET_LABEL.
27972 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
27973 missing argument to extract_bit_field call.
27974 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
27976 2017-05-15 Martin Liska <mliska@suse.cz>
27979 * gcc.c (process_command): Do not allow empty argument of -o option.
27981 2017-05-15 Renlin Li <renlin.li@arm.com>
27983 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
27984 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
27985 * config/aarch64/constraints.md (Usf): Add long call check.
27986 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
27987 (call_value): Likewise.
27988 (sibcall): Likewise.
27989 (sibcall_value): Likewise.
27991 (call_value_insn): New.
27992 (sibcall_insn): Update rtx pattern.
27993 (sibcall_value_insn): Likewise.
27994 (call_internal): Remove.
27995 (call_value_internal): Likewise.
27996 (sibcall_internal): Likewise.
27997 (sibcall_value_internal): Likewise.
27998 (call_reg): Likewise.
27999 (call_symbol): Likewise.
28000 (call_value_reg): Likewise.
28001 (call_value_symbol): Likewise.
28003 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
28006 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
28008 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
28010 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
28011 compatible with CCGOCmode and with CCZmode.
28013 2017-05-14 Martin Sebor <msebor@redhat.com>
28015 PR middle-end/77671
28016 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
28017 (gimple_fold_builtin_snprintf): Same.
28018 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
28019 (gimple_fold_builtin_snprintf): Same.
28020 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
28021 of character types.
28022 (is_call_safe): New function.
28023 (try_substitute_return_value): Call it.
28024 (try_simplify_call): New function.
28025 (pass_sprintf_length::handle_gimple_call): Call it.
28027 2017-05-14 Martin Sebor <msebor@redhat.com>
28029 PR middle-end/80669
28030 * builtins.c (expand_builtin_stpncpy): Simplify.
28032 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
28034 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
28035 * config/i386/i386.h
28036 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
28037 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
28038 (struct machine_function): Add new members call_ms2sysv,
28039 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
28040 (struct machine_frame_state): New fields sp_realigned and
28041 sp_realigned_offset.
28042 * config/i386/i386.c
28043 (enum xlogue_stub): New enum.
28044 (enum xlogue_stub_sets): New enum.
28045 (class xlogue_layout): New class.
28046 (struct ix86_frame): New fields stack_realign_allocate_offset,
28047 stack_realign_offset and outlined_save_offset. Modify comments to
28048 detail stack layout when using out-of-line stubs.
28049 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
28050 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
28051 -mcall-ms2sysv-xlogues.
28052 (stub_managed_regs): New static variable.
28053 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
28054 registers managed by out-of-line stub.
28055 (disable_call_ms2sysv_xlogues): New function.
28056 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
28057 m->call_ms2sysv when appropriate and compute frame layout for
28059 (sp_valid_at, fp_valid_at): New inline functions.
28060 (choose_basereg): New function.
28061 (choose_baseaddr): Add align parameter, use choose_basereg and modify
28063 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
28064 Use align parameter of choose_baseaddr to generated aligned SSE movs
28066 (pro_epilogue_adjust_stack): Modify to track
28067 machine_frame_state::sp_realigned.
28068 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
28069 (ix86_nsaved_sseregs): Likewise.
28070 (ix86_emit_save_regs): Likewise.
28071 (ix86_emit_save_regs_using_mov): Likewise.
28072 (ix86_emit_save_sse_regs_using_mov): Likewise.
28073 (get_scratch_register_on_entry): Likewise.
28074 (gen_frame_set): New function.
28075 (gen_frame_load): Likewise.
28076 (gen_frame_store): Likewise.
28077 (emit_outlined_ms2sysv_save): Likewise.
28078 (emit_outlined_ms2sysv_restore): Likewise.
28079 (ix86_expand_prologue): Modify stack re-alignment code and call
28080 emit_outlined_ms2sysv_save when appropriate.
28081 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
28082 parameter rtx_insn *insn, which allows the function to be used to only
28083 generate the notes.
28084 (ix86_expand_epilogue): Modify validity checks of frame and stack
28085 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
28086 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
28087 * config/i386/predicates.md
28088 (save_multiple): New predicate.
28089 (restore_multiple): Likewise.
28090 * config/i386/sse.md
28091 (save_multiple<mode>): New pattern.
28092 (save_multiple_realign<mode>): Likewise.
28093 (restore_multiple<mode>): Likewise.
28094 (restore_multiple_and_return<mode>): Likewise.
28095 (restore_multiple_leave_return<mode>): Likewise.
28096 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
28098 2017-05-14 Julia Koval <julia.koval@intel.com>
28100 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
28101 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
28102 (__builtin_ia32_xsetbv): New builtins.
28103 * config/i386/i386.c (ix86_expand_special_args_builtin):
28105 (ix86_expand_builtin): Special expand for new intrinsics.
28106 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
28107 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
28108 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
28110 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28112 * cfganal.c (inverted_post_order_compute): Change argument type
28114 * cfganal.h (inverted_post_order_compute): Adjust prototype.
28115 * df-core.c (rest_of_handle_df_initialize): Adjust.
28116 (rest_of_handle_df_finish): Likewise.
28117 (df_analyze_1): Likewise.
28118 (df_analyze): Likewise.
28119 (loop_inverted_post_order_compute): Change argument to be a vec *.
28120 (df_analyze_loop): Adjust.
28121 (df_get_n_blocks): Likewise.
28122 (df_get_postorder): Likewise.
28123 * df.h (struct df_d): Change field to be a vec.
28124 * lcm.c (compute_laterin): Adjust.
28125 (compute_available): Likewise.
28126 * lra-lives.c (lra_create_live_ranges_1): Likewise.
28127 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
28128 * tree-ssa-pre.c (compute_antic): Likewise.
28130 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28132 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
28133 (depth_first_search::depth_first_search): Change structure init
28134 function to this constructor.
28135 (depth_first_search::add_bb): Rename function to this member.
28136 (depth_first_search::execute): Likewise.
28137 (flow_dfs_compute_reverse_finish): Adjust.
28139 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28141 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
28142 (longest_simple_path): Likewise.
28143 * shrink-wrap.c (spread_components): Likewise.
28144 (disqualify_problematic_components): Likewise.
28145 (emit_common_heads_for_components): Likewise.
28146 (emit_common_tails_for_components): Likewise.
28147 (insert_prologue_epilogue_for_components): Likewise.
28149 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28151 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
28154 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28156 * df-core.c (df_set_blocks): Start using auto_bitmap.
28157 (df_compact_blocks): Likewise.
28158 * df-problems.c (df_rd_confluence_n): Likewise.
28159 * df-scan.c (df_insn_rescan_all): Likewise.
28160 (df_process_deferred_rescans): Likewise.
28161 (df_update_entry_block_defs): Likewise.
28162 (df_update_exit_block_uses): Likewise.
28163 (df_entry_block_bitmap_verify): Likewise.
28164 (df_exit_block_bitmap_verify): Likewise.
28165 (df_scan_verify): Likewise.
28166 * lra-constraints.c (lra_constraints): Likewise.
28167 (undo_optional_reloads): Likewise.
28168 (lra_undo_inheritance): Likewise.
28169 * lra-remat.c (calculate_gen_cands): Likewise.
28170 (do_remat): Likewise.
28171 * lra-spills.c (assign_spill_hard_regs): Likewise.
28172 (spill_pseudos): Likewise.
28173 * tree-ssa-pre.c (bitmap_set_and): Likewise.
28174 (bitmap_set_subtract_values): Likewise.
28176 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28178 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
28179 management with auto_bitmap.
28180 (fix_inter_tick): Likewise.
28181 (fix_recovery_deps): Likewise.
28182 * ira.c (add_store_equivs): Likewise.
28183 (find_moveable_pseudos): Likewise.
28184 (split_live_ranges_for_shrink_wrap): Likewise.
28185 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
28186 (rtx_reuse_manager::seen_def_p): Likewise.
28187 (rtx_reuse_manager::set_seen_def): Likewise.
28188 * print-rtl.h (class rtx_reuse_manager): Likewise.
28190 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28192 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
28194 (migrate_btr_def): Likewise.
28195 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
28196 * df-core.c (loop_post_order_compute): Likewise.
28197 (loop_inverted_post_order_compute): Likewise.
28198 * hsa-common.h: Likewise.
28199 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
28200 * init-regs.c (initialize_uninitialized_regs): Likewise.
28201 * ipa-inline.c (resolve_noninline_speculation): Likewise.
28202 (inline_small_functions): Likewise.
28203 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
28204 * ira.c (combine_and_move_insns): Likewise.
28205 (build_insn_chain): Likewise.
28206 * loop-invariant.c (find_invariants): Likewise.
28207 * lower-subreg.c (propagate_pseudo_copies): Likewise.
28208 * predict.c (tree_predict_by_opcode): Likewise.
28209 (predict_paths_leading_to): Likewise.
28210 (predict_paths_leading_to_edge): Likewise.
28211 (estimate_loops_at_level): Likewise.
28212 (estimate_loops): Likewise.
28213 * shrink-wrap.c (try_shrink_wrapping): Likewise.
28214 (spread_components): Likewise.
28215 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
28216 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
28217 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
28218 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
28219 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
28220 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
28221 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
28222 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
28223 (mark_threaded_blocks): Likewise.
28224 (thread_through_all_blocks): Likewise.
28225 * tree-ssa.c (verify_ssa): Likewise.
28226 (execute_update_addresses_taken): Likewise.
28227 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
28229 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28231 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
28233 (post_order_compute): Likewise.
28234 (inverted_post_order_compute): Likewise.
28235 (pre_and_rev_post_order_compute_fn): Likewise.
28237 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28239 * genrecog.c (int_set::int_set): Explicitly construct our
28240 auto_vec base class.
28241 * vec.h (auto_vec::auto_vec): New constructor.
28243 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28245 * bitmap.h (class auto_bitmap): New constructor taking
28246 bitmap_obstack * argument.
28248 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28250 * bitmap.h (class auto_bitmap): Change type of m_bits to
28251 bitmap_head, and adjust ctor / dtor and member operators.
28253 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
28255 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
28256 when returned register mode doesn't match original mode.
28258 2017-05-12 Jeff Law <law@redhat.com>
28259 Jakub Jelinek <jakub@redhat.com>
28261 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
28262 we look for cc setter after the compare-elim changes.
28263 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
28264 within the vector to match what compare-elim now expects.
28265 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
28266 (xorsi3_flags, one_cmplsi2_flags): Likewise.
28268 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
28269 after the compare-elim changes.
28270 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
28271 the vector to match what compare-elim now expects.
28272 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
28273 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
28274 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
28275 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
28276 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
28278 * config/visium/visium.c (single_set_and_flags): Fix where
28279 we look for cc setter after the compare-elim changes.
28280 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
28281 with the vector to match what compare-elim now expects.
28282 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
28283 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
28284 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
28285 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
28286 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
28287 (neg<mode>2_insn_set_overflow): Likewise.
28289 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
28291 PR middle-end/79794
28292 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
28293 maybe_expand_insn call, set ops[0].target. If still set after call,
28294 set alt_rtl. Add extra arg to recursive calls.
28295 (extract_bit_field): Add alt_rtl argument. Pass to
28297 * expmed.h (extract_bit_field): Fix prototype.
28298 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
28299 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
28300 to extract_bit_field_calls.
28301 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
28302 Pass alt_rtl to extract_bit_field calls.
28303 * calls.c (store_unaligned_arguments_into_psuedos)
28304 load_register_parameters): Pass extra NULL to extract_bit_field calls.
28305 * optabs.c (maybe_legitimize_operand): Clear op->target when call
28307 * optabs.h (struct expand_operand): Add target bitfield.
28309 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28311 * compare-elim.c (try_eliminate_compare): Canonicalize
28312 operation with embedded compare to
28313 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
28314 (set (reg) (operation)].
28316 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
28318 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28321 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
28322 cost of adding a carry flag for ADC instruction.
28323 [case MINUS]: Ignore the cost of subtracting a carry flag
28324 for SBB instruction.
28326 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
28328 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
28330 * config/rs6000/bmiintrin.h: New file.
28331 * config/rs6000/bmi2intrin.h: New file.
28332 * config/rs6000/x86intrin.h: New file.
28334 2017-05-12 Jeff Law <law@redhat.com>
28336 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
28339 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
28341 PR middle-end/80707
28342 * tree-cfg.c: Remove cfg edges of unreachable case statements.
28344 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28346 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28347 early expansion of vector divide builtins.
28348 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
28349 builtins identified as having unsigned arguments.
28351 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28353 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
28354 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
28355 expansion of vector logical operations (and, andc, or, xor,
28358 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28360 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
28361 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
28363 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28365 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28366 early GIMPLE expansion of vector multiplies.
28368 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28370 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
28371 TARGET_HAVE_MOVT conditional.
28372 (movt splitter): Likewise.
28374 2017-05-12 Richard Biener <rguenther@suse.de>
28376 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
28377 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28378 Fold all stmts not inplace.
28380 2017-05-12 Richard Biener <rguenther@suse.de>
28382 PR tree-optimization/80713
28383 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
28384 inserted_exprs bit for not removed stmts.
28386 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
28388 PR middle-end/69921
28389 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
28390 parallelized" attribute for parallelized OpenACC kernels.
28391 * omp-offload.c (execute_oacc_device_lower): Use it.
28393 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
28394 Set "oacc kernels" attribute.
28395 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
28396 parameter. Adjust all users.
28397 (oacc_fn_attrib_kernels_p): Remove function.
28398 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
28399 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
28400 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
28401 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
28402 assert "oacc kernels" attribute is set.
28404 2017-05-11 Carl Love <cel@us.ibm.com>
28406 * config/rs6000/rs6000-c: Add support for built-in functions
28407 vector unsigned char vec_popcnt (vector signed char)
28408 vector unsigned char vec_popcnt (vector unsigned char)
28409 vector unsigned short vec_popcnt (vector signed short)
28410 vector unsigned short vec_popcnt (vector unsigned short)
28411 vector unsigned int vec_popcnt (vector signed int)
28412 vector unsigned int vec_popcnt (vector unsigned int)
28413 vector unsigned long long vec_popcnt (vector signed long long)
28414 vector unsigned long long vec_popcnt (vector unsigned long long)
28415 vector signed long long vec_slo (vector signed long long,
28416 vector signed char)
28417 vector signed long long vec_slo (vector signed long long,
28418 vector unsigned char)
28419 vector unsigned long long vec_slo (vector unsigned long long,
28420 vector signed char)
28421 vector unsigned long long vec_slo (vector unsigned long long,
28422 vector unsigned char)
28423 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
28424 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
28425 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
28426 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
28427 * doc/extend.texi: Update the built-in documentation file for the
28428 new built-in functions.
28430 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
28432 * attribs.h (sorted_attr_string): Move machine independent
28433 functions for target clone support from the i386 port to common
28434 code. Rename ix86_function_versions to common_function_versions.
28435 Rename make_name to make_unique_name.
28436 (common_function_versions): Likewise.
28437 (make_unique_name): Likewise.
28438 (make_dispatcher_decl): Likewise.
28439 (is_function_default_version): Likewise.
28440 * attribs.c (attr_strcmp): Likewise.
28441 (sorted_attr_string): Likewise.
28442 (common_function_versions): Likewise.
28443 (make_unique_name): Likewise.
28444 (make_dispatcher_decl): Likewise.
28445 (is_function_default_version): Likewise.
28446 * config/i386/i386.c (attr_strcmp): Likewise.
28447 (sorted_attr_string): Likewise.
28448 (ix86_function_versions): Likewise.
28449 (make_name): Likewise.
28450 (make_dispatcher_decl): Likewise.
28451 (is_function_default_version): Likewise.
28452 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
28454 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28457 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
28458 Account for direct move costs for vec_construct of integer
28461 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
28464 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
28465 (UNSPEC_STX_ATOMIC): Ditto.
28466 (loaddi_via_sse): New insn.
28467 (storedi_via_sse): Ditto.
28468 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
28469 Update corresponding peephole2 patterns.
28470 (atomic_storedi_fpu): Ditto.
28472 2017-05-11 Julia Koval <julia.koval@intel.com>
28474 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
28475 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
28477 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
28478 (__builtin_ia32_rsqrt14ss_mask): New builtins.
28479 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
28481 2017-05-11 Nathan Sidwell <nathan@acm.org>
28483 * graphite-poly.c: Include dumpfile.h.
28485 * dumpfle.h (dump_function): Declare here ...
28486 * tree-dump.h (dump_function): ... not here.
28487 * dumpfile.c: #include tree-cfg.h.
28488 (dump_function): Move here from ...
28489 * tree-dump.c (dump_function): ... here.
28490 * gimplify.c: #include splay-tree.h, not tree-dump.h.
28491 * graphite-poly.c: Don't include tree-dump.h.
28492 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
28493 * print-tree.c: Likewise.
28494 * stor-layout.c: Likewise.
28495 * tree-nested.c: Likewise.
28497 * dumpfile.c (dump_start): Use TDF_FLAGS.
28498 (dump_enable_all): Fix TDF_KIND check thinko.
28500 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
28502 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
28503 array entries to represent two legal parameterizations of the
28504 overloaded __builtin_cmpb function, as represented by the
28505 P6_OV_BUILTIN_CMPB constant.
28506 (altivec_resolve_overloaded_builtin): Add special case handling
28507 for the __builtin_cmpb function, as represented by the
28508 P6_OV_BUILTIN_CMPB constant.
28509 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
28510 (BU_P6_64BIT_2): New macro.
28511 (BU_P6_OVERLOAD_2): New macro
28512 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
28513 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
28514 (CMPB): Add overload support to represent both 32-bit and 64-bit
28515 compare-bytes function.
28516 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
28517 support for TARGET_CMPB.
28518 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
28519 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28520 documentation of the __builtin_cmpb overloaded built-in function.
28522 2017-05-11 Richard Biener <rguenther@suse.de>
28524 PR tree-optimization/80705
28525 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
28526 bases are not vectorizable.
28528 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28530 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
28531 when counting register pressure.
28533 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28535 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
28536 (iv_ca_more_deps): Renamed to ...
28537 (iv_ca_compare_deps): ... this.
28538 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
28540 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28542 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
28544 (determine_group_iv_costs): ... here.
28545 (find_inv_vars_cb): Record inv var if it's not recorded before.
28547 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28549 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
28550 (get_shiftadd_cost): Ditto.
28552 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28554 * tree-ssa-address.c: Include header file.
28555 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
28557 (add_to_parts): Refactor.
28558 (addr_to_parts): New parameter. Update use of move_hint_to_base.
28559 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
28562 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28564 PR tree-optimization/53090
28565 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
28567 (extract_cond_operands): Detect condition with IV on both sides
28568 and return COMP_IV_EXPR_2.
28569 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
28570 (rewrite_use_compare): Simplify by removing call to function
28571 extract_cond_operands.
28573 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28575 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
28576 (extract_cond_operands): Detect condition comparing against non-
28577 invariant bound and return appropriate enum value.
28578 (find_interesting_uses_cond): Update use of extract_cond_operands.
28579 Handle its return value accordingly.
28580 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
28582 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28584 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
28585 nonlinear iv_use computation in loop invariant sensitive way.
28587 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28589 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
28590 (find_iv_candidates): Call relate_compare_use_with_all_cands.
28592 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28594 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
28595 (dump_cand): Support iv_cand.inv_exprs.
28596 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
28598 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
28601 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28603 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
28605 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
28606 as local function. Include necessary header files.
28607 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
28609 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28611 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
28613 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28615 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
28616 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
28617 RSHIFT_EXPR and BIT_NOT_EXPR.
28619 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28621 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
28622 (adjust_setup_cost): New parameter supporting round up adjustment.
28623 (struct address_cost_data): Delete.
28624 (force_expr_to_var_cost): Don't bound cost with spill_cost.
28625 (split_address_cost, ptr_difference_cost): Delete.
28626 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
28627 (struct ainc_cost_data): New struct.
28628 (get_address_cost_ainc): New function.
28629 (get_address_cost, get_computation_cost): Reimplement.
28630 (determine_group_iv_cost_address): Record inv_expr for all uses of
28632 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
28633 (iv_ca_has_deps): Reimplemented to ...
28634 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
28636 (iv_ca_extend): Call iv_ca_more_deps.
28638 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28640 * tree-ssa-address.c (struct mem_address): Move to header file.
28641 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
28642 * tree-ssa-address.h (struct mem_address): Move from C file.
28643 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
28645 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28647 * tree-affine.h (aff_combination_type): New interface.
28648 (aff_combination_zero_p): Remove static.
28649 (aff_combination_const_p): New interface.
28650 (aff_combination_singleton_var_p): New interfaces.
28652 2017-05-11 Richard Biener <rguenther@suse.de>
28654 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28655 Skip unreachable blocks and destinations.
28656 (eliminate): Move stmt removal and fixup ...
28657 (fini_eliminate): ... here. Skip inserted exprs.
28658 (pass_pre::execute): Move fini_pre after fini_eliminate.
28659 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
28660 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
28661 PRE to get rid of dead code that has invalid SSA form and
28662 split critical edges again.
28664 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28666 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28668 2017-05-11 Richard Biener <rguenther@suse.de>
28670 * passes.c (execute_function_todo): Verify loops if they are
28671 said to be up-to-date.
28672 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
28673 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
28675 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
28678 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
28679 handle calling assemble_external ourself.
28682 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
28683 modes with zero size. Enhance comment.
28685 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28687 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
28688 built-ins for vec_xl and vec_xst with short and char pointer
28691 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
28693 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
28694 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
28695 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
28696 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
28697 (_mm_maskz_min_round_ss): New intrinsics.
28698 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
28699 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
28700 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
28701 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
28702 (__builtin_ia32_minss_mask_round): New builtins.
28703 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28704 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28705 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
28707 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
28708 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
28710 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
28713 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
28715 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
28716 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
28717 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
28718 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
28719 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
28720 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
28721 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
28722 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28723 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
28724 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
28725 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
28726 (__builtin_ia32_mulss_mask_round): New builtins.
28727 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28728 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28729 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
28731 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
28732 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
28734 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
28737 2017-05-10 Julia Koval <julia.koval@intel.com>
28739 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
28740 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
28741 (_mm256_setr_m128i): New intrinsics.
28743 2017-05-10 Julia Koval <julia.koval@intel.com>
28745 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
28746 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
28747 (_mm_maskz_rcp14_ss): New intrinsics.
28748 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
28749 (__builtin_ia32_rcp14ss_mask): New builtins.
28750 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
28752 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
28754 PR tree-optimization/51513
28755 * tree-cfg.c (gimple_seq_unreachable_p): New function.
28756 (assert_unreachable_fallthru_edge_p): Use it.
28757 (group_case_labels_stmt): Likewise.
28758 * tree-cfg.h: Prototype it.
28759 * stmt.c: Include cfghooks.h and tree-cfg.h.
28760 (emit_case_dispatch_table) <gap_label>: New local variable.
28761 Use it to fill dispatch table gaps.
28762 Test for default_label before updating probabilities.
28763 (expand_case) <default_label>: Remove unneeded initialization.
28764 Test for unreachable default case statement and remove its edge.
28765 Set default_label accordingly.
28766 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
28768 2017-05-10 Carl Love <cel@us.ibm.com>
28770 * config/rs6000/rs6000-c: Add support for built-in functions
28771 vector signed char vec_neg (vector signed char)
28772 vector signed short int vec_neg (vector short int)
28773 vector signed int vec_neg (vector signed int)
28774 vector signed long long vec_neg (vector signed long long)
28775 vector float vec_neg (vector float)
28776 vector double vec_neg (vector double)
28777 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
28779 * config/rs6000/altivec.h: Add define for vec_neg
28780 * doc/extend.texi: Update the built-in documentation for the
28781 new built-in functions.
28783 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28785 PR tree-optimization/77644
28786 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
28788 2017-05-10 Nathan Sidwell <nathan@acm.org>
28790 * dumpfile.h (TDI_lang_all): New.
28791 (TDF_KIND): New. Renumber others
28792 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
28794 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
28796 (get_dump_file_name): Adjust suffix generation.
28797 (dump_enable_all): Use TDF_KIND.
28798 * doc/invoke.texi (-fdump-lang-all): Document.
28800 * dumpfile.h: Tabify.
28802 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
28805 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
28806 Move member access before delete.
28808 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
28810 * tree-inline.c (expand_call_inline): Split block at stmt
28813 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
28816 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
28817 are now unused after splitting mov{sf,sd}_hardfloat.
28818 (f32_lr2): Likewise.
28819 (f32_lm): Likewise.
28820 (f32_lm2): Likewise.
28821 (f32_li): Likewise.
28822 (f32_li2): Likewise.
28823 (f32_lv): Likewise.
28824 (f32_sr): Likewise.
28825 (f32_sr2): Likewise.
28826 (f32_sm): Likewise.
28827 (f32_sm2): Likewise.
28828 (f32_si): Likewise.
28829 (f32_si2): Likewise.
28830 (f32_sv): Likewise.
28831 (f32_dm): Likewise.
28832 (f32_vsx): Likewise.
28833 (f32_av): Likewise.
28834 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
28835 For movsf, order stores so the VSX stores occur before the GPR
28836 store which encourages the register allocator to use a traditional
28837 FPR instead of a GPR. For movsd, order the stores so that the GPR
28838 store comes before the VSX stores to allow the power6 to work.
28839 This is due to the power6 not having a 32-bit integer store
28840 instruction from a FPR.
28841 (movsf_hardfloat): Likewise.
28842 (movsd_hardfloat): Likewise.
28844 2017-05-09 Martin Sebor <msebor@redhat.com>
28846 PR translation/80280
28847 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
28850 PR translation/80280
28851 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
28852 data member added in r247778.
28853 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
28855 2017-05-09 Nathan Sidwell <nathan@acm.org>
28857 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
28859 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
28862 2017-05-09 Marek Polacek <polacek@redhat.com>
28864 * doc/invoke.texi: Fix typo.
28866 2017-05-09 Richard Biener <rguenther@suse.de>
28868 * tree-vrp.c (vrp_val_is_max): Adjust comment.
28869 (vrp_val_is_min): Likewise.
28870 (set_value_range_to_value): Likewise.
28871 (set_value_range_to_nonnegative): Likewise.
28872 (gimple_assign_nonzero_p): Likewise.
28873 (gimple_stmt_nonzero_p): Likewise.
28874 (vrp_int_const_binop): Likewise. Remove unreachable case.
28875 (adjust_range_with_scev): Adjust comments.
28876 (compare_range_with_value): Likewise.
28877 (extract_range_from_phi_node): Likewise.
28878 (test_for_singularity): Likewise.
28880 2017-05-09 Richard Biener <rguenther@suse.de>
28882 * tree-vrp.c (get_single_symbol): Add assert that we don't
28883 get overflowed constants as invariant part.
28884 (compare_values_warnv): Add comment before the TREE_NO_WARNING
28885 checks. Use wi::cmp instead of recursing for integer constants.
28886 (compare_values): Just ignore whether we assumed undefined
28887 overflow instead of failing the compare.
28888 (extract_range_for_var_from_comparison_expr): Add comment before the
28889 TREE_NO_WARNING sets.
28890 (test_for_singularity): Likewise.
28891 (extract_range_from_comparison): Do not disable optimization
28892 when we assumed undefined overflow.
28893 (extract_range_basic): Remove init of unused var.
28895 2017-05-09 Richard Biener <rguenther@suse.de>
28897 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
28898 (extract_range_from_multiplicative_op_1): Adjust.
28899 (extract_range_from_binary_expr_1): Use int_const_binop.
28901 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
28904 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
28905 rs6000_store_data_bypass_p in seven define_bypass directives and
28906 in several comments.
28907 * config/rs6000/rs6000-protos.h: Add prototype for
28908 rs6000_store_data_bypass_p function.
28909 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
28910 function implements slightly different (rs6000-specific) semantics
28911 than store_data_bypass_p, returning false rather than aborting
28912 with assertion error when arguments do not satisfy the
28913 requirements of store data bypass.
28914 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
28915 rs6000_store_data_bypass_p.
28917 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
28919 * config/xtensa/xtensa-protos.h
28920 (xtensa_initial_elimination_offset): New declaration.
28921 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28922 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
28923 macro definition, add case for FRAME_POINTER_REGNUM when
28924 FRAME_GROWS_DOWNWARD.
28925 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
28926 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
28927 xtensa_initial_elimination_offset.
28929 2017-05-08 Nathan Sidwell <nathan@acm.org>
28931 * doc/invoke.texi: Alphabetize -fdump options.
28933 2017-05-08 Martin Sebor <msebor@redhat.com>
28935 PR translation/80280
28936 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
28938 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
28940 * target.def (compute_frame_layout): New optional target hook.
28941 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
28942 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
28943 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
28945 * reload1.c (verify_initial_elim_offsets): Likewise.
28946 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
28947 (use_simple_return_p): Call arm_compute_frame_layout if needed.
28948 (arm_get_frame_offsets): Split up into this ...
28949 (arm_compute_frame_layout): ... and this function.
28951 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
28953 * config/aarch64/constraints.md (Usa): New constraint.
28954 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
28956 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
28958 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
28959 with_multilib_list after it has been checked.
28961 2017-05-08 Richard Biener <rguenther@suse.de>
28963 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
28964 (bitmap_set_subtract_values): Likewise.
28966 2017-05-08 Richard Biener <rguenther@suse.de>
28968 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
28969 (gimple_assign_nonzero): ... this and remove strict_overflow_p
28971 (gimple_stmt_nonzero_warnv_p): Rename to ...
28972 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
28974 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
28975 (extract_range_basic): Adjust, do not disable propagation on
28976 strict overflow sensitive simplification.
28977 (vrp_visit_cond_stmt): Likewise.
28979 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
28981 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
28982 body size unconditionally.
28984 2017-05-07 Jeff Law <law@redhat.com>
28987 2017-05-06 Jeff Law <law@redhat.com>
28988 PR tree-optimization/78496
28989 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28992 PR tree-optimization/78496
28993 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28994 (simplify_stmt_using_ranges): Call it.
28995 (vrp_dom_walker::before_dom_children): Extract equivalences
28996 from an ASSERT_EXPR with an equality comparison against a
28999 2017-05-06 Jeff Law <law@redhat.com>
29001 PR tree-optimization/78496
29002 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
29005 PR tree-optimization/78496
29006 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
29007 (simplify_stmt_using_ranges): Call it.
29008 (vrp_dom_walker::before_dom_children): Extract equivalences
29009 from an ASSERT_EXPR with an equality comparison against a
29012 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
29014 * lra-constraints.c (lra_copy_reg_equiv): New function.
29015 (split_reg): Use it to copy equivalence information from the
29016 original register to the spill register.
29018 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
29020 PR rtl-optimization/75964
29021 * simplify-rtx.c (simplify_const_relational_operation): Remove
29022 invalid handling of comparisons of integer ABS.
29024 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
29026 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
29027 initialize to zero.
29028 (init_regs): Remove declaration.
29029 (function_arg_advance_32): Initialize error_p as boolean variable.
29031 2017-05-05 Nathan Sidwell <nathan@acm.org>
29033 * store-motion.c (remove_reachable_equiv_notes): Reformat long
29034 lines. Use for (;;).
29036 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29038 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
29039 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
29040 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
29041 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
29042 VF=2 that require versioning.
29044 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29046 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
29049 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29051 * diagnostic.h (diagnostic_override_option_index): Convert from
29052 macro to inline function.
29054 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29056 * diagnostic.c (last_module_changed_p): New function.
29057 (set_last_module): New function.
29058 (diagnostic_report_current_module): Convert macro usage to
29059 the above functions.
29060 * diagnostic.h (diagnostic_context::last_module): Strengthen
29061 from const line_map * to const line_map_ordinary *.
29062 (diagnostic_last_module_changed): Delete macro.
29063 (diagnostic_set_last_module): Delete macro.
29065 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29067 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
29068 with diagnostic_report_diagnostic.
29069 (diagnostic_n_impl_richloc): Likewise.
29070 * diagnostic.h (report_diagnostic): Delete macro.
29071 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
29072 with diagnostic_report_diagnostic.
29073 * substring-locations.c (format_warning_va): Likewise.
29075 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29077 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
29078 save/restor of format_spec. Move option-printing code to...
29079 (print_option_information): ...this new function, and
29080 reimplement by simply printing to the pretty_printer,
29081 rather than appending to the format string.
29083 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29085 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
29086 handling logic into...
29087 (update_effective_level_from_pragmas): ...this new function.
29089 2017-05-04 Andrew Waterman <andrew@sifive.com>
29091 * config/riscv/riscv.opt (mstrict-align): New option.
29092 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
29093 (SLOW_UNALIGNED_ACCESS): Define.
29094 (riscv_slow_unaligned_access): Declare.
29095 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
29097 (riscv_slow_unaligned_access): New variable.
29098 (rocket_tune_info): Set slow_unaligned_access to true.
29099 (optimize_size_tune_info): Set slow_unaligned_access to false.
29100 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
29101 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
29102 (riscv_option_override): Set riscv_slow_unaligned_access.
29103 * doc/invoke.texi: Add -mstrict-align to RISC-V.
29105 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
29107 * config/riscv/riscv.md: Unify indentation.
29109 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29114 * config/rs6000/rs6000.md (u code attribute): Add FIX and
29116 (extendsi<mode>2): Add support for doing sign extension via
29117 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
29118 don't have ISA 3.0 instructions.
29119 (extendsi<mode>2 splitter): Likewise.
29120 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
29121 generate the normal insns since SImode can now go in vector
29122 registers. Disallow the special UNSPECs needed for previous
29123 machines to hide SImode being used. Add new insns
29124 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
29125 (fix_trunc<mode>si2_stfiwx): Likewise.
29126 (fix_trunc<mode>si2_internal): Likewise.
29127 (fixuns_trunc<mode>si2): Likewise.
29128 (fixuns_trunc<mode>si2_stfiwx): Likewise.
29129 (fctiw<u>z_<mode>_smallint): Likewise.
29130 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
29131 of floating point to 32-bit integer from doing a direct move to
29132 the GPR registers to do a store.
29133 (fctiwz_<mode>): Break long line.
29135 2017-05-05 Bin Cheng <bin.cheng@arm.com>
29137 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
29138 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
29139 (addr_list, addr_offset_valid_p): New.
29140 (split_address_groups): Check offset validity with above function.
29141 (gt-tree-ssa-loop-ivopts.h): Include header file.
29143 2017-05-05 Nathan Sidwell <nathan@acm.org>
29145 * config.gcc (arm*-*-*): Add missing 'fi'.
29147 2017-05-05 Steve Ellcey <sellcey@cavium.com>
29149 * doc/invoke.texi (-fopt-info): Explicitly say order of options
29150 included in -fopt-info does not matter.
29151 * doc/optinfo.texi (-fopt-info): Fix description of default
29152 behavour. Explicitly say order of options included in -fopt-info
29155 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29157 * config.gcc: Allow combinations of aprofile and rmprofile values for
29158 --with-multilib-list.
29159 * config/arm/t-multilib: New file.
29160 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
29161 variables. Remove setting of ISA and floating-point ABI in
29162 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
29163 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
29164 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
29165 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
29167 * config/arm/t-rmprofile: Likewise except for the matches changes.
29168 * doc/install.texi (--with-multilib-list): Document the combination of
29169 aprofile and rmprofile values and warn about pitfalls in doing that.
29171 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29173 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
29174 (movdi_aarch64): Likewise.
29176 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29178 PR tree-optimization/80632
29179 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
29181 (build_arrays): Initialize it for virtual phis.
29182 (fix_phi_nodes): Use it for virtual phis.
29184 PR tree-optimization/80558
29185 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
29186 [x, y] op z into [x op, y op z] for op & or | if conditions
29189 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
29190 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29193 * config/arm/arm.md (use_literal_pool): Remove.
29194 (64-bit immediate split): No longer takes cost into consideration
29195 if arm_disable_literal_pool is enabled.
29196 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
29197 used when arm_disable_literal_pool is enabled.
29198 (arm_max_const_double_inline_cost): Remove use of
29199 arm_disable_literal_pool.
29200 (push_minipool_fix): Add assert.
29201 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
29202 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
29203 (no_literal_pool_sf_immediate): New.
29205 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29207 PR tree-optimization/80613
29208 * tree-ssa-dce.c (propagate_necessity): Remove cases for
29209 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
29211 2017-05-05 Richard Biener <rguenther@suse.de>
29213 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
29215 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
29217 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
29218 of this flag from insn conditions due to removal from r247495.
29220 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29222 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
29224 (arm_early_store_addr_dep_ptr): Likewise.
29225 * config/arm/aarch-common-protos.h
29226 (arm_early_load_addr_dep_ptr): Add prototype.
29227 (arm_early_store_addr_dep_ptr): Likewise.
29228 * config/arm/cortex-a53.md: Add new bypasses.
29230 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29232 * tree.c (next_type_uid): Change type to unsigned.
29233 (type_hash_canon): Decrement back next_type_uid if
29234 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
29235 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
29238 2017-05-04 Martin Sebor <msebor@redhat.com>
29240 * builtins.c: Fix a trivial typo in a comment.
29242 PR middle-end/79234
29243 * builtins.c (check_sizes): Adjust to handle reading past the end.
29244 Avoid printing excessive upper bound of ranges. Use %E to print
29245 tree nodes instead of converting them to %wu.
29246 (expand_builtin_memchr): New function.
29247 (compute_dest_size): Rename...
29248 (compute_objsize): ...to this.
29249 (expand_builtin_memcpy): Adjust.
29250 (expand_builtin_mempcpy): Adjust.
29251 (expand_builtin_strcat): Adjust.
29252 (expand_builtin_strcpy): Adjust.
29253 (check_strncat_sizes): Adjust.
29254 (expand_builtin_strncat): Adjust.
29255 (expand_builtin_strncpy): Adjust and simplify.
29256 (expand_builtin_memset): Adjust.
29257 (expand_builtin_bzero): Adjust.
29258 (expand_builtin_memcmp): Adjust.
29259 (expand_builtin): Handle memcmp.
29260 (maybe_emit_chk_warning): Check strncat just once.
29262 2017-05-04 Martin Sebor <msebor@redhat.com>
29264 PR preprocessor/79214
29265 PR middle-end/79222
29266 PR middle-end/79223
29267 * builtins.c (check_sizes): Add inlining context and issue
29268 warnings even when -Wno-system-headers is set.
29269 (check_strncat_sizes): Same.
29270 (expand_builtin_strncat): Same.
29271 (expand_builtin_memmove): New function.
29272 (expand_builtin_stpncpy): Same.
29273 (expand_builtin): Handle memmove and stpncpy.
29275 2017-05-04 Bin Cheng <bin.cheng@arm.com>
29277 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
29278 which is not used any more.
29280 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29282 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
29284 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29286 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
29287 (cortexa53_tunings): Likewise.
29288 (cortexa57_tunings): Likewise.
29289 (cortexa72_tunings): Likewise.
29290 (cortexa73_tunings): Likewise.
29292 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29294 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
29295 Set loop alignment to 8.
29297 2017-05-04 Martin Sebor <msebor@redhat.com>
29299 PR translation/80280
29300 * builtins.c (expand_builtin_object_size): Add missing quoting to
29301 %D and like directives.
29302 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
29303 (hsa_type_for_tree_type): Same.
29304 (verify_function_arguments): Same.
29305 * symtab.c (symbol_table::change_decl_assembler_name): Same.
29306 * varasm.c (get_section): Same.
29309 2017-05-04 Martin Sebor <msebor@redhat.com>
29311 PR translation/80280
29312 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
29314 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29316 * config/aarch64/aarch64.c (generic_addrcost_table):
29317 Change HI/TI mode setting.
29319 2017-05-04 Martin Jambor <mjambor@suse.cz>
29321 PR tree-optimization/80622
29322 * tree-sra.c (comes_initialized_p): New function.
29323 (build_accesses_from_assign): Only set write lazily when
29324 comes_initialized_p is false.
29325 (analyze_access_subtree): Use comes_initialized_p.
29326 (propagate_subaccesses_across_link): Assert !comes_initialized_p
29327 instead of testing for PARM_DECL.
29329 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29331 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
29332 constraint on operand 0 to allow more general addressing modes.
29333 Adjust output template.
29334 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
29336 * config/aarch64/aarch64-protos.h
29337 (aarch64_address_valid_for_prefetch_p): Declare prototype.
29338 * config/aarch64/constraints.md (Dp): New address constraint.
29339 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
29342 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
29344 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
29345 update use of estimate_ipcp_clone_size_and_time.
29346 (estimate_local_effects): Update use of
29347 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
29348 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
29349 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
29350 Return nonspecialized time.
29352 2017-05-04 Richard Biener <rguenther@suse.de>
29354 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
29355 for the last VUSE which def dominates the PHI. Directly call
29357 (get_continuation_for_phi_1): Remove.
29359 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
29361 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
29362 to explain the use of truncating division. Cap the number of
29363 iterations to the maximum given by nb_iterations_upper_bound,
29366 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
29368 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
29369 * configure: Regenerate.
29370 * config.in: Regenerate.
29371 * config/i386/driver-mingw32.c: new file.
29372 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
29373 * config.host: Link driver-mingw32.o on MinGW host.
29374 * doc/install.texi: Document new --enable-mingw-wildcard configure
29377 2017-05-04 Marek Polacek <polacek@redhat.com>
29379 PR tree-optimization/80612
29380 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
29382 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29383 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
29385 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
29386 (movt splitter): Likewise.
29387 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
29388 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
29389 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
29390 block for Thumb-1 with MOVT.
29391 (thumb2_legitimate_address_p): Move code block ...
29392 (can_avoid_literal_pool_for_label_p): ... into this new function.
29393 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
29395 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
29396 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
29397 "M-profile targets with the MOVT instruction".
29399 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29401 * config/arm/arm-builtins.c (arm_init_builtins): Rename
29402 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
29403 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
29405 2017-05-04 Martin Liska <mliska@suse.cz>
29407 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
29408 variable cond_code.
29410 2017-05-04 Richard Biener <rguenther@suse.de>
29412 * tree.c (array_at_struct_end_p): Handle arrays at struct
29413 end with flexarrays more conservatively. Refactor and treat
29414 arrays of arrays or aggregates more strict. Fix
29415 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
29416 * tree.c (array_at_struct_end_p): Adjust prototype.
29417 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
29418 * gimple-fold.c (get_range_strlen): Likewise.
29419 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
29421 2017-05-04 Richard Biener <rguenther@suse.de>
29423 PR tree-optimization/31130
29424 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
29426 (supports_overflow_infinity): Likewise.
29427 (is_negative_overflow_infinity): Likewise.
29428 (is_positive_overflow_infinity): Likewise.
29429 (is_overflow_infinity): Likewise.
29430 (stmt_overflow_infinity): Likewise.
29431 (overflow_infinity_range_p): Likewise.
29432 (usable_range_p): Remove as always returning true.
29433 (make_overflow_infinity): Remove.
29434 (negative_overflow_infinity): Likewise.
29435 (positive_overflow_infinity): Likewise.
29436 (avoid_overflow_infinity): Likewise.
29437 (set_value_range): Adjust accordingly.
29438 (set_value_range_to_nonnegative): Likewise, remove now unused
29439 overflow_infinity arg.
29440 (vrp_operand_equal_p): Adjust.
29441 (update_value_range): Likewise.
29442 (range_int_cst_singleton_p): Likewise.
29443 (operand_less_p): Likewise.
29444 (compare_values_warnv): Likewise.
29445 (extract_range_for_var_from_comparison_expr): Likewise.
29446 (vrp_int_const_binop): Likewise.
29447 (zero_nonzero_bits_from_vr): Likewise.
29448 (extract_range_from_multiplicative_op_1): Likewise.
29449 (extract_range_from_binary_expr_1): Likewise.
29450 (extract_range_from_unary_expr): Likewise.
29451 (extract_range_from_comparison): Likewise.
29452 (extract_range_basic): Likewise.
29453 (adjust_range_with_scev): Likewise.
29454 (compare_ranges): Likewise.
29455 (compare_range_with_value): Likewise.
29456 (dump_value_range): Likewise.
29457 (test_for_singularity): Likewise, remove strict_overflow_p parameter
29459 (simplify_cond_using_ranges): Adjust.
29461 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
29463 * brig-builtins.def: Added a builtin for class_f64.
29464 * builtin-types.def: Added a builtin type needed by class_f64.
29466 2017-05-03 Jason Merrill <jason@redhat.com>
29468 * timevar.def: Add TV_CONSTEXPR.
29470 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29472 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
29474 2017-05-03 Martin Jambor <mjambor@suse.cz>
29476 * ipa-prop.c (ipa_update_after_lto_read): Removed.
29477 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
29478 * ipa-cp.c (ipcp_propagate_stage): Do not call
29479 ipa_update_after_lto_read.
29480 * ipa-inline.c (ipa_inline): Likewise.
29482 2017-05-03 Martin Jambor <mjambor@suse.cz>
29484 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
29485 tag. Added a default constructor and a destructor.
29486 (ipa_edge_args_sum_t): New class;
29487 (ipa_edge_args_sum): Declare.
29488 (ipa_edge_args_vector): Remove declaration.
29489 (IPA_EDGE_REF): Use ipa_edge_args_sum.
29490 (ipa_free_edge_args_substructures): Remove declaration.
29491 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
29492 (ipa_edge_args_info_available_for_edge_p): Likewise.
29493 * ipa-prop.c (ipa_edge_args_vector): Removed.
29494 (edge_removal_hook_holder): Likewise.
29495 (edge_duplication_hook_holder): Likewise.
29496 (ipa_edge_args_sum): New variable.
29497 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
29498 ipa_edge_args_vector.
29499 (ipa_free_edge_args_substructures): Likewise.
29500 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
29501 ipa_edge_args_vector.
29502 (ipa_edge_removal_hook): Turned into method
29503 ipa_edge_args_sum_t::remove.
29504 (ipa_edge_duplication_hook): Turned into method
29505 ipa_edge_args_sum_t::duplicate.
29506 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
29507 registering edge hooks.
29508 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
29509 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
29510 ipa_edge_args_sum instead of ipa_edge_args_vector.
29511 * ipa-profile.c (ipa_profile): Likewise.
29513 2017-05-03 Martin Jambor <mjambor@suse.cz>
29515 * symbol-summary.h (function_summary): New method exists.
29516 (function_summary::symtab_removal): Deallocate through release.
29517 (call_summary): New class.
29518 (gt_ggc_mx): New overload.
29519 (gt_pch_nx): Likewise.
29520 (gt_pch_nx): Likewise.
29522 2017-05-03 Jeff Law <law@redhat.com>
29524 PR tree-optimization/78496
29525 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
29526 from simplify_cond_using_ranges. Split off code to walk
29527 backwards through casts into ...
29528 (simplify_cond_using_ranges_2): New function.
29529 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
29530 (execute_vrp): After identifying jump threads, call
29531 simplify_cond_using_ranges_2.
29533 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
29536 * ipa-inline.h (inline_summary): Add ctor.
29537 (create_ggc): Do not use ggc_cleared_alloc.
29539 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
29540 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29542 * gcc.c (handle_braces): Support escaping in switch matching
29544 * doc/invoke.texi (Spec Files): Document it.
29545 Remove superfluous @code markup in items.
29547 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29549 * diagnostic-show-locus.c (struct column_range): New struct.
29550 (get_affected_columns): New function.
29551 (get_printed_columns): New function.
29552 (struct correction): New struct.
29553 (correction::ensure_capacity): New function.
29554 (correction::ensure_terminated): New function.
29555 (struct line_corrections): New struct.
29556 (line_corrections::~line_corrections): New dtor.
29557 (line_corrections::add_hint): New function.
29558 (layout::print_trailing_fixits): Reimplement in terms of the new
29560 (selftest::test_overlapped_fixit_printing): New function.
29561 (selftest::diagnostic_show_locus_c_tests): Call it.
29563 2017-05-03 Nathan Sidwell <nathan@acm.org>
29565 Canonicalize canonical type hashing
29566 * tree.h (type_hash_canon_hash): Declare.
29567 * tree.c (type_hash_list, attribute_hash_list): Move into
29568 type_hash_canon_hash.
29569 (build_type_attribute_qual_variant): Break out hash code calc into
29570 type_hash_canon_hash.
29571 (type_hash_canon_hash): New. Generic type hash computation.
29572 (build_range_type_1, build_array_type_1, build_function_type,
29573 build_method_type_directly, build_offset_type, build_complex_type,
29574 make_vector_type): Call it.
29576 2017-05-03 Richard Biener <rguenther@suse.de>
29578 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
29579 When all DRs have unknown misaligned do not always peel
29580 when there is a store but apply the same costing model as if
29581 there were only loads.
29583 2017-05-03 Richard Biener <rguenther@suse.de>
29586 PR tree-optimization/80492
29587 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
29588 compare_base_decls returning dont-know properly.
29590 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
29592 * config/arm/iterators.md (CCSI): New mode iterator.
29593 (arch): New mode attribute.
29594 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
29595 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
29596 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
29597 code iterator for success result mode.
29598 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
29599 the corresponding new insn generators.
29601 2017-05-03 Bin Cheng <bin.cheng@arm.com>
29604 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29605 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29607 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
29609 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
29610 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
29611 (DDR_A): Wrap DDR argument in brackets.
29612 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
29613 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
29614 (DDR_REVERSED_P): Likewise.
29616 2017-05-03 Jakub Jelinek <jakub@redhat.com>
29618 PR tree-optimization/79472
29619 * tree-switch-conversion.c (struct switch_conv_info): Add
29620 contiguous_range and default_case_nonstandard fields.
29621 (collect_switch_conv_info): Compute contiguous_range and
29622 default_case_nonstandard fields, don't clear final_bb if
29623 contiguous_range and only the default case doesn't have the required
29625 (check_all_empty_except_final): Set default_case_nonstandard instead
29626 of failing if contiguous_range and the default case doesn't have empty
29628 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
29629 and only the default case doesn't have the required constants. Skip
29631 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
29632 if default_case_nonstandard.
29633 (build_constructors): Build constant 1 just once. Assert that default
29634 values aren't inserted in between cases if contiguous_range. Skip
29636 (build_arrays): Skip virtual phis.
29637 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
29638 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
29639 Handle virtual phis.
29640 (gen_inbound_check): Handle default_case_nonstandard case.
29641 (process_switch): Adjust check_final_bb caller. Call
29642 gather_default_values with the first non-default case instead of
29643 default case if default_case_nonstandard.
29645 2017-05-02 Nathan Sidwell <nathan@acm.org>
29647 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
29648 check. Fix formatting.
29650 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
29652 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
29653 errors when comparing specialized and unspecialized times.
29655 2017-05-02 David Malcolm <dmalcolm@redhat.com>
29657 * diagnostic-show-locus.c
29658 (layout::should_print_annotation_line_p): Make private.
29659 (layout::print_annotation_line): Make private.
29660 (layout::annotation_line_showed_range_p): Make private.
29661 (layout::show_ruler): Make private.
29662 (layout::print_source_line): Make private. Pass in line and
29663 line_width, rather than calling location_get_source_line. Drop
29665 (layout::print_leading_fixits): New method.
29666 (layout::print_any_fixits): Rename to...
29667 (layout::print_trailing_fixits): ...this, and make private.
29668 Don't print newline fixits.
29669 (diagnostic_show_locus): Move logic for printing one row into...
29670 (layout::print_line): ...this new function. Move the
29671 location_get_source_line call and error-handling from
29672 print_source_line to here. Call print_leading_fixits, and rename
29673 print_any_fixits to print_trailing_fixits.
29674 (selftest::test_fixit_insert_containing_newline): Update now that
29675 newlines are partially supported.
29676 (selftest::test_fixit_insert_containing_newline_2): New test.
29677 (selftest::test_fixit_replace_containing_newline): Update comments.
29678 (selftest::diagnostic_show_locus_c_tests): Call the new test.
29679 * edit-context.c (class added_line): New class.
29680 (class edited_line): Describe newline handling in comment.
29681 (edited_line::actually_edited_p): New method.
29682 (edited_line::print_content): Delete redundant decl.
29683 (edited_line::m_predecessors): New field.
29684 (edited_file::print_content): Call edited_line::print_content.
29685 (edited_file::print_diff): Update to support newlines.
29686 (edited_file::print_diff_hunk): Likewise.
29687 (edited_file::print_run_of_changed_lines): New function.
29688 (edited_file::print_diff_line): Convert to...
29689 (print_diff_line): ...this.
29690 (edited_file::get_effective_line_count): New function.
29691 (edited_line::edited_line): Initialize new field m_predecessors.
29692 (edited_line::~edited_line): Clean up m_predecessors.
29693 (edited_line::apply_fixit): Handle newlines.
29694 (edited_line::get_effective_line_count): New function.
29695 (edited_line::print_content): New function.
29696 (edited_line::print_diff_lines): New function.
29697 (selftest::test_applying_fixits_insert_containing_newline): New
29699 (selftest::test_applying_fixits_replace_containing_newline): New
29701 (selftest::insert_line): New function.
29702 (selftest::test_applying_fixits_multiple_lines): Add example of
29704 (selftest::edit_context_c_tests): Call the new tests.
29706 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29708 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
29709 parameter cand. Update dump information.
29710 (get_computation_cost): Update uses.
29712 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29714 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
29715 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
29716 (get_computation_at, rewrite_use_address): Update use of
29717 get_computation_aff.
29719 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29721 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
29722 (get_computation): Delete.
29723 (get_computation_cost): Implement like get_computation_cost_at.
29724 Use get_computation_at.
29725 (get_computation_cost_at): Delete.
29726 (rewrite_use_nonlinear_expr): Use get_computation_at.
29727 (rewrite_use_compare, remove_unused_ivs): Ditto.
29729 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29731 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
29733 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29735 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
29736 (ivopts_global_cost_for_size): Rename parameter and update uses.
29737 (iv_ca_recount_cost): Update uses.
29738 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
29739 candidates seperately in n_invs and n_cands.
29740 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
29742 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29744 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
29745 (find_inv_vars_cb): New.
29746 (find_depends): Renamed to ...
29747 (find_inv_vars): ... this.
29748 (add_candidate_1, force_var_cost): Call find_inv_vars.
29749 (split_address_cost, determine_group_iv_cost_cond): Ditto.
29751 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29753 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
29754 inv_vars. Add inv_exprs.
29755 (struct iv_cand): Rename depends_on to inv_vars.
29756 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
29757 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
29758 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
29759 (dump_cand): Dump inv_vars.
29760 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
29761 (record_invariant, find_depends, add_candidate_1): Ditto.
29762 (set_group_iv_cost, force_var_cost): Ditto.
29763 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
29764 (get_computation_cost_at, get_computation_cost): Ditto.
29765 (determine_group_iv_cost_generic): Ditto.
29766 (determine_group_iv_cost_address): Ditto.
29767 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
29768 (determine_group_iv_costs): Ditto.
29769 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
29770 (iv_ca_set_remove_invariants): Renamed to ...
29771 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
29772 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
29773 (iv_ca_set_add_invariants): Renamed to ...
29774 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
29775 (iv_ca_set_cp): Use iv_ca_set_add_invs.
29776 (iv_ca_has_deps): Support inv_vars and inv_exprs.
29777 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
29778 (create_new_ivs): Remove useless dump.
29780 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29782 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
29784 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
29785 (iv_ca_set_no_cp, create_new_iv): Ditto.
29787 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29789 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29791 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29793 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
29794 function tree_check2.
29796 2017-05-02 Martin Liska <mliska@suse.cz>
29798 * doc/gcov.texi: Add missing preposition.
29799 * gcov.c (function_info::function_info): Properly fill up
29800 all member variables.
29802 2017-05-02 Tamar Christina <tamar.christina@arm.com>
29804 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
29806 2017-05-02 Tamar Christina <tamar.christina@arm.com>
29808 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
29810 2017-05-02 Martin Liska <mliska@suse.cz>
29813 * lto-streamer-in.c (lto_read_tree_1): Remove
29814 LTO_STREAMER_DEBUG.
29815 * lto-streamer.c (struct tree_hash_entry): Likewise.
29816 (struct tree_entry_hasher): Likewise.
29817 (tree_entry_hasher::hash): Likewise.
29818 (tree_entry_hasher::equal): Likewise.
29819 (lto_streamer_init): Likewise.
29820 (lto_orig_address_map): Likewise.
29821 (lto_orig_address_get): Likewise.
29822 (lto_orig_address_remove): Likewise.
29823 * lto-streamer.h: Likewise.
29824 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
29825 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
29827 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
29829 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
29830 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
29831 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
29832 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
29833 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
29834 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
29835 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
29836 (mm_maskz_sub_ss): New intrinsics.
29837 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29838 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
29839 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
29840 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
29841 (__builtin_ia32_subss_mask_round): New builtins.
29842 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29843 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29844 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
29846 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
29847 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
29849 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
29852 2017-05-02 Martin Jambor <mjambor@suse.cz>
29854 PR tree-optimization/78687
29855 * tree-sra.c (access): New field parent.
29856 (process_subtree_disqualification): New function.
29857 (disqualify_candidate): Call it.
29858 (build_accesses_from_assign): Reset write flag if creating an
29860 (build_access_subtree): Fill in parent field and also prpagate
29861 down grp_write flag.
29862 (create_artificial_child_access): New parameter set_grp_write, set
29863 grp_write to its value.
29864 (propagate_subaccesses_across_link): Also propagate grp_write flag
29866 (propagate_all_subaccesses): Push the closest parent back to work
29867 queue if add_access_to_work_queue returned true.
29869 2017-05-02 Richard Biener <rguenther@suse.de>
29871 * common.opt (fstrict-overflow): Alias negative to fwrapv.
29872 * doc/invoke.texi (fstrict-overflow): Remove all traces of
29873 -fstrict-overflow documentation.
29874 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
29875 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
29876 flag_strict_overflow.
29877 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
29878 * lto-opts.c (lto_write_options): Do not stream it.
29879 * lto-wrapper.c (merge_and_complain): Do not handle it.
29880 * opts.c (default_options_table): Do not set -fstrict-overflow.
29881 (finish_options): Likewise do not clear it when sanitizing.
29882 * simplify-rtx.c (simplify_const_relational_operation): Do not
29883 test flag_strict_overflow.
29885 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29887 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
29888 using enabled attribute.
29889 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
29890 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
29891 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
29892 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
29893 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
29894 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
29895 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
29896 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
29897 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
29898 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
29900 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29902 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
29904 2017-05-02 Richard Biener <rguenther@suse.de>
29906 PR tree-optimization/80591
29908 2017-04-10 Richard Biener <rguenther@suse.de>
29910 * tree-ssa-structalias.c (find_func_aliases): Properly handle
29913 2017-05-02 Richard Biener <rguenther@suse.de>
29915 PR tree-optimization/80549
29916 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
29917 (cleanup_tree_cfg_noloop): Create forwarders to known loop
29918 headers if they do not have a preheader.
29920 2017-05-02 Martin Liska <mliska@suse.cz>
29923 * common.opt: Fix typo.
29924 * doc/invoke.texi: Likewise.
29926 2017-05-01 Jan Beulich <jbeulich@suse.com>
29928 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
29929 swapping, add (x,x,m,x,n) alternative.
29931 2017-05-01 Nathan Sidwell <nathan@acm.org>
29933 * calls.c (combine_pending_stack_adjustment_and_call): Remove
29934 unnecessary unadjusted_alignment check.
29936 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29939 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
29941 * gimplify.c (gimplify_cilk_detach): New function.
29942 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
29943 * tree-core.h: Document EXPR_CILK_SPAWN.
29944 * tree.h (EXPR_CILK_SPAWN): Define.
29946 2017-05-01 David Malcolm <dmalcolm@redhat.com>
29948 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
29949 to use new fixit_hint representation, using the "replace" logic.
29950 (get_line_span_for_fixit_hint): Likewise.
29951 (layout::print_any_fixits): Likewise.
29952 (selftest::test_one_liner_many_fixits): Rename to...
29953 (selftest::test_one_liner_many_fixits_1): ...this, and update
29954 comment and expected output to reflect that the multiple fix-it
29955 hints are now consolidated into one insertion.
29956 (selftest::test_one_liner_many_fixits_2): New test.
29957 (selftest::test_diagnostic_show_locus_one_liner): Update for
29959 (selftest::test_fixit_consolidation): Update for fix-it API
29961 * diagnostic.c (print_parseable_fixits): Likewise.
29962 * edit-context.c (edited_line::m_line_events): Convert from
29963 auto_vec <line_event *> to auto_vec <line_event>.
29964 (class line_event): Convert from abstract base class to a concrete
29965 class, taking over the role of replace_event.
29966 (class insert_event): Delete.
29967 (class replace_event): Rename to class line_event. Convert to
29969 (edit_context::add_fixits): Reimplement.
29970 (edit_context::apply_insert): Delete.
29971 (edit_context::apply_replace): Rename to...
29972 (edit_context::apply_fixit): ...this. Convert to half-open range.
29973 (edited_file::apply_insert): Delete.
29974 (edited_file::apply_replace): Rename to...
29975 (edited_file::apply_fixit): ...this.
29976 (edited_line::~edited_line): Drop deletion of events.
29977 (edited_line::apply_insert): Delete.
29978 (edited_line::apply_replace): Rename to...
29979 (edited_line::apply_fixit): ...this. Convert to half-open range.
29980 Update for change to type of m_line_events.
29981 * edit-context.h (edit_context::apply_insert): Delete.
29982 (edit_context::apply_replace): Rename to...
29983 (edit_context::apply_fixit): ...this.
29985 2017-05-01 Martin Sebor <msebor@redhat.com>
29987 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
29990 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
29993 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
29994 __get_cpuid_max returns 0.
29995 (__get_cpuid_count): Ditto.
29997 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
29999 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
30000 replacement expression is another instance of one of its arguments.
30002 2017-05-01 Jakub Jelinek <jakub@redhat.com>
30005 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
30006 check for stack push/pop autoinc.
30007 * config/i386/i386.c (ix86_agi_dependent): Return false
30008 if the only reason why modified_in_p returned true is that
30009 addr is SP based and set_insn is a push or pop.
30011 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
30013 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
30016 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
30019 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
30020 (account_size_time): Use two predicates - exec_pred and
30022 (evaluate_conditions_for_known_args): Compute both clause and
30024 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
30025 (inline_summary_t::duplicate): Update.
30026 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
30028 (compute_inline_parameters): Likewise.
30029 (estimate_edge_size_and_time): Update caluclation of time.
30030 (estimate_node_size_and_time): Compute both time and nonspecialized
30032 (estimate_ipcp_clone_size_and_time): Update.
30033 (inline_merge_summary): Update.
30034 (do_estimate_edge_time): Update.
30035 (do_estimate_edge_size): Update.
30036 (do_estimate_edge_hints): Update.
30037 (inline_read_section, inline_write_summary): Stream both new predicates.
30038 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
30040 (compute_inlined_call_time): Cleanup.
30041 (big_speedup_p): Update.
30042 (edge_badness): Update.
30043 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
30044 (size_time_entry): Replace predicate by exec_predicate and
30045 nonconst_predicate.
30046 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
30047 (estimate_edge_time): Return also nonspec_time.
30048 (reset_edge_growth_cache): Update.
30050 2017-04-29 Jakub Jelinek <jakub@redhat.com>
30052 PR rtl-optimization/80491
30053 * ifcvt.c (noce_process_if_block): When looking for x setter
30054 with missing else_bb, don't check only the insn right before
30055 cond_earliest, but look for the last insn that x is modified in
30056 within the same bb.
30058 PR rtl-optimization/80491
30059 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
30061 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
30063 PR tree-optimization/80487
30064 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
30066 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30068 PR tree-optimization/79697
30069 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
30070 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
30071 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
30073 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
30074 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
30076 2017-04-28 Martin Sebor <msebor@redhat.com>
30078 PR tree-optimization/80523
30079 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
30080 (init_target_to_host_charmap, target_to_host, target_strtol10): New
30082 (maybe_warn, format_directive, parse_directive): Use new functions.
30083 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
30085 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
30087 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
30089 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
30091 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
30092 target_header_dir): Set correctly.
30093 * configure: Regenerated.
30094 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30095 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30096 instead of SYSTEM_HEADER_DIR.
30098 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
30100 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
30101 (estimate_local_effects): Likewise.
30102 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
30103 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
30104 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
30105 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
30106 do_estimate_edge_time, estimate_edge_time): Likewise.
30107 * ipa-inline-analysis.c (estimate_node_size_and_time,
30108 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
30109 (estimate_time_after_inlining): Remove.
30111 2017-04-28 Martin Liska <mliska@suse.cz>
30113 * doc/gcov.texi: Enhance documentation of gcov.
30115 2017-04-28 Martin Liska <mliska@suse.cz>
30117 * doc/gcov.texi: Sort options in alphabetic order.
30118 * doc/gcov-dump.texi: Likewise.
30119 * doc/gcov-tool.texi: Likewise.
30120 * gcov.c (print_usage): Likewise.
30121 * gcov-dump.c (print_usage): Likewise.
30122 * gcov-tool.c (print_merge_usage_message): Likewise.
30123 (print_rewrite_usage_message): Likewise.
30124 (print_overlap_usage_message): Likewise.
30126 2017-04-28 Martin Liska <mliska@suse.cz>
30128 PR gcov-profile/53915
30129 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
30131 2017-04-28 Martin Liska <mliska@suse.cz>
30133 PR gcov-profile/79891
30134 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
30135 is marked by compiler as living on a line.
30136 (get_cycles_count): Remove usage of the union.
30137 (output_intermediate_file): Likewise.
30138 (find_source): Fix GNU coding style.
30139 (accumulate_line_counts): Remove old non-all block mode.
30140 (output_lines): Remove usage of the union.
30141 * profile.c (output_location): Include all BBs, even if
30142 belonging to a same line (and file) as a previous BB.
30144 2017-04-28 Martin Liska <mliska@suse.cz>
30146 * gcov.c (process_args): Handle new argument 'w'.
30147 (read_graph_file): Assign ID to BBs.
30148 (output_branch_count): Display BB # if verbose flag is set.
30149 (output_lines): Likewise for arcs.
30150 (print_usage): Add '--verbose' option help.
30151 * doc/gcov.texi: Document --verbose (-w) option.
30153 2017-04-28 Martin Liska <mliska@suse.cz>
30155 * gcov.c (struct block_location_info): New struct.
30156 (process_file): Fill up the new structure.
30157 (read_graph_file): Replace usage of encoding by the newly added
30159 (add_line_counts): Likewise.
30160 (accumulate_line_counts): Remove usage of the union.
30161 (function_info::function_info): New function.
30162 (function_info::~function_info): Likewise.
30163 (process_file): Call delete instead of release_function.
30164 (release_function): Release the function.
30165 (release_structures): Call delete instead of release_function.
30166 (solve_flow_graph): Replace usage of num_blocks.
30167 (find_exception_blocks): Likewise.
30168 (output_lines): Fix GNU coding style.
30170 2017-04-28 Martin Liska <mliska@suse.cz>
30173 * coverage.c (coverage_remove_note_file): New function.
30174 * coverage.h: Declare the function.
30175 * toplev.c (finalize): Clean if an error has been seen.
30177 2017-04-28 Martin Liska <mliska@suse.cz>
30179 PR gcov-profile/80031
30180 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
30181 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
30182 * gcov.c (read_graph_file): Read just number of blocks.
30183 * profile.c (branch_prob): Do not stream 0 flags per a basic
30186 2017-04-28 Martin Liska <mliska@suse.cz>
30188 * gcov-dump.c (tag_*): Add new argument to declarations.
30189 (dump_gcov_file): Likewise.
30190 (tag_blocks): Add and use new argument depth.
30191 (tag_arcs): Likewise.
30192 (tag_lines): Likewise.
30193 (tag_counters): Likewise.
30194 (tag_summary): Likewise.
30195 (dump_working_sets): Use depth to do a proper indentation.
30197 2017-04-28 Jakub Jelinek <jakub@redhat.com>
30200 * cgraph.h (symtab_node::debug_symtab): No longer inline.
30201 * symtab.c (symtab_node::debug_symtab): Move definition here.
30203 2017-04-28 Richard Biener <rguenther@suse.de>
30205 * lto-streamer.h (LTO_major_version): Bump to 7.
30207 2017-04-28 Richard Biener <rguenther@suse.de>
30209 * tree-vrp.c (assert_info): New struct.
30210 (add_assert_info): New helper.
30211 (register_edge_assert_for_2): Refactor to add asserts to a vector
30213 (register_edge_assert_for_1): Likewise.
30214 (register_edge_assert_for): Likewise.
30215 (finish_register_edge_assert_for): New helper actually registering
30216 asserts where live on edge.
30217 (find_conditional_asserts): Adjust.
30218 (find_switch_asserts): Likewise.
30219 (evrp_dom_walker::try_find_new_range): Generalize.
30220 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
30222 2017-04-27 Marek Polacek <polacek@redhat.com>
30225 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
30226 arg10 and arg11 to itype.
30228 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
30230 * doc/extend.texi (Object Size Checking): Improve grammar.
30232 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
30235 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
30236 that the logic for permitting reciprocal estimates matches that
30239 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30242 * tree.c (type_cache_hasher::equal): Only compare
30243 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
30244 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
30245 non-aggregate element types.
30246 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
30247 about the flag on ARRAY_TYPEs in the comment, formatting fix.
30249 2017-04-27 Richard Biener <rguenther@suse.de>
30251 PR middle-end/80533
30252 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
30253 stripping ARRAY_REFs from MEM_EXPR make sure we're not
30254 keeping a reference to a trailing array.
30256 2017-04-27 Richard Biener <rguenther@suse.de>
30258 PR middle-end/80539
30259 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
30260 being in loop-closed SSA form conservatively.
30261 (chrec_fold_multiply_poly_poly): Likewise.
30263 2017-04-27 Tamar Christina <tamar.christina@arm.com>
30265 PR middle-end/79665
30266 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
30267 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
30269 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30272 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
30273 (aarch64_function_arg_alignment): Return unsigned int again, but still
30274 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
30275 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30276 Don't emit -Wpsabi note.
30277 (aarch64_function_arg_boundary): Likewise.
30278 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30281 2017-04-26 Nathan Sidwell <nathan@acm.org>
30283 * tree.h (crc32_unsigned_n): Declare.
30284 (crc32_unsigned, crc32_unsigned): Make inline.
30285 * tree.c (crc32_unsigned_bits): Replace with ...
30286 (crc32_unsigned_n): ... this.
30287 (crc32_unsigned, crc32_byte): Remove.
30288 (crc32_string): Remove unnecessary braces.
30290 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30292 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
30293 * ipa-inline-analysis.c (MAX_TIME): Remove.
30294 (account_size_time): Use sreal for time.
30295 (dump_inline_summary): Update.
30296 (estimate_function_body_sizes): Update.
30297 (estimate_edge_size_and_time): Update.
30298 (estimate_calls_size_and_time): Update.
30299 (estimate_node_size_and_time): Update.
30300 (inline_merge_summary): Update.
30301 (inline_update_overall_summary): Update.
30302 (estimate_time_after_inlining): Update.
30303 (inline_read_section): Update.
30304 (inline_write_summary): Update.
30305 * ipa-inline.c (compute_uninlined_call_time): Update.
30306 (compute_inlined_call_time): Update.
30307 (recursive_inlining): Update.
30308 (inline_small_functions): Update.
30309 (dump_overall_stats): Update.
30310 * ipa-inline.h: Include sreal.h.
30311 (size_time_entry): Turn time to sreal.
30312 (inline_summary): Turn self_time nad time to sreal.
30314 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30316 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
30318 (sreal::stream_out, sreal::stream_in): New.
30319 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
30321 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30323 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
30326 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
30329 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
30331 (dimode_scalar_chain::compute_convert_gain): Ditto.
30332 (dimode_scalar_chain::make_vector_copies): Ditto.
30333 (dimode_scalar_chain::convert_reg): Ditto.
30334 (dimode_scalar_chain::convert_insn): Ditto.
30335 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
30336 (VI248_AVX512BW_1): New mode iterator.
30337 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
30338 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
30341 2017-04-25 Martin Sebor <msebor@redhat.com>
30343 PR tree-optimization/80497
30344 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
30345 constants are representable in HOST_WIDE_INT.
30346 (parse_directive): Ditto.
30348 2017-04-25 Martin Sebor <msebor@redhat.com>
30351 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
30352 (new_zero_array): Adjust signature.
30353 (dom_info::dom_init): Used unsigned rather that size_t.
30354 (dom_info::dom_info): Same.
30356 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30357 Jakub Jelinek <jakub@redhat.com>
30360 * config/arm/arm.c: Include gimple.h.
30361 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30362 returns negative, increment ncrn only if it returned positive.
30363 (arm_needs_doubleword_align): Return int instead of bool,
30364 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
30365 members, but if there is any such non-FIELD_DECL
30366 > PARM_BOUNDARY aligned decl, return -1 instead of false.
30367 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30368 returns negative, increment nregs only if it returned positive.
30369 (arm_setup_incoming_varargs): Likewise.
30370 (arm_function_arg_boundary): Emit -Wpsabi note if
30371 arm_needs_doubleword_align returns negative, return
30372 DOUBLEWORD_ALIGNMENT only if it returned positive.
30374 2017-04-25 Marek Polacek <polacek@redhat.com>
30377 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30378 first argument to type.
30380 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
30383 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
30384 type checks to test for compatibility instead of equality.
30386 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30387 Jakub Jelinek <jakub@redhat.com>
30390 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
30392 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
30393 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
30394 the alignment computation, but return their maximum in warn_alignment.
30395 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30396 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
30398 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
30399 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30402 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30404 * config/arc/simdext.md (dmpyh): Fix typo.
30406 2017-04-25 Richard Biener <rguenther@suse.de>
30408 PR tree-optimization/80492
30409 * alias.c (compare_base_decls): Handle registers with asm
30410 specification conservatively.
30411 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
30412 compare_base_decls returning dont-know properly.
30414 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30416 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
30417 (legitimate_offset_address_p): New function.
30418 (arc_legitimate_address_p): Use above function.
30420 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30422 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
30424 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30426 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
30427 ACCH registers whenever they are available.
30429 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30431 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
30432 double regs fix when not used.
30434 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30436 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
30438 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
30439 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
30441 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30443 * config/arc/arc.c (arc_output_addsi): Check for h-register class
30444 when emitting short ADD instructions.
30446 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30448 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
30450 (cmpsi_cc_c_insn): Likewise.
30451 (cbranchsi4_scratch): Compute proper instruction length using
30452 compact_hreg_operand.
30453 * config/arc/predicates.md (compact_hreg_operand): New predicate.
30455 2017-04-25 Richard Biener <rguenther@suse.de>
30457 PR middle-end/80509
30458 * passes.c (pass_manager::pass_manager): Initialize
30459 m_name_to_pass_map.
30461 2017-04-25 Richard Biener <rguenther@suse.de>
30463 PR tree-optimization/79201
30464 * tree-ssa-sink.c (statement_sink_location): Handle calls.
30466 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30469 * config/s390/vector.md: Split MEM->GPR vector moves for
30470 non-s_operand addresses.
30472 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30475 * config/s390/predicates.md (reload_const_wide_int_operand): New
30477 * config/s390/s390.md ("movti"): Remove d/P alternative.
30478 ("movti_bigconst"): New pattern definition.
30480 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30483 * s390-protos.h (s390_expand_cs_hqi): Removed.
30484 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
30485 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
30486 modes as well as CCZ1mode and CCZmode.
30487 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
30488 signature of s390_emit_compare_and_swap.
30489 (s390_expand_cs_hqi): Likewise, make static.
30490 (s390_expand_cs_tdsi): Generate an explicit compare before trying
30491 compare-and-swap, in some cases.
30492 (s390_expand_cs): Wrapper function.
30493 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
30495 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
30496 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
30497 patterns for small and large integers. Forbid symref memory operands.
30498 Move expander to s390.c. Require cc register.
30499 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
30500 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
30501 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
30502 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
30503 symref memory operands. Remove CC mode and call s390_match_ccmode
30505 ("atomic_exchange<mode>"): Allow and implement all integer modes.
30507 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30509 * config/s390/s390.md (define_peephole2): New peephole to help
30510 combining the load-and-test pattern with volatile memory.
30512 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30514 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
30515 with CCZmode for TARGET_Z196.
30517 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30519 PR rtl-optimization/80501
30520 * combine.c (make_compound_operation_int): Set subreg_code to SET
30521 even for AND with mask of the sign bit of mode.
30523 PR rtl-optimization/80500
30524 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
30525 sum's initial value.
30527 2017-04-25 Julian Brown <julian@codesourcery.com>
30528 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30530 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
30532 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
30534 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
30536 2017-04-25 Julian Brown <julian@codesourcery.com>
30537 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30539 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
30540 (thunderx2t99_sha): New Reservation.
30542 2017-04-25 Julian Brown <julian@codesourcery.com>
30543 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30545 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
30546 type for 1-element load.
30548 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
30550 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
30552 2017-04-24 Martin Jambor <mjambor@suse.cz>
30554 PR tree-optimization/80293
30555 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
30556 char arrays not totally scalarizable if it is false.
30557 (analyze_all_variable_accesses): Pass correct value in the new
30558 parameter. Add a statistics counter.
30560 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
30562 PR middle-end/79931
30563 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
30565 2017-04-24 Richard Biener <rguenther@suse.de>
30567 PR tree-optimization/80494
30568 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
30569 out for complex types.
30571 2017-04-24 Richard Biener <rguenther@suse.de>
30573 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
30574 * tree-ssa-sccvn.c (print_scc): Print SCC size.
30575 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
30576 (DFS): Adjust and never fail.
30577 (sccvn_dom_walker::fail): Remove.
30578 (sccvn_dom_walker::before_dom_children): Adjust.
30579 (run_scc_vn): Likewise and never fail.
30580 * tree-ssa-pre.c (pass_pre::execute): Adjust.
30581 (pass_fre::execute): Likewise.
30583 2017-04-24 Richard Biener <rguenther@suse.de>
30585 PR tree-optimization/79725
30586 * tree-ssa-sink.c (statement_sink_location): Return whether
30587 failure reason was zero uses. Move that check later.
30588 (sink_code_in_bb): Deal with zero uses by removing the stmt
30591 2017-04-24 Richard Biener <rguenther@suse.de>
30594 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
30595 pointer-based references.
30597 2017-04-24 Richard Biener <rguenther@suse.de>
30600 * pass_manager.h (pass_manager::operator new): Remove.
30601 (pass_manager::operator delete): Likewise.
30602 * passes.c (pass_manager::operator new): Remove.
30603 (pass_manager::operator delete): Likewise.
30604 (pass_manager::pass_manager): Zero individual pass members.
30606 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
30609 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
30610 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
30611 Check "XEXP (src, 1)" operand here.
30612 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
30613 Check "XEXP (src, 1)" operand here.
30614 (dimode_scalar_chain::make_vector_copies): Detect count register
30615 of a shift instruction. Zero extend count register from QImode
30616 to DImode to satisfy vector shift pattern count operand predicate.
30617 Substitute vector shift count operand with a DImode copy.
30618 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
30621 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
30623 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
30624 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
30625 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
30626 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
30627 (UNSPEC_NOREX_MEM): Remove definition.
30629 2017-04-21 Richard Biener <rguenther@suse.de>
30631 PR tree-optimization/79547
30632 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
30633 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
30634 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
30635 without any constraints.
30637 2017-04-21 Richard Biener <rguenther@suse.de>
30639 PR tree-optimization/78847
30640 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
30642 2017-04-21 Richard Biener <rguenther@suse.de>
30644 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
30645 (build_distinct_type_copy): Likewise.
30646 (build_variant_type_copy): Likewise.
30647 * tree.c (build_qualified_type): Pass down mem-stat info.
30648 (build_distinct_type_copy): Likewise.
30649 (build_variant_type_copy): Likewise.
30651 2017-04-21 Richard Biener <rguenther@suse.de>
30653 PR tree-optimization/80237
30654 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
30656 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
30657 for a simplified result.
30659 2016-04-21 Richard Biener <rguenther@suse.de>
30661 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
30662 sth as strict as a simple_iv but a chrec without symbols and an
30663 operand defined in the loop we are peeling (and not some subloop).
30664 (propagate_constants_for_unrolling): Propagate all constants.
30666 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
30669 * config/i386/i386.c (print_reg): Remove assert for disalowed
30670 regno values, call output_operand_lossage instead.
30672 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
30675 * config/i386/constraints.md (Yc): New register constraint.
30676 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
30677 Use Yc constraint for alternative 2 of operand 0. Remove
30678 preferred_for_speed attribute.
30680 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
30682 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
30683 lastprivate clauses in SIMT case.
30685 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
30687 * doc/invoke.texi (-Wextra-semi): Document new warning option.
30689 2017-04-20 Richard Biener <rguenther@suse.de>
30691 PR tree-optimization/57796
30692 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
30693 as N scalar stores.
30694 (vect_model_load_cost): Cost gathers as N scalar loads.
30696 2017-04-20 Richard Biener <rguenther@suse.de>
30698 * ggc-page.c (ggc_allocated_p): Rename to ...
30699 (safe_lookup_page_table_entry): ... this and return the lookup
30701 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
30703 2017-04-20 Richard Biener <rguenther@suse.de>
30705 PR tree-optimization/80453
30706 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
30707 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
30708 from the conditions.
30709 (vn_phi_eq): Pass them down.
30710 (vn_phi_lookup): Record them.
30711 (vn_phi_insert): Likewise.
30713 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
30715 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
30716 uninitialized variable warning to avoid buffer overrun.
30718 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
30721 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
30722 is suppressed for '{ 0 }' in C.
30724 2017-04-20 Jakub Jelinek <jakub@redhat.com>
30726 * BASE-VER: Set to 8.0.0.
30728 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
30730 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
30731 priority .init_array and .fini_array section with SECTION_NOTYPE
30734 2017-04-20 Jakub Jelinek <jakub@redhat.com>
30736 PR middle-end/80423
30737 * tree.h (build_array_type): Add typeless_storage default argument.
30738 * tree.c (type_cache_hasher::equal): Also compare
30739 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
30740 (build_array_type): Add typeless_storage argument, set
30741 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
30743 (build_nonshared_array_type): Adjust build_array_type_1 caller.
30744 (build_array_type): Likewise. Add typeless_storage argument.
30746 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
30747 Jakub Jelinek <jakub@redhat.com>
30749 PR tree-optimization/80426
30750 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
30751 operation on symbolic operands, also compute the overflow for the
30752 invariant part when the operation degenerates into a negation.
30754 2017-04-19 Jakub Jelinek <jakub@redhat.com>
30757 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
30758 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
30761 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
30763 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
30766 * config/avr/avr.c (tree.h): Include it.
30767 (cgraph.h): Include it.
30768 (avr_encode_section_info): Don't warn for uninitialized progmem
30769 variable if it's just an alias.
30771 2017-04-19 Richard Biener <rguenther@suse.de>
30774 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
30775 when needed by AutoPGO.
30777 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
30780 * doc/lto.texi: Remove an extra 'that'.
30782 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
30784 PR rtl-optimization/80429
30785 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
30786 are only used in debug insns.
30788 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
30789 Vladimir Makarov <vmakarov@redhat.com>
30791 * config/sparc/predicates.md (input_operand): Add comment. Return
30792 true for any memory operand when LRA is in progress.
30793 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
30795 2017-04-18 Jeff Law <law@redhat.com>
30798 * mips.md ({return,simple_return}_internal): Do not overwrite
30801 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30803 PR tree-optimization/80443
30804 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
30805 instead of adding 1, subtract -1 and similarly instead of subtracting
30808 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
30810 PR rtl-optimization/80357
30811 * haifa-sched.c (tmp_bitmap): New variable.
30812 (model_recompute): Handle duplicate use records.
30813 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
30814 (free_global_sched_pressure_data): Free it.
30816 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
30819 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
30820 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30821 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30822 instead of SYSTEM_HEADER_DIR.
30824 2017-04-18 Jeff Law <law@redhat.com>
30826 PR middle-end/80422
30827 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
30828 predecessors after walking up the insn chain.
30830 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30833 * dwarf2out.c (modified_type_die): Try harder not to emit internal
30834 sizetype type into debug info.
30836 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30839 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
30840 unneeded test for TARGET_UPPER_REGS_SF.
30841 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
30843 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30846 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
30847 instead of gsi_after_labels.
30849 2017-04-18 Jeff Law <law@redhat.com>
30851 * regcprop.c (maybe_mode_change): Avoid creating copies of the
30855 2017-04-13 Jeff Law <law@redhat.com>
30856 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30857 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30859 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
30862 * config/avr/avr.c (intl.h): Include it.
30863 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
30865 2017-04-18 Martin Liska <mliska@suse.cz>
30867 PR gcov-profile/78783
30868 * gcov-tool.c (gcov_output_files): Validate that destination
30869 file is either removed by the tool or by a user.
30871 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30872 Guy Benyei <guybe@mellanox.com>
30874 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
30875 block, and do not negate it, the stored id is already negative.
30877 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30879 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
30881 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
30884 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
30885 masks of options that should be turned off if the VSX vector
30886 options are turned off.
30887 (OTHER_P8_VECTOR_MASKS): Likewise.
30888 (OTHER_VSX_VECTOR_MASKS): Likewise.
30889 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
30890 rs6000_disable_incompatible_switches to validate no type switches
30892 (rs6000_incompatible_switch): New function to disallow turning on
30893 other vector options if -mno-vsx, -mno-power8-vector, or
30894 -mno-power9-vector are specified.
30896 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30898 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
30900 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30902 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
30903 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
30904 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
30905 (ARG_POINTER_CFA_OFFSET): Likewise.
30907 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30909 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
30910 conditions to take advantage of various optimizations.
30912 2017-04-13 Jeff Law <law@redhat.com>
30914 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30915 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30916 (zero_extendsidi2_dext): Likewise.
30918 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30921 * fold-const.c (fold_ternary_loc): Revert
30922 use op0 instead of fold_convert_loc (loc, type, arg0) part of
30925 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
30927 PR rtl-optimization/80343
30928 * lra-remat.c (update_scratch_ops): Assign original hard reg to
30929 new scratch pseudo.
30931 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
30934 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
30935 to ubsan_encode_value.
30937 2017-04-13 Jeff Law <law@redhat.com>
30939 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
30940 appearing in DEBUG_INSNs.
30942 2017-04-13 Martin Liska <mliska@suse.cz>
30944 PR gcov-profile/80413
30945 * gcov-io.c (gcov_write_string): Copy to buffer just when
30946 allocated size is greater than zero.
30948 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30951 * dwarf2out.c (decls_for_scope): Ignore declarations of
30952 current_function_decl in BLOCK_NONLOCALIZED_VARS.
30954 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
30957 * ipa-visibility.c (non_local_p): Fix typos.
30958 (localize_node): When localizing symbol in same comdat group,
30959 dissolve the group only when we know external symbols are going
30961 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
30963 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30965 PR tree-optimization/79390
30966 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
30967 order does not result in usable sequence, retry with reversed operand
30973 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
30974 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
30975 op0 instead of fold_convert_loc (loc, type, arg0).
30977 2017-04-12 Jeff Law <law@redhat.com>
30979 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
30980 has a delay slot in the generated code.
30982 * config/cris/cris.md (cris_preferred_reload_class): Return
30983 GENNONACR_REGS rather than GENERAL_REGS.
30985 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30988 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
30989 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
30990 signedness of the result type.
30992 2017-04-12 Richard Biener <rguenther@suse.de>
30993 Jeff Law <law@redhat.com>
30995 PR tree-optimization/80359
30996 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
30997 trim stores to TARGET_MEM_REFs.
30999 2017-04-12 Richard Biener <rguenther@suse.de>
31001 PR tree-optimization/79390
31002 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
31003 threading case even more.
31005 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
31008 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
31009 for quad_address_p for TImode, instead of just not indexed_address.
31011 2017-04-12 Richard Biener <rguenther@suse.de>
31012 Bernd Edlinger <bernd.edlinger@hotmail.de>
31014 PR middle-end/79671
31015 * alias.c (component_uses_parent_alias_set_from): Handle
31016 TYPE_TYPELESS_STORAGE.
31017 (get_alias_set): Likewise.
31018 * tree-core.h (tree_type_common): Add typeless_storage flag.
31019 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
31020 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
31021 for types containing members with TYPE_TYPELESS_STORAGE.
31022 (place_field): Likewise.
31023 (layout_type): Likewise for ARRAY_TYPE.
31024 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
31025 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
31026 TYPE_TYPELESS_STORAGE.
31027 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
31029 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31032 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
31033 first argument to type.
31035 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31039 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
31040 CONST0_RTX (mode) rather than const0_rtx where appropriate.
31041 (rs6000_expand_binop_builtin): Likewise.
31042 (rs6000_expand_ternop_builtin): Likewise; also add missing
31043 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
31044 vshasigma built-ins.
31045 * doc/extend.texi: Document that vec_xxpermdi's third argument
31046 must be a constant.
31048 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
31050 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
31051 Use shift_const cost parameter when calculating gain of STV shifts.
31053 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
31055 PR rtl-optimization/70478
31056 * lra-constraints.c (process_alt_operands): Check memory for
31057 disfavoring memory insn operand.
31059 2017-04-11 Jakub Jelinek <jakub@redhat.com>
31061 PR middle-end/80100
31062 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
31063 left shift in unsigned HOST_WIDE_INT type.
31065 PR rtl-optimization/80385
31066 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
31067 (not (neg X)) into (plus X -1) for complex or non-integral modes.
31070 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
31071 if they have any depend clauses.
31073 2017-04-11 Martin Liska <mliska@suse.cz>
31076 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
31077 * ipa-split.c (split_function): Create a local comdat symbol
31078 if caller is in a comdat group.
31080 2017-04-11 Martin Liska <mliska@suse.cz>
31083 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
31086 2017-04-11 Martin Sebor <msebor@redhat.com>
31088 PR middle-end/80364
31089 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
31090 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
31092 (directive::set_width, directive::set_precision, format_character):
31094 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
31097 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
31100 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
31101 conflict, set target->arch_name instead of target->cpu_name.
31103 2017-04-11 Richard Biener <rguenther@suse.de>
31105 PR tree-optimization/80374
31106 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
31107 build_zero_cst, remove fold_convertible_p check again.
31109 2017-04-11 Martin Liska <mliska@suse.cz>
31112 * ubsan.c (instrument_object_size): Do not instrument register
31115 2017-04-11 Jakub Jelinek <jakub@redhat.com>
31118 * config/i386/i386-builtin-types.def
31119 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
31120 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
31121 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
31122 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
31123 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
31124 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
31125 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
31126 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
31127 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
31128 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
31129 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
31130 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
31131 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
31132 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
31133 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
31134 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
31135 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
31136 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
31137 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
31138 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
31139 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
31140 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
31141 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
31142 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
31143 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
31144 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
31145 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
31146 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
31147 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
31148 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
31149 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
31150 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
31151 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
31152 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
31153 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
31154 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
31155 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
31156 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
31157 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
31158 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
31159 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
31160 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
31161 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
31162 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
31163 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
31164 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
31166 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
31167 flag to second_arg_count, handle 4 argument function type _COUNT
31168 aliases, handle second_arg_count on second argument rather than last.
31170 2017-04-10 Jeff Law <law@redhat.com>
31172 PR tree-optimization/80374
31173 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
31174 record anything if we can not convert integer_zero_node to the
31177 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
31180 * config/rs6000/rs6000.c (rs6000_option_override_internal):
31181 Enhance special handling given to the TARGET_P9_MINMAX option in
31182 relation to certain other options.
31184 2017-04-10 Bin Cheng <bin.cheng@arm.com>
31186 PR tree-optimization/80153
31187 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
31188 remove POINTER_PLUS_EXPR's base part directly, rather than through
31191 2017-04-10 Richard Biener <rguenther@suse.de>
31192 Bin Cheng <bin.cheng@arm.com>
31194 PR tree-optimization/80153
31195 * tree-affine.c (aff_combination_to_tree): Get base pointer from
31196 the first element of pointer type aff_tree. Build result expr in
31198 (add_elt_to_tree): Convert to type unconditionally. Remove other
31199 fold_convert calls.
31200 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
31201 (rewrite_use_nonlinear_expr): Check invariant using iv information.
31203 2017-04-10 Richard Biener <rguenther@suse.de>
31205 * tree-ssa-structalias.c (find_func_aliases): Properly handle
31208 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
31210 PR rtl-optimization/70478
31211 * lra-constraints.c (curr_small_class_check): New.
31212 (update_and_check_small_class_inputs): New.
31213 (process_alt_operands): Update curr_small_class_check. Disfavor
31214 alternative insn memory operands. Check available regs for small
31217 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31220 * config/mips/mips.opt (-mvirt): Update description.
31221 * doc/invoke.texi (-mvirt): Likewise.
31223 2017-04-10 Richard Biener <rguenther@suse.de>
31225 PR middle-end/80362
31226 * fold-const.c (fold_binary_loc): Look at unstripped ops when
31227 looking for NEGATE_EXPR in -A / -B to A / B folding.
31229 2017-04-10 Martin Liska <mliska@suse.cz>
31231 PR gcov-profile/80224
31232 * gcov.c (print_usage): Fix usage string.
31233 (get_gcov_intermediate_filename): Remove.
31234 (output_gcov_file): Use both for normal and intermediate format.
31235 (generate_results): Do not initialize special file for
31236 intermediate format.
31238 2017-04-10 Richard Biener <rguenther@suse.de>
31240 PR tree-optimization/80304
31241 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
31244 2017-04-10 Nathan Sidwell <nathan@acm.org>
31247 * config/rs6000/rs6000.c (rs6000_vector_type): New.
31248 (rs6000_init_builtins): Use it.
31250 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31252 * config/arm/arm.md (<mrc>): Add mode to SET source.
31253 (<mrrc>): Likewise.
31255 2017-04-10 Richard Biener <rguenther@suse.de>
31257 PR middle-end/80344
31258 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
31260 2017-04-10 Jakub Jelinek <jakub@redhat.com>
31263 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
31264 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
31265 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
31266 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
31267 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
31268 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
31269 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
31270 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
31271 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
31272 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
31273 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
31274 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
31275 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
31276 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
31277 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
31278 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
31279 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
31280 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
31281 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
31282 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
31283 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
31284 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
31285 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
31287 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
31289 PR rtl-optimization/70478
31290 * lra-constraints.c: Reverse the last patch.
31292 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
31294 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
31295 Add comment for WCHAR_T.
31297 2017-04-08 Martin Liska <mliska@suse.cz>
31300 2017-04-07 Martin Liska <mliska@suse.cz>
31303 * ipa-split.c (split_function): Add function part to a same comdat
31306 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31309 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
31311 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
31313 * rs6000/rs6000.c (vec_load_pendulum): Rename...
31314 (vec_pairing): ...to this.
31315 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
31316 (rs6000_sched_init): Adjust for name change.
31317 (struct rs6000_sched_context): Likewise.
31318 (rs6000_init_sched_context): Likewise.
31319 (rs6000_set_sched_context): Likewise.
31321 2017-04-07 Jakub Jelinek <jakub@redhat.com>
31327 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
31329 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
31330 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
31331 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
31333 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
31335 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
31337 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31339 PR rtl-optimization/70703
31340 * ira-color.c (update_conflict_hard_regno_costs): Use
31341 int64_t instead of HOST_WIDE_INT.
31343 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31345 PR rtl-optimization/70478
31346 * lra-constraints.c (process_alt_operands): Disfavor alternative
31347 insn memory operands.
31349 2017-04-07 Jeff Law <law@redhat.com>
31351 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
31352 CALL and NOTE_INSN_CALL_ARG_LOCATION.
31354 2017-04-07 Martin Liska <mliska@suse.cz>
31357 * config/aarch64/aarch64.c (aarch64_process_target_attr):
31358 Show error message instead of an ICE.
31360 2017-04-07 Martin Liska <mliska@suse.cz>
31363 * ipa-split.c (split_function): Add function part to a same comdat
31366 2017-04-07 Richard Biener <rguenther@suse.de>
31368 PR middle-end/80341
31369 * tree.c (get_unwidened): Also handle ! for_type case for
31371 * convert.c (do_narrow): Split out from ...
31372 (convert_to_integer_1): ... here. Do not pass final truncation
31373 type to get_unwidened for TRUNC_DIV_EXPR.
31375 2017-04-07 Richard Biener <rguenther@suse.de>
31377 * tree-affine.c (wide_int_ext_for_comb): Take type rather
31379 (aff_combination_const): Adjust.
31380 (aff_combination_scale): Likewise.
31381 (aff_combination_add_elt): Likewise.
31382 (aff_combination_add_cst): Likewise.
31383 (aff_combination_convert): Likewise.
31384 (add_elt_to_tree): Likewise. Remove unused argument.
31385 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
31387 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
31389 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
31391 * config/arm/arm.c (arm_default_short_enums): Use
31392 ARM_DEFAULT_SHORT_ENUMS.
31393 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
31395 2017-04-06 Jakub Jelinek <jakub@redhat.com>
31398 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
31399 members with redundant out-of-class redeclaration.
31401 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31404 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
31405 * config/i386/i386.md (*zero_extendsidi2):
31406 Add (?*x,*x) and (?*v,*v) alternatives.
31408 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31411 * config/i386/i386.c (ix86_expand_builtin)
31412 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
31413 mode from insn data. Convert operands to insn operand mode.
31414 Copy operands that don't satisfy insn predicate to a register.
31416 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
31418 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
31421 2017-04-06 Richard Biener <rguenther@suse.de>
31423 PR tree-optimization/80334
31424 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
31425 preserve alignment of accesses.
31427 2017-04-06 Richard Biener <rguenther@suse.de>
31429 PR tree-optimization/80262
31430 * tree-sra.c (build_ref_for_offset): Preserve address-space
31432 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
31433 Drop useless address-space information on MEM_REF offsets.
31435 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
31437 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
31439 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31441 PR rtl-optimization/70703
31442 * ira-color.c (update_conflict_hard_regno_costs): Use
31443 HOST_WIDE_INT instead of long.
31445 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
31448 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
31449 not defined for x86_64 target. Add -mmmx target option when __SSE2__
31451 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
31452 for x86_64 target. Handle -m3dnowa option.
31454 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31456 PR rtl-optimization/70703
31457 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
31458 (update_conflict_hard_regno_costs): Use long instead of unsigned
31459 arithmetic for cost calculation.
31461 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31462 Bernd Edlinger <bernd.edlinger@hotmail.de>
31465 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
31468 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
31471 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
31472 ptr_mode with Pmode throughout.
31473 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
31474 into probe_stack_range and use DImode.
31476 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
31479 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
31480 call_eh_return is true.
31482 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31484 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
31485 Initialize last_match_fntype_index.
31487 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31490 * tree-nvr.c: Include internal-fn.h.
31491 (pass_return_slot::execute): Ignore internal calls without
31494 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31495 Richard Biener <rguenther@suse.de>
31498 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
31499 captures used multiple times, except for the last use.
31500 * generic-match-head.c: Include gimplify.h.
31502 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31504 PR tree-optimization/79390
31505 * target.h (struct noce_if_info): Declare.
31506 * targhooks.h (default_noce_conversion_profitable_p): Declare.
31507 * target.def (noce_conversion_profitable_p): New target hook.
31508 * ifcvt.h (struct noce_if_info): New type, moved from ...
31509 * ifcvt.c (struct noce_if_info): ... here.
31510 (noce_conversion_profitable_p): Renamed to ...
31511 (default_noce_conversion_profitable_p): ... this. No longer
31513 (noce_try_store_flag_constants, noce_try_addcc,
31514 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
31515 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
31516 instead of noce_conversion_profitable_p.
31517 * config/i386/i386.c: Include ifcvt.h.
31518 (ix86_option_override_internal): Don't override
31519 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
31520 (ix86_noce_conversion_profitable_p): New function.
31521 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
31522 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
31523 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
31524 * doc/tm.texi: Regenerated.
31526 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31528 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
31531 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31534 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
31535 instructions for small multiply cores.
31537 2017-04-04 Jeff Law <law@redhat.com>
31539 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
31541 (mips_expand_vec_perm_const): Initialize elements in orig_perm
31542 that are not set by the loop over the elements.
31544 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31547 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
31548 int mode, convert_modes it to mode as unsigned, otherwise use
31549 lowpart_subreg to mode rather than SImode.
31550 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
31551 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
31552 Use DImode instead of SImode for the shift count operand.
31553 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
31556 2017-04-04 Richard Biener <rguenther@suse.de>
31558 PR middle-end/80281
31559 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
31560 arithmetic done for the negate or the plus. Simplify.
31561 (A - (-B) -> A + B): Likewise.
31562 * fold-const.c (split_tree): Make sure to not negate pointers.
31564 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
31566 PR rtl-optimization/60818
31567 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
31568 a compare of comparisons with the thing compared if this results
31569 in a different machine mode.
31571 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
31573 * alias.c (base_alias_check): Fix typo in comment.
31574 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
31575 * cgraphunit.c (symbol_table::compile): Likewise.
31576 * collect2.c (maybe_run_lto_and_relink): Likewise.
31577 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
31578 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
31579 * config/avr/avr.c (avr_map_op_t): Likewise.
31580 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
31581 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
31582 * config/epiphany/epiphany.md (movcc): Likewise.
31583 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
31584 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
31586 * config/mips/mips.c (mips_save_restore_reg): Likewise.
31587 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
31588 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
31589 * config/sh/sh.c (sh_rtx_costs): Likewise.
31590 * fold-const.c (fold_truth_andor): Likewise.
31591 * genautomata.c (collapse_flag): Likewise.
31592 * gengtype.h (struct type::u::s): Likewise.
31593 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
31594 * input.c (FORMAT_AMOUNT): Likewise.
31595 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
31596 (known_aggs_to_agg_replacement_list): Likewise.
31597 * ipa-inline-analysis.c: Likewise.
31598 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
31599 * ipa-polymorphic-call.c
31600 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
31601 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
31602 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
31604 * modulo-sched.c (apply_reg_moves): Likewise.
31605 * omp-expand.c (build_omp_regions_1): Likewise.
31606 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
31607 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
31608 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
31609 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
31610 * value-prof.c: Likewise.
31611 * var-tracking.c (val_reset): Likewise.
31613 2017-04-03 Richard Biener <rguenther@suse.de>
31615 PR tree-optimization/80275
31616 * fold-const.c (split_address_to_core_and_offset): Handle
31619 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
31621 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
31622 descriptors is at least equal to that of functions.
31624 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
31626 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
31628 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
31631 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
31632 (mov<IMOD4:mode>): New expander.
31633 (*mov<IMOD4:mode>_internal): New insn and split pattern.
31635 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
31637 PR rtl-optimization/79405
31638 * fwprop.c (propagations_left): New variable.
31639 (forward_propagate_into): Decrement it.
31640 (fwprop_init): Initialize it.
31641 (fw_prop): If the variable has reached zero, stop propagating.
31642 (fwprop_addr): Ditto.
31644 2017-03-31 Jakub Jelinek <jakub@redhat.com>
31647 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
31648 a FUNCTION_DECL, pass it as decl instead of origin to
31651 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
31653 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
31656 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
31659 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
31660 TARGET_VSX_SMALL_INTEGER.
31662 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31664 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
31665 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
31667 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31669 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
31670 extraction from odd-numbered MSA register.
31672 2017-03-31 Jakub Jelinek <jakub@redhat.com>
31674 PR middle-end/80173
31675 * expmed.c (store_bit_field_1): Don't attempt to create
31676 a word subreg out of hard registers wider than word if they
31677 have HARD_REGNO_NREGS of 1 for their mode.
31679 PR middle-end/80163
31680 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
31681 conversions to integer types wider than word and pointer.
31684 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
31685 (rtx_equal_for_cselib_p): Pass 0 to it.
31686 * cselib.c (cselib_hasher::equal): Likewise.
31687 (rtx_equal_for_cselib_1): Add depth argument. If depth
31688 is 128, don't look up VALUE locs and punt. Increment
31689 depth in recursive calls when walking VALUE locs.
31691 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
31693 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
31694 (make_gcov_file_name): Use the canonical path name for generating
31696 (read_line): Fix handling of files with ascii null bytes.
31698 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
31700 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
31701 to initialise a vector register instead
31702 of using a const_int.
31704 2017-03-30 Jakub Jelinek <jakub@redhat.com>
31706 PR translation/80189
31707 * gimplify.c (omp_default_clause): Use %qs instead of %s in
31708 diagnostic messages.
31710 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
31713 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
31714 (dfp_diex_<mode>): Update mode of operand 1.
31715 * doc/extend.texi (dxex, dxexq): Document change to return type.
31716 (diex, diexq): Document change to argument type.
31718 2017-03-30 Martin Jambor <mjambor@suse.cz>
31721 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
31722 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
31723 it reflects the signature changes performed at the callee side.
31724 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
31725 to cgraph_build_function_type_skip_args.
31726 (build_function_decl_skip_args): Adjust call to the above function.
31728 2017-03-30 Jakub Jelinek <jakub@redhat.com>
31731 * config/i386/sse.md
31732 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
31733 register as dest whenever it is a MEM not rtx_equal_p to the
31734 corresponding dup operand, and when forcing into reg move the
31735 reg into the memory afterwards.
31736 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
31737 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
31738 for the force_reg mode.
31739 (avx512vl_vextractf128<mode>): Use register as dest either
31740 always when a MEM, or when it is a MEM not rtx_equal_p to the
31741 corresponding dup operand, or even not when it is a CONST_VECTOR
31742 depending on the mode and lo vs. hi.
31743 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
31745 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
31746 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
31747 Likewise. Require that operands[2] is even.
31748 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
31749 Remove extraneous parens. Require that operands[2] is a multiple
31751 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
31752 operands[0] is a MEM if <mask_applied>, the predicates/constraints
31753 disallow memory then.
31755 2017-03-30 Richard Biener <rguenther@suse.de>
31757 PR tree-optimization/77498
31758 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
31759 to non-constants over backedges.
31761 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
31763 PR rtl-optimization/80233
31764 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
31765 as last_combined_insn. Do not test for BARRIER_P separately.
31767 2017-03-29 Andreas Schwab <schwab@suse.de>
31770 * calls.c (prepare_call_address): Convert funexp to Pmode before
31771 copying to temp reg.
31773 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31775 PR tree-optimization/80158
31776 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
31777 Handle possible future case of more than one alternate
31779 (replace_rhs_if_not_dup): Likewise.
31780 (replace_one_candidate): Likewise.
31782 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
31784 PR rtl-optimization/80193
31785 * ira.c (ira): Do not check allocation for LRA.
31787 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
31789 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
31790 (nvptx_output_simt_exit): Declare.
31791 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
31792 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
31793 (init_softstack_frame): Move initialization of crtl->is_leaf to...
31794 (nvptx_declare_function_name): ...here. Emit declaration of local
31795 memory space buffer for omp_simt_enter insn.
31796 (nvptx_output_unisimt_switch): New.
31797 (nvptx_output_softstack_switch): New.
31798 (nvptx_output_simt_enter): New.
31799 (nvptx_output_simt_exit): New.
31800 * config/nvptx/nvptx.h (struct machine_function): New fields
31801 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
31802 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
31803 (UNSPECV_SIMT_EXIT): Ditto.
31804 (omp_simt_enter_insn): New insn.
31805 (omp_simt_enter): New expansion.
31806 (omp_simt_exit): New insn.
31807 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
31809 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
31810 (expand_GOMP_SIMT_ENTER_ALLOC): New.
31811 (expand_GOMP_SIMT_EXIT): New.
31812 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
31813 (GOMP_SIMT_ENTER_ALLOC): Ditto.
31814 (GOMP_SIMT_EXIT): Ditto.
31815 * target-insns.def (omp_simt_enter): New insn.
31816 (omp_simt_exit): Ditto.
31817 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
31819 (lower_rec_simd_input_clauses): Implement SIMT privatization.
31820 (lower_rec_input_clauses): Likewise.
31821 (lower_lastprivate_clauses): Handle SIMT privatization.
31823 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
31824 (ompdevlow_adjust_simt_enter): New.
31825 (find_simtpriv_var_op): New.
31826 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
31827 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
31829 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
31830 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
31831 (copy_decl_for_dup_finish): Ditto.
31833 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
31835 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
31838 * config/i386/i386.c (ix86_option_override_internal): Always
31839 allow -mpreferred-stack-boundary=3 for 64-bit targets.
31841 2017-03-28 Bin Cheng <bin.cheng@arm.com>
31843 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
31845 2017-03-28 Bin Cheng <bin.cheng@arm.com>
31847 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
31848 mark new edge's irreducible flag accordign to it.
31849 (vect_do_peeling): Check loop preheader edge's irreducible flag
31850 and pass it to function slpeel_add_loop_guard.
31852 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
31854 PR tree-optimization/80218
31855 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
31856 Update block frequencies and counts.
31858 2017-03-28 Richard Biener <rguenther@suse.de>
31860 PR tree-optimization/78644
31861 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
31862 of a simplification result we may not use it at all.
31864 2017-03-28 Richard Biener <rguenther@suse.de>
31867 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
31868 without arguments, generate default definition of a SSA name.
31870 2017-03-28 Richard Biener <rguenther@suse.de>
31872 PR middle-end/80222
31873 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
31874 TYPE_REF_CAN_ALIAS_ALL references.
31875 * fold-const.c (fold_indirect_ref_1): Likewise.
31877 2017-03-28 Martin Liska <mliska@suse.cz>
31880 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
31881 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
31883 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31884 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
31886 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
31887 (EXTRA_SPECS): Define.
31888 (SUBTARGET_EXTRA_SPECS): Likewise.
31889 (SUBTARGET_CPP_SPEC): Likewise.
31890 * config/arc/elf.h (EXTRA_SPECS): Renamed to
31891 SUBTARGET_EXTRA_SPECS.
31892 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
31894 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31896 * config/arc/simdext.md (vst64_insn): Update pattern.
31897 (vld32wh_insn): Likewise.
31898 (vld32wl_insn): Likewise.
31899 (vld64_insn): Likewise.
31900 (vld32_insn): Likewise.
31902 2017-03-28 Marek Polacek <polacek@redhat.com>
31905 * fold-const.c (fold_comparison): Use protected_set_expr_location
31906 instead of SET_EXPR_LOCATION.
31908 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
31910 * tree.c (add_expr): Avoid name lookup warning.
31912 2017-03-27 Jeff Law <law@redhat.com>
31914 PR tree-optimization/80216
31915 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
31916 function name. Limit recursion depth.
31917 (record_temporary_equivalences): Corresponding changes.
31919 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
31921 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
31924 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31927 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
31929 * cfgcleanup.c (reg_note_cfa_p): New array.
31930 (insns_have_identical_cfa_notes): New function.
31931 (old_insns_match_p): Don't cross-jump in between /f
31932 and non-/f instructions. If both i1 and i2 are frame related,
31933 verify all CFA notes, their order and content.
31935 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
31938 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
31939 HImode and SImode with zero extend to DImode to one insn.
31940 (bswap<mode>2_extenddi): Likewise.
31941 (bswapsi2_extenddi): Likewise.
31942 (bswaphi2_extendsi): Likewise.
31943 (bswaphi2): Combine bswap HImode and SImode into one insn.
31944 Separate memory insns from swapping register.
31945 (bswapsi2): Likewise.
31946 (bswap<mode>2): Likewise.
31947 (bswaphi2_internal): Delete, no longer used.
31948 (bswapsi2_internal): Likewise.
31949 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
31950 store, and gpr<-gpr swap insns.
31951 (bswap<mode>2_store): Likewise.
31952 (bswaphi2_reg): Register only splitter, combine with the splitter.
31953 (bswaphi2 splitter): Likewise.
31954 (bswapsi2_reg): Likewise.
31955 (bswapsi2 splitter): Likewise.
31956 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
31957 the insns into load, store, and register/register insns.
31958 (bswapdi2_ldbrx): Likewise.
31959 (bswapdi2_load): Likewise.
31960 (bswapdi2_store): Likewise.
31961 (bswapdi2_reg): Likewise.
31963 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
31965 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
31966 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
31968 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
31971 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
31973 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31974 special handling for target option conflicts between dform
31975 options (-mpower9-dform, -mpower9-dform-vector,
31976 -mpower9-dform-scalar) and -mno-direct-move.
31978 2017-03-27 Richard Biener <rguenther@suse.de>
31980 PR tree-optimization/80181
31981 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
31983 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31985 * config/arc/predicates.md (move_double_src_operand): Replace the
31986 call to move_double_src_operand with a call to address_operand.
31988 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31990 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
31991 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
31992 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
31994 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31996 * config/arc/predicates.md (long_immediate_loadstore_operand):
31997 Consider scaled addresses cases.
31999 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32001 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
32002 restored when in interrupt.
32003 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
32004 doesn't have delay slot.
32006 2017-03-27 Richard Biener <rguenther@suse.de>
32009 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
32010 inlined thunk clones.
32012 2017-03-27 Jakub Jelinek <jakub@redhat.com>
32015 * asan.c (instrument_derefs): Copy over last operand from
32016 original COMPONENT_REF to the new COMPONENT_REF with
32017 DECL_BIT_FIELD_REPRESENTATIVE.
32018 * ubsan.c (instrument_object_size): Likewise.
32020 2017-03-27 Richard Biener <rguenther@suse.de>
32022 PR tree-optimization/80170
32023 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
32024 sure DR/SCEV didnt fold in constants we do not see when looking
32025 at the reference base alignment.
32027 2017-03-27 Richard Biener <rguenther@suse.de>
32029 PR middle-end/80171
32030 * gimple-fold.c (fold_ctor_reference): Properly guard against
32031 NULL return value from canonicalize_constructor_val.
32033 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
32036 * config/i386/i386.c (ix86_expand_builtin)
32037 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
32038 flags reg setting and flags reg using instructions.
32039 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
32040 clobbering instructions to zero extend op2.
32042 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
32044 * doc/install.texi (Configuration) <--with-aix-soname>:
32045 Update link to AIX ld.
32047 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
32049 PR rtl-optimization/80160
32050 PR rtl-optimization/80159
32051 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
32052 reg_alternate_class into account.
32054 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
32057 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
32058 to consider in curr_insn_transform.
32060 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32062 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
32063 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
32064 and emit_mode_inner.
32066 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32068 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
32069 argument to the overloaded builtin variants. Use the new flag to
32070 deprecate certain builtin variants.
32071 * config/s390/s390-builtin-types.def: Add new builtin types.
32072 * config/s390/s390-builtins.h: Support new flags field for
32073 overloaded builtins.
32074 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
32075 (s390_macro_to_expand): Enable vector float data type.
32076 (s390_cpu_cpp_builtins_internal): Indicate support of the new
32077 builtins by incrementing the __VEC__ version number.
32078 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
32080 (s390_resolve_overloaded_builtin): Emit error messages depending
32081 on the builtin flags.
32082 * config/s390/s390.c (s390_expand_builtin): Support additional
32083 flags argument. Change error message to match the messages
32084 emitted in s390-c.c.
32085 * config/s390/s390.md: New UNSPEC_* constants.
32086 (op_type): Add new instruction types.
32087 * config/s390/vecintrin.h: Add new builtins and test data class
32089 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
32090 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
32091 (VEC_INEXACT, VEC_NOINEXACT): New constants.
32092 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
32093 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
32094 ("vec_mergel<mode>"): V_HW -> VEC_HW.
32096 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
32097 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
32098 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
32099 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
32101 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
32102 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
32103 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
32104 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
32106 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
32107 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
32108 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
32109 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
32110 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
32111 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
32112 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
32114 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
32115 ("vec_scatter_element<V_HW_4:mode>_DI")
32116 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
32117 ("vec_fpint<mode>", "vflls")
32118 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
32119 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
32120 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
32121 ("*vec_cmphe<mode>_cc"): ... these.
32123 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
32124 mode constant instead of magic value.
32126 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32128 * config/s390/s390.c (s390_expand_vec_compare): Support other
32129 vector floating point modes than just V2DF.
32130 (s390_expand_vcond): Likewise.
32131 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
32132 (s390_cannot_change_mode_class): Prevent mode changes between TF
32133 and V1TF in vector registers.
32134 * config/s390/s390.md (DF, SF): New mode attributes.
32135 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
32136 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
32137 SFmode support for VRs.
32138 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
32140 (VFT, VF_HW): New mode iterators.
32141 (vw, sdx): New mode attributes.
32142 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
32143 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
32144 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
32145 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
32146 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
32147 also the new vector floating point modes. Renaming to ...
32149 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
32150 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
32151 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
32152 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
32153 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
32154 ("vec_unordered<mode>"): ... these.
32156 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
32157 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
32158 ("*vec_extendv2df"): New insn definitions.
32160 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32162 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
32163 ("mulditi3_2", "*muldi3_sign"): New patterns.
32164 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
32165 rename the pattern definition.
32167 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32169 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
32171 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
32173 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32175 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
32176 instruction if possible.
32177 * config/s390/vector.md (vec_halfnumelts): New mode
32179 ("*vec_vllezlf<mode>"): New pattern.
32181 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32183 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
32184 ("popcountv4si2", "popcountv2di2"): Rename to ...
32185 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
32186 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
32188 ("popcount<mode>2_vxe"): New pattern.
32190 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32192 * common/config/s390/s390-common.c (processor_flags_table): Add
32194 * config.gcc: Add arch12.
32195 * config/s390/driver-native.c (s390_host_detect_local_cpu):
32196 Default to arch12 for unknown CPU model numbers.
32197 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
32198 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
32199 PROCESSOR_max sanity check.
32200 * config/s390/s390-opts.h (enum processor_type): Add
32202 * config/s390/s390.c (processor_table): Add arch12.
32203 (s390_expand_builtin): Add check for B_VXE flag.
32204 (s390_issue_rate): Add PROCESSOR_ARCH12.
32205 (s390_get_sched_attrmask): Likewise.
32206 (s390_get_unit_mask): Likewise.
32207 (s390_sched_score): Enable z13 scheduling for arch12.
32208 (s390_sched_reorder): Likewise.
32209 (s390_sched_variable_issue): Likewise.
32210 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
32212 (s390_tune_attr): Use z13 scheduling also for arch12.
32213 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
32214 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
32215 (TARGET_VXE_P): New macros.
32216 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
32217 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
32218 * config/s390/s390.opt: Add arch12 as processor_type.
32220 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32222 * config/s390/s390.md
32223 ("fixuns_truncdddi2", "fixuns_trunctddi2")
32224 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
32225 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
32227 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
32228 Rename expanders to ...
32230 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
32231 ("fixuns_truncdddi2_emu"): ... these.
32233 ("fixuns_trunc<mode>si2_emu"): New expander.
32235 ("*fixuns_truncdfdi2_z13"): Rename to ...
32236 ("*fixuns_truncdfdi2_vx"): ... this.
32238 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32240 * config/s390/2964.md: Remove the single element vector compare
32241 instructions which are no longer used.
32242 * config/s390/s390.c (s390_select_ccmode): Remove handling of
32244 (s390_canonicalize_comparison): Remove handling of DFmode
32246 (s390_expand_vec_compare_scalar): Remove function.
32247 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
32248 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
32250 ("*cmp<mode>_ccs"): Add wfcdb instruction.
32252 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32254 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
32256 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
32257 will anyway by matched by mov<mode>_64dfp.
32259 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32261 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
32262 vlef/vstef. Add missing operand to vleif.
32264 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32266 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
32267 pair for all vector types with 64 bit elements.
32268 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
32269 * config/s390/vector.md (V_HW_64): ... here.
32270 (V_128_NOSINGLE): New mode iterator.
32271 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
32272 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
32273 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
32274 ("*vec_load_pairv2di"): Change to ...
32275 ("*vec_load_pair<mode>"): ... this one.
32277 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32279 * config/s390/constraints.md: Add comments.
32280 (jKK): Reject element sizes > 8 bytes.
32281 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
32283 * config/s390/s390.md: Add the s_operand checks formerly in
32284 s390_split_ok_p to various splitters where they are still
32286 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
32287 for 128 bit vectors. Plus two splitters.
32289 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32291 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
32294 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32297 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
32298 error if the boundary argument is not constant.
32300 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32302 PR rtl-optimization/80112
32303 * loop-doloop.c (doloop_condition_get): Don't check condition
32304 if cmp isn't SET with IF_THEN_ELSE src.
32306 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32308 PR tree-optimization/80158
32309 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
32310 replacing a candidate statement, also replace it for the
32311 candidate's alternate interpretation.
32312 (replace_rhs_if_not_dup): Likewise.
32313 (replace_one_candidate): Likewise.
32315 2017-03-24 Richard Biener <rguenther@suse.de>
32317 PR tree-optimization/80167
32318 * graphite-isl-ast-to-gimple.c
32319 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
32321 (translate_isl_ast_to_gimple::get_rename): Likewise.
32323 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32325 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
32326 handling of certain combinations of target options, including the
32327 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
32328 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
32330 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32333 * config/arm/arm.md (*load_multiple): Add reload_completed to
32334 matching condition.
32336 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32337 Richard Biener <rguenth@suse.de>
32339 PR tree-optimization/79908
32340 PR tree-optimization/80136
32341 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32342 been cast away, gimplify_and_add suffices.
32344 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
32346 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
32348 2017-03-23 Richard Biener <rguenther@suse.de>
32350 PR tree-optimization/80032
32351 * gimplify.c (gimple_push_cleanup): Forced unconditional
32352 cleanups still have to go to the conditional_cleanups
32355 2017-03-22 Jakub Jelinek <jakub@redhat.com>
32357 PR tree-optimization/80072
32358 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
32360 (next_operand_entry_id): Change type to unsigned int.
32361 (sort_by_operand_rank): Make sure to return the right return value
32362 even if unsigned fields are bigger than INT_MAX.
32363 (struct oecount): Change cnt and id type to unsigned int.
32364 (oecount_hasher::equal): Formatting fix.
32365 (oecount_cmp): Make sure to return the right return value
32366 even if unsigned fields are bigger than INT_MAX.
32367 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
32370 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
32371 TREE_READONLY on result if writing it more than once.
32374 * doc/invoke.texi (-fsanitize=thread): Document that with
32375 -fnon-call-exceptions atomics are not able to throw
32379 * tsan.c: Include tree-eh.h.
32380 (instrument_builtin_call): Call maybe_clean_eh_stmt or
32381 maybe_clean_or_replace_eh_stmt where needed.
32382 (instrument_memory_accesses): Add cfg_changed argument.
32383 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
32384 if it returned true.
32385 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
32387 PR rtl-optimization/63191
32388 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
32389 wrapper function, moved the whole old content into ...
32390 (ix86_delegitimize_address_1): ... this. New inline function.
32391 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
32392 true as last argument instead of ix86_delegitimize_address.
32394 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32396 * config/aarch64/aarch64.c (generic_branch_cost): Copy
32397 cortexa57_branch_cost.
32399 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32401 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
32403 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32406 * doc/md.texi (Constraints): Document wA constraint.
32407 * config/rs6000/constraints.md (wA): New.
32408 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
32409 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
32410 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
32411 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
32413 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
32416 * gimplify.c (is_oacc_declared): New function.
32417 (oacc_default_clause): Use it to set default flags for acc declared
32418 variables inside parallel regions.
32419 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
32420 declared variables.
32421 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
32422 declare attribute to any decl as necessary.
32424 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
32427 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
32428 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
32429 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
32430 (arm_arch_lpae): This.
32431 * config/arm/arm.c (arm_arch7ve): Rename into ...
32432 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
32433 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
32436 2017-03-22 Martin Liska <mliska@suse.cz>
32439 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
32440 error message instead of an ICE.
32442 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32444 * doc/extend.texi (6.11 Additional Floating Types): Revise.
32446 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
32448 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
32450 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32453 2017-03-21 Martin Sebor <msebor@redhat.com>
32455 * doc/extend.texi: Use "cannot" instead of "can't."
32456 * doc/hostconfig.texi: Same.
32457 * doc/install.texi: Same.
32458 * doc/invoke.texi: Same.
32459 * doc/loop.texi: Same.
32460 * doc/md.texi: Same.
32461 * doc/objc.texi: Same.
32462 * doc/rtl.texi: Same.
32463 * doc/tm.texi: Same.
32464 * doc/tm.texi.in: Same.
32465 * doc/trouble.texi: Same.
32467 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
32470 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
32471 (collect_checksum_attributes): Set it.
32472 (die_checksum_ordered): Use it.
32474 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32476 PR tree-optimization/79908
32477 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
32478 change: For a VA_ARG whose LHS has been cast away, use
32479 force_gimple_operand to construct the side effects.
32481 2017-03-21 David Malcolm <dmalcolm@redhat.com>
32483 PR translation/80001
32484 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
32485 more amenable to translation.
32486 (oacc_loop_auto_partitions): Likewise.
32488 2017-03-21 Marek Polacek <polacek@redhat.com>
32489 Martin Sebor <msebor@redhat.com>
32491 PR tree-optimization/80109
32492 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
32493 on INTEGRAL_TYPE_P.
32495 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32496 Segher Boessenkool <segher@kernel.crashing.org>
32499 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
32500 check reg_used_between_p between insn and one of succ or succ2
32501 depending on if succ is artificial insn not inserted into insn
32504 2017-03-21 Martin Liska <mliska@suse.cz>
32506 PR gcov-profile/80081
32507 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
32508 * doc/gcc.texi: Include gcov-dump stuff.
32509 * doc/gcov-dump.texi: New file.
32511 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
32513 PR rtl-optimization/79150
32514 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
32515 conditional jump, if the jump is the last insn of the loop.
32517 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32518 Richard Biener <rguenth@suse.de>
32520 PR tree-optimization/79908
32521 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32522 been cast away, use force_gimple_operand to construct the side
32525 2017-03-21 Martin Liska <mliska@suse.cz>
32527 PR libfortran/79956
32528 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
32531 2017-03-21 Brad Spengler <spender@grsecurity.net>
32534 * plugin.c (htab_hash_plugin): New function.
32535 (add_new_plugin): Use it and adjust.
32536 (parse_plugin_arg_opt): Adjust.
32537 (init_one_plugin): Likewise.
32539 2017-03-21 Richard Biener <rguenther@suse.de>
32541 PR tree-optimization/80032
32542 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
32543 if set force the cleanup to happen unconditionally.
32544 (gimplify_target_expr): Push inserted clobbers with force_uncond
32545 to avoid them being removed by control-dependent DCE.
32547 2017-03-21 Richard Biener <rguenther@suse.de>
32549 PR tree-optimization/80122
32550 * tree-inline.c (copy_bb): Do not expans va-arg packs or
32551 va_arg_pack_len when the inlined call stmt requires pack
32553 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
32555 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32558 * tsan.c (instrument_builtin_call): If the memory model argument
32559 is not a constant, assume it is valid.
32562 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
32565 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
32567 PR rtl-optimization/79910
32568 * combine.c (can_combine_p): Do not allow combining an I0 or I1
32569 if its dest is used by an insn before I2 (other than the combined
32570 insns themselves, which are properly handled already).
32572 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
32575 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
32577 * combine.c (record_used_regs): New static function.
32578 (try_combine): Handle situations where there is an additional
32579 instruction between I2 and I3 which needs to have a LOG_LINK
32583 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
32585 * combine.c (try_combine): Delete redundant i1 test. Call
32586 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
32588 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32591 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
32592 alternatives 13/14.
32594 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32596 PR tree-optimization/80054
32597 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
32598 the optimization if a PHI or any of its arguments is not dominated
32599 by the candidate's basis. Use gphi* rather than gimple* as
32601 (replace_profitable_candidates): Clean up a gimple* variable that
32602 should be a gphi* variable.
32604 2017-03-20 Martin Sebor <msebor@redhat.com>
32607 * doc/extend.texi (attribute constructor): Document present limitation.
32609 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
32612 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
32613 __POWER9_VECTOR__ #ifdef control, change template definition to
32614 use Power9-specific built-in function.
32615 (vec_any_eq): Likewise.
32616 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
32617 to control outcomes from this test.
32618 (vector_ae_<mode>p): For VEC_F modes, likewise.
32620 2017-03-20 Ian Lance Taylor <iant@google.com>
32622 * config/i386/i386.c (ix86_function_regparm): Save an extra
32623 register for -fsplit-stack with DECL_STATIC_CHAIN.
32625 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
32628 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
32629 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
32631 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
32633 * config/riscv/riscv.c (riscv_print_operand): Use "fence
32635 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
32638 2017-03-20 Marek Polacek <polacek@redhat.com>
32641 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
32643 2017-03-20 Richard Biener <rguenther@suse.de>
32645 PR tree-optimization/80113
32646 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
32647 allocate extra SSA name for PHI def.
32648 (add_close_phis_to_outer_loops): Likewise.
32649 (add_close_phis_to_merge_points): Likewise.
32650 (copy_loop_close_phi_args): Likewise.
32651 (copy_cond_phi_nodes): Likewise.
32653 2017-03-20 Martin Liska <mliska@suse.cz>
32655 PR middle-end/79753
32656 * tree-chkp.c (chkp_build_returned_bound): Do not build
32657 returned bounds for a LHS that's not a BOUNDED_P type.
32659 2017-03-20 Martin Liska <mliska@suse.cz>
32663 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
32664 COMPLEX_CST and VECTOR_CST.
32666 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32669 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
32670 target operand. A new splitter adds the clobber statement in case
32671 the target operand is dead anyway.
32673 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
32675 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
32676 to age-old versions of binutils and glibc.
32678 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
32680 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
32682 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
32684 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
32686 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
32688 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
32689 requirement for binutils 2.13.
32691 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
32693 * combine.c (try_combine): Delete redundant i1 test. Call
32694 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
32696 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
32698 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
32699 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
32701 <riscv64-*-elf>: Re-arrange section
32702 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
32703 <riscv32-*-linux>: Likewise.
32704 <riscv64-*-elf>: Likewise
32705 <riscv64-*-linux>: Likewise.
32707 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
32710 * aarch64.opt(verbose-cost-dump): Fix typo.
32712 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
32715 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
32716 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
32718 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
32720 * reload.c (find_reloads): When reloading a nonoffsettable address,
32721 use RELOAD_OTHER for it and its address reloads.
32723 PR rtl-optimization/79910
32724 * combine.c (record_used_regs): New static function.
32725 (try_combine): Handle situations where there is an additional
32726 instruction between I2 and I3 which needs to have a LOG_LINK
32729 2017-03-17 Jeff Law <law@redhat.com>
32731 PR tree-optimization/71437
32732 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
32733 conditional in the hash table first.
32734 (vrp_dom_walker::before_dom_children): Extract condition from
32735 ASSERT_EXPR. Record condition, its inverion and any implied
32736 conditions as well.
32738 2017-03-17 Marek Polacek <polacek@redhat.com>
32739 Markus Trippelsdorf <markus@trippelsdorf.de>
32741 PR tree-optimization/80079
32742 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
32745 2017-03-17 Richard Biener <rguenther@suse.de>
32747 PR middle-end/80075
32748 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
32749 Properly verify the LHS before the RHS possibly claims to be
32751 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
32754 2017-03-17 Martin Jambor <mjambor@suse.cz>
32756 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
32757 (List of -O2 options): Likewise.
32758 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
32761 2017-03-17 Tom de Vries <tom@codesourcery.com>
32763 * gcov-dump.c (print_usage): Print bug_report_url.
32765 2017-03-17 Richard Biener <rguenther@suse.de>
32767 PR middle-end/80050
32768 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
32769 (parser::peek): Likewise.
32771 2017-03-17 Richard Biener <rguenther@suse.de>
32773 PR tree-optimization/80048
32774 * sese.c (free_sese_info): Properly release rename_map and
32775 copied_bb_map elements.
32777 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
32779 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
32780 Add linked-list forward and backlinks. Insert on
32781 construction, remove on destruction.
32782 (class pass_store_merging): Add m_stores_head field.
32783 (pass_store_merging::terminate_and_process_all_chains):
32784 Iterate over m_stores_head list.
32785 (pass_store_merging::terminate_all_aliasing_chains):
32787 (pass_store_merging::execute): Check for debug stmts first.
32788 Push new chains onto the m_stores_head stack.
32790 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
32793 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
32794 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
32795 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
32797 2017-03-16 Jeff Law <law@redhat.com>
32799 PR tree-optimization/71437
32800 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
32801 member function. Implementation moved into after_dom_children
32802 member function and into the threader's thread_outgoing_edges
32804 (dom_opt_dom_walker::after_dom_children): Simplify by moving
32805 some code into new thread_outgoing_edges.
32806 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
32807 definition. Simplify marker handling (do it here). Assume we always
32808 have the available expression and the const/copies tables.
32809 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
32811 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
32812 * tree-vrp.c (equiv_stack): No longer file scoped.
32813 (vrp_dom_walker): New class.
32814 (vrp_dom_walker::before_dom_children): New member function.
32815 (vrp_dom_walker::after_dom_children): Likewise.
32816 (identify_jump_threads): Setup domwalker. Use it rather than
32817 walking edges in a random order by hand. Simplify setup/finalization.
32818 (finalize_jump_threads): Remove.
32819 (vrp_finalize): Do not call identify_jump_threads here.
32820 (execute_vrp): Do it here instead and call thread_through_all_blocks
32823 PR tree-optimization/71437
32824 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
32826 (simplify_stmt_for_jump_threading): Add basic_block argument. All
32828 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
32829 (dom_opt_dom_walker::thread_across_edge): Remove
32830 handle_dominating_asserts argument. All callers changed.
32831 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
32832 changes. Remove calls to lhs_of_dominating_assert. Other
32833 uses of handle_dominating_asserts turn into unconditional code
32834 (simplify_control_stmt_condition_1): Likewise.
32835 (simplify_control_stmt_condition): Likewise.
32836 (thread_through_normal_block, thread_across_edge): Likewise.
32837 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
32838 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
32839 object if it is not an SSA_NAME.
32840 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
32841 before calling into the VRP specific simplifiers.
32842 (identify_jump_threads): Remove handle_dominating_asserts
32845 2017-03-16 Jakub Jelinek <jakub@redhat.com>
32848 * tree-diagnostic.c (default_tree_printer): No longer static.
32849 * tree-diagnostic.h (default_tree_printer): New prototype.
32851 2017-03-16 Tamar Christina <tamar.christina@arm.com>
32853 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
32854 Change ins into fmov.
32856 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32858 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
32859 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
32860 Use h_con constraint for operand 1.
32861 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
32862 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
32864 2017-03-15 Jeff Law <law@redhat.com>
32866 PR tree-optimization/71437
32867 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
32868 (record_temporary_equivalences): Use it.
32870 PR tree-optimization/71437
32871 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
32872 tree-ssa-scopedtables.
32873 (lookup_avail_expr, build_and_record_new_cond): Likewise.
32874 (record_conditions, record_cond, vuse_eq): Likewise.
32875 (record_edge_info): Adjust to API tweak of record_conditions.
32876 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
32877 (record_temporary_equivalences, optimize_stmt): Likewise.
32878 (eliminate_redundant_computations): Likewise.
32879 (record_equivalences_from_stmt): Likewise.
32880 * tree-ssa-scopedtables.c: Include options.h and params.h.
32881 (vuse_eq): New function, moved from tree-ssa-dom.c
32882 (build_and_record_new_cond): Likewise.
32883 (record_conditions): Likewise. Accept vector of conditions rather
32884 than edge_equivalence structure for first argument.
32885 for the first argument.
32886 (avail_exprs_stack::lookup_avail_expr): New member function, moved
32887 from tree-ssa-dom.c.
32888 (avail_exprs_stack::record_cond): Likewise.
32889 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
32890 from tree-ssa-dom.c.
32891 (avail_exprs_stack): Add new member functions lookup_avail_expr
32893 (record_conditions): Declare.
32895 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
32898 * lra-constraints.c (process_alt_operands): Increase reject for
32899 reloading an input/output operand.
32901 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
32904 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
32905 insns to convert from signed/unsigned char/short to IEEE 128-bit
32907 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
32909 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
32912 * config/i386/i386.c (ix86_vector_duplicate_value): Create
32913 subreg of inner mode for values already in registers.
32915 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
32917 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
32918 iteration reg is used after the loop.
32920 2017-03-14 Martin Sebor <msebor@redhat.com>
32922 PR tree-optimization/79800
32923 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
32924 precision in negative-positive range.
32925 (format_floating): Call non-const overload with adjusted precision.
32927 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
32930 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
32933 2017-03-14 Martin Sebor <msebor@redhat.com>
32935 PR middle-end/80020
32936 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
32937 * builtins.def (aligned_alloc): Use it.
32940 * Makefile.in (GTFILES): Add calls.c.
32941 * calls.c: Include "gt-calls.h".
32943 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
32945 PR rtl-optimization/79728
32946 * regs.h (struct target_regs): New field
32947 x_contains_allocatable_regs_of_mode.
32948 (contains_allocatable_regs_of_mode): New macro.
32949 * reginfo.c (init_reg_sets_1): Initialize it, and change
32950 contains_reg_of_mode so it includes global regs as well.
32951 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
32952 rather than contains_regs_of_mode.
32954 2017-03-14 Martin Liska <mliska@suse.cz>
32956 * doc/invoke.texi: Document options that can't be combined with
32957 -fcheck-pointer-bounds.
32959 2017-03-14 Martin Liska <mliska@suse.cz>
32961 PR middle-end/79831
32962 * doc/invoke.texi (-Wchkp): Document the option.
32964 2017-03-14 Martin Liska <mliska@suse.cz>
32966 * Makefile.in: Install gcov-dump.
32968 2017-03-14 Martin Liska <mliska@suse.cz>
32970 * multiple_target.c (expand_target_clones): Bail out for
32971 an invalid attribute.
32973 2017-03-14 Richard Biener <rguenther@suse.de>
32975 * alias.c (struct alias_set_entry): Pack properly.
32976 * cfgloop.h (struct loop): Likewise.
32977 * cse.c (struct set): Likewise.
32978 * ipa-utils.c (struct searchc_env): Likewise.
32979 * loop-invariant.c (struct invariant): Likewise.
32980 * lra-remat.c (struct cand): Likewise.
32981 * recog.c (struct change_t): Likewise.
32982 * rtl.h (struct address_info): Likewise.
32983 * symbol-summary.h (function_summary): Likewise.
32984 * tree-loop-distribution.c (struct partition): Likewise.
32985 * tree-object-size.c (struct object_size_info): Likewise.
32986 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
32987 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
32988 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
32989 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
32990 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
32991 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
32992 (struct _stmt_vec_info): Likewise.
32994 2017-03-14 Martin Liska <mliska@suse.cz>
32997 * multiple_target.c (create_dispatcher_calls): Check that
32998 a target can create a function dispatcher.
33000 2017-03-14 Martin Liska <mliska@suse.cz>
33003 * multiple_target.c (expand_target_clones): Drop local.local
33004 flag for default implementation.
33006 2017-03-14 Richard Biener <rguenther@suse.de>
33008 PR tree-optimization/80030
33009 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
33011 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
33013 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
33014 gcc_fallthrough() instead of __attribute__((fallthrough));
33016 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
33018 * doc/gcc.texi: Remove "up" link to (DIR).
33019 * doc/gccint.texi: Ditto.
33021 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
33023 * doc/install.texi (Specific) <avr>: Remove reference to
33026 2017-03-13 Jeff Law <law@redhat.com>
33028 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
33029 attribute rather than comments.
33031 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
33032 match_scratch operand is highest.
33034 2017-03-13 Martin Liska <mliska@suse.cz>
33036 PR middle-end/78339
33037 * ipa-pure-const.c (warn_function_noreturn): If the declarations
33038 is a CHKP clone, use original declaration.
33040 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33042 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
33043 (arc_conditional_register_usage): Use a different allocation order
33044 when optimizing for size.
33045 * common/config/arc/arc-common.c (arc_option_optimization_table):
33046 Section anchors default on when optimizing for size.
33048 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33050 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
33052 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33054 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
33055 * config/arc/arc.md (cpu_facility): Add cd variant.
33056 (*movqi_insn): Add code density variant.
33057 (*movhi_insn): Likewise.
33058 (*movqi_insn): Likewise.
33059 (*addsi3_mixed): Likewise.
33060 (subsi3_insn): Likewise.
33062 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33064 * config/arc/arc.md (movsi_cond_exec): Update constraint.
33066 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33068 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
33069 expressions with MINUS and UNARY ops.
33071 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33074 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
33076 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
33077 between vec_select and vector argument.
33078 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
33079 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
33080 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
33081 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
33082 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
33083 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
33085 2017-03-13 Richard Biener <rguenther@suse.de>
33088 * params.def (vect-max-peeling-for-alignment): Fix typo.
33090 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33092 * doc/install.texi (Specific) <mips-*-*>: Remove description of
33093 issue that only occurred with binutils below 2.18.
33095 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33097 * doc/install.texi (Specific) <cris-axis-elf>: No longer
33098 refer to binutils 2.11/2.12 minimum.
33100 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33102 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
33103 ftp.kernel.org and simplify binutils requirement.
33105 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
33107 * doc/invoke.texi (Warning Options): Fix spelling of link-time
33109 (Optimize Options): Ditto. Also remove redundancy.
33111 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33113 PR translation/79848
33114 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
33116 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
33117 to G_ to avoid double translation.
33119 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33121 PR translation/79923
33122 * auto-profile.c (get_combined_location): Convert leading
33123 character of diagnostics to lower case and remove trailing period.
33124 (read_profile): Likewise for various diagnostics.
33125 * config/arm/arm.c (arm_option_override): Remove trailing period
33126 from various diagnostics.
33127 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
33128 (msp430_expand_delay_cycles): Likewise.
33130 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33133 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
33134 full command-line argument, rather than just "str".
33135 (aarch64_validate_march): Likewise.
33136 (aarch64_validate_mtune): Likewise.
33138 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
33140 PR rtl-optimization/78911
33141 * lra-assigns.c (must_not_spill_p): New function.
33142 (spill_for): Use it.
33144 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33146 PR tree-optimization/79981
33147 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
33148 ATOMIC_COMPARE_EXCHANGE ifn result.
33149 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
33150 IFN_ATOMIC_COMPARE_EXCHANGE.
33152 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33155 * opts.c (parse_sanitizer_options): Add missing question mark to
33156 "did you mean" message.
33158 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33160 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
33162 (VMULEUH_UNS): Likewise.
33163 (VMULOUB_UNS): Likewise.
33164 (VMULOUH_UNS): Likewise.
33165 * config/rs6000/rs6000.c (builtin_function_type): Remove
33166 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
33168 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33171 * read-rtl-function.c (function_reader::read_rtx_operand): Update
33172 x with result of extra_parsing_for_operand_code_0.
33173 (function_reader::extra_parsing_for_operand_code_0): Convert
33174 return type from void to rtx, returning x. When reading
33175 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
33176 larger size containing struct block_symbol.
33178 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
33180 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
33181 -mfloat128-hardware without -m64.
33183 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
33186 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
33187 entries to the case statement that marks unsigned arguments to
33188 overloaded functions.
33190 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33192 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
33193 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
33195 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
33198 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
33199 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
33201 2017-03-10 Martin Liska <mliska@suse.cz>
33205 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
33206 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
33209 2017-03-10 Olivier Hainque <hainque@adacore.com>
33211 * tree-switch-conversion (array_value_type): Start by resetting
33212 candidate type to it's main variant.
33214 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33216 PR rtl-optimization/79909
33217 * combine.c (try_combine): Use simplify_replace_rtx on individual
33218 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
33219 of the whole CALL_INSN_FUNCTION_USAGE.
33221 PR tree-optimization/79972
33222 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
33223 get_range_info on SSA_NAMEs. Formatting fixes.
33225 2017-03-10 Richard Biener <rguenther@suse.de>
33226 Jakub Jelinek <jakub@redhat.com>
33228 PR tree-optimization/77975
33229 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
33230 edge to be constant.
33231 (get_val_for): For constant x return it. Formatting fix.
33232 (loop_niter_by_eval): Avoid pointless looping if the next iteration
33233 would use the same bases as the current one.
33235 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33237 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
33238 instead of vec_select for V1TImode.
33239 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
33241 (VSX_LE_128): Add V1TI to this mode iterator.
33242 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
33243 (*vsx_le_perm_store_<mode>): Likewise.
33244 (pre-reload splitter for VSX stores): Likewise.
33245 (post-reload splitter for VSX stores): Likewise.
33246 (*vsx_xxpermdi2_le_<mode>): Likewise.
33247 (*vsx_lxvd2x2_le_<mode>): Likewise.
33248 (*vsx_stxvd2x2_le_<mode>): Likewise.
33250 2017-03-09 Michael Eager <eager@eagercon.com>
33252 Correct failures with --enable-checking=yes,rtl.
33254 * config/microblaze/microblaze.c (microblaze_expand_shift):
33255 Replace GET_CODE test with CONST_INT_P and INTVAL test with
33256 test for const0_rtx.
33257 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
33258 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
33260 2017-03-09 Richard Biener <rguenther@suse.de>
33262 PR tree-optimization/79977
33263 * graphite-scop-detection.c (scop_detection::merge_sese):
33264 Handle the case of extra exits to blocks dominating the entry.
33266 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
33268 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
33271 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
33273 PR rtl-optimization/79949
33274 * lra-constraints.c (process_alt_operands): Check memory when
33275 trying to predict a cycle. Print about the overall increase.
33277 2017-03-09 Richard Biener <rguenther@suse.de>
33279 PR middle-end/79971
33280 * gimple-expr.c (useless_type_conversion_p): Preserve
33281 TYPE_SATURATING for fixed-point types.
33283 2017-03-09 Richard Biener <rguenther@suse.de>
33286 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
33287 alignment of BLKmode params.
33289 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33292 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
33293 (VALL_NO_V2Q): Likewise.
33295 * config/aarch64/aarch64-simd.md
33296 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
33298 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
33299 VALL_NO_V2Q mode iterator.
33300 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
33302 2017-03-09 Martin Liska <mliska@suse.cz>
33304 PR tree-optimization/79631
33305 * tree-chkp-opt.c (chkp_is_constant_addr): Call
33306 tree_int_cst_sign_bit just for INTEGER constants.
33308 2017-03-09 Martin Liska <mliska@suse.cz>
33312 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
33315 2017-03-09 Marek Polacek <polacek@redhat.com>
33318 * tree.c (inchash::add_expr): Handle TREE_VEC.
33320 2017-03-09 Martin Liska <mliska@suse.cz>
33323 (chkp_narrow_size_and_offset): New function.
33324 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
33325 (void chkp_parse_bit_field_ref): New function.
33326 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
33327 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
33329 2017-03-09 Martin Liska <mliska@suse.cz>
33332 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
33333 (chkp_find_bounds_1): Remove gcc_unreachable.
33335 2017-03-09 Jakub Jelinek <jakub@redhat.com>
33338 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
33339 BUILT_IN_SYNC*, determine the access type from the size suffix and
33340 always build a MEM_REF with that type. Handle forgotten
33341 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
33344 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
33345 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
33346 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
33347 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
33348 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
33349 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
33350 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
33351 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
33352 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
33353 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
33354 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
33355 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
33356 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
33357 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
33358 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
33359 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
33360 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
33361 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
33362 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
33363 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
33364 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
33365 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
33366 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
33367 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
33368 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
33369 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
33370 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
33371 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
33372 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
33373 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
33374 definitions outside of __OPTIMIZE__ guarded section.
33377 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
33378 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
33379 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
33380 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
33383 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33385 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
33386 ("vfenez<mode>"): Add missing constraints.
33388 2017-03-08 Martin Sebor <msebor@redhat.com>
33391 * config/nds32/nds32.c (nds32_option_override):
33392 Fix misspelled diagnostic.
33394 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33397 * gimplify.c (gimplify_omp_for): Replace index var in outer
33398 taskloop statement with an artificial variable and add
33399 OMP_CLAUSE_PRIVATE clause for it.
33401 2017-03-08 Richard Biener <rguenther@suse.de>
33403 PR tree-optimization/79955
33404 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
33405 for accesses that are completely outside of the variable.
33407 2017-03-08 Andrew Haley <aph@redhat.com>
33409 PR tree-optimization/79943
33410 * tree-ssa-loop-split.c (compute_new_first_bound): When
33411 calculating the new upper bound, (END-BEG) should be added, not
33414 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33416 * config/avr/avr.md (setmemhi): Make sure match_dup
33417 operand number comes before match_scratch.
33419 2017-03-08 Richard Biener <rguenther@suse.de>
33421 PR tree-optimization/79920
33422 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
33423 with ncopies == 1 to ...
33424 (vect_transform_slp_perm_load): ... here. Properly compute
33425 all element loads by iterating VF times over the group. Do
33426 not handle ncopies (computed in a broken way) in
33427 vect_create_mask_and_perm.
33429 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33432 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
33433 is a uniform vector, use uniform_vector_p return value instead of
33434 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
33436 2017-03-07 Marek Polacek <polacek@redhat.com>
33438 PR middle-end/79809
33439 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
33440 (alloca_call_type): Likewise.
33442 2017-03-07 Martin Liska <mliska@suse.cz>
33444 * gcov.c (process_args): Put comment to correct location.
33446 2017-03-07 Martin Liska <mliska@suse.cz>
33448 PR middle-end/68270
33449 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
33450 Use array_at_struct_end_p instead of DECL_CHAIN (field).
33451 (chkp_narrow_bounds_for_field): Likewise.
33452 (chkp_parse_array_and_component_ref): Pass one more argument to
33455 2017-03-07 Richard Biener <rguenther@suse.de>
33457 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
33460 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
33462 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
33463 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
33465 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33468 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
33469 to end of description.
33470 (PARAM_MAX_STORES_TO_MERGE): Likewise.
33472 2017-03-07 Jakub Jelinek <jakub@redhat.com>
33474 PR rtl-optimization/79901
33475 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
33477 (*avx512f_<code><mode>3<mask_name>): ... this.
33478 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
33479 iterator instead of VI8_AVX2_AVX512BW.
33481 PR rtl-optimization/79901
33482 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
33483 min/max expander, expand it using expand_vec_cond_expr.
33486 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
33489 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33492 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
33493 to void * for PCH reasons.
33494 * dwarf2out.c (output_loc_operands, output_die): Cast
33495 v.val_vec.array to unsigned char *.
33497 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
33500 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
33503 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
33505 PR rtl-optimization/79571
33506 * lra-constraints.c (process_alt_operands): Calculate static
33507 reject and subtract it from overall when only addresses will be
33510 2017-03-06 Julia Koval <julia.koval@intel.com>
33513 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
33514 incoming stack boundary to 128 for 64-bit targets.
33516 2017-03-06 Richard Biener <rguenther@suse.de>
33518 PR tree-optimization/79894
33519 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
33520 to NULL after folding it.
33522 2017-03-06 Richard Biener <rguenther@suse.de>
33524 PR tree-optimization/79824
33525 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
33526 check disabling peeling for gaps.
33528 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
33530 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
33531 attributes): Document gettimeofday.
33533 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33535 * config/s390/s390.c (s390_option_override_internal): Set
33536 PARAM_MIN_VECT_LOOP_BOUND
33538 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33540 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
33541 * config/s390/s390.md: Likewise.
33543 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33546 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
33547 (<avx2_avx512>_perm<mode>): Rename to ...
33548 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
33550 (<avx512>_perm<mode>_mask): Rename to ...
33551 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
33553 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
33554 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
33555 instead of VI8F_256_512.
33556 (avx512f_perm<mode>): New define_expand.
33557 (avx512f_perm<mode>_mask): Likewise.
33558 (avx512f_perm<mode>_1<mask_name>): New define_insn.
33559 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
33561 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33563 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
33564 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
33566 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
33568 2017-03-06 Martin Liska <mliska@suse.cz>
33571 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
33572 when having a SSA NAME w/o VAR_DECL assigned to it.
33574 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33576 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
33577 msa_dpsub_<su>_d): Fix MODE for vec_select.
33579 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33581 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
33583 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
33585 2017-03-06 Richard Biener <rguenther@suse.de>
33587 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
33588 * plugin.c (register_plugin_info): Likewise.
33589 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
33591 2017-03-05 Jakub Jelinek <jakub@redhat.com>
33593 * config/i386/sse.md (sse_storehps, sse_storelps,
33594 avx_<castmode><avxsizesuffix>_<castmode>,
33595 avx512f_<castmode><avxsizesuffix>_<castmode>,
33596 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
33597 in condition that at least one operand is not a MEM.
33599 2017-03-03 Jakub Jelinek <jakub@redhat.com>
33601 PR middle-end/79805
33602 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
33603 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
33605 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
33606 gimple_call_nothrow_p flag based on whether original builtin can throw.
33607 If it can, emit following stmts on the fallthrough edge.
33608 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
33609 don't create new bb if inserting just debug stmts on the edge, try to
33610 insert them on the fallthru bb or just reset debug stmts.
33612 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
33615 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
33616 restore recog_data (including the operand rtxes inside it) around
33617 the call to get_insn_template.
33619 2017-03-03 Martin Sebor <msebor@redhat.com>
33621 PR tree-optimization/79699
33622 * context.c (context::~context): Free MPFR caches to avoid
33623 a memory leak on program exit.
33625 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33627 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
33628 Use wide_int::ulow () instead of .elt (0).
33630 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
33632 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
33633 (*pushxf): Limit oF constraint to 32bit targets and add oC
33634 constraint for 64bit targets.
33635 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
33636 (*pushdf): Change rmF constraint to rmC.
33638 2017-03-03 Martin Liska <mliska@suse.cz>
33640 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
33641 Remove unused variable.
33643 2017-03-03 Jakub Jelinek <jakub@redhat.com>
33646 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
33647 is a memory operand, increase num_memory.
33648 (ix86_expand_args_builtin): Likewise.
33650 2017-03-03 Jan Hubicka <jh@suse.cz>
33653 * ipa-devirt.c (maybe_record_node): Properly handle
33654 __cxa_pure_virtual visibility.
33656 2017-03-03 Martin Liska <mliska@suse.cz>
33658 PR tree-optimization/79803
33659 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
33661 (pass_loop_prefetch::execute): Disabled optimization if an
33662 assumption about L1 cache size is not met.
33664 2017-03-03 Martin Liska <mliska@suse.cz>
33666 PR rtl-optimization/79574
33667 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
33668 (hash_scan_set): Likewise.
33669 (dump_hash_table): Likewise.
33670 (hoist_code): Likewise.
33672 2017-03-03 Richard Biener <rguenther@suse.de>
33674 * fixed-value.c (fixed_from_string): Restore use of elt (1)
33675 in place of uhigh ().
33676 (fixed_convert_from_real): Likewise.
33678 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
33681 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
33683 2017-03-03 Richard Biener <rguenther@suse.de>
33685 PR middle-end/79818
33686 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
33687 TYPE_OVERFLOW_UNDEFINED check.
33689 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33691 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
33693 (vector_ae_<mode>_p): Likewise.
33694 (vector_nez_<mode>_p): Likewise.
33695 (vector_ne_v2di_p): Likewise.
33696 (vector_ae_v2di_p): Likewise.
33697 (vector_ne_<mode>_p): Likewise.
33698 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
33700 (vsx_tsqrt<mode>2_fe): Likewise.
33702 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
33705 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
33707 2017-03-02 Jakub Jelinek <jakub@redhat.com>
33709 PR rtl-optimization/79780
33710 * cprop.c (one_cprop_pass): When second and further conditional trap
33711 in a single basic block is turned into an unconditional trap, turn it
33712 into a deleted note to avoid RTL verification failures.
33714 2017-03-02 Richard Biener <rguenther@suse.de>
33716 * fold-const.c (const_binop): Use ulow () instead of elt (0).
33718 2017-03-02 Richard Biener <rguenther@suse.de>
33720 PR tree-optimization/79345
33722 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
33723 param and abort the walk, returning -1 if it is hit.
33724 (walk_aliased_vdefs): Take a limit param and pass it on.
33725 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
33726 defaulting to 0 and return a signed int.
33727 * tree-ssa-uninit.c (struct check_defs_data): New struct.
33728 (check_defs): New helper.
33729 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
33730 about uninitialized memory.
33731 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
33732 bogus uninitialized warning.
33733 (fixed_convert_from_real): Likewise.
33735 2017-03-02 Bin Cheng <bin.cheng@arm.com>
33737 PR tree-optimization/66768
33738 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
33739 iv_use if base object can't be determined.
33741 2017-03-02 Jakub Jelinek <jakub@redhat.com>
33743 PR tree-optimization/79345
33744 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
33745 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
33746 (get_pattern_stats): Initialize it.
33747 * genemit.c (gen_expand): Verify match_scratch numbers come after
33748 match_operand/match_dup numbers.
33749 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
33750 match_scratch numbers.
33751 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
33753 * config/s390/s390.md (trunctdsd2): Likewise.
33755 2017-03-02 Richard Biener <rguenther@suse.de>
33757 * wide-int.h (wide_int_storage::operator=): Implement in terms
33760 2017-03-02 Richard Biener <rguenther@suse.de>
33762 PR tree-optimization/79777
33763 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
33764 the to insert expression to sth existing.
33766 2017-03-01 Martin Sebor <msebor@redhat.com>
33768 PR middle-end/79692
33769 * gimple-ssa-sprintf.c
33770 (directive::known_width_and_precision): New function.
33771 (format_integer): Use it.
33772 (get_mpfr_format_length): Consider the full range of precision
33773 when computing %g output with the # flag. Set the likely byte
33774 count to 3 rather than 1 when precision is indeterminate.
33775 (format_floating): Correct the lower bound of precision.
33777 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33779 * doc/invoke.texi: Document default code model for 64-bit Linux.
33781 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33784 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
33785 udiv rather than div since input pattern is unsigned.
33787 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
33789 * config/i386/i386.c (print_reg): Warn for values of
33790 unsupported size in integer register.
33792 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
33795 * config/rs6000/predicates.md (current_file_function_operand): Do
33796 not allow self calls to be local if the function is replaceable.
33798 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
33801 * config/rs6000/altivec.h (vec_ctz and others): Change the
33802 preprocessor macro that controls conditional compilation from
33803 _ARCH_PWR9 to __POWER9_VECTOR__.
33804 (vec_all_ne): Change parameterization of __altivec_scalar_pred
33805 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
33806 control (instead of _ARCH_PWR9 control) so that template
33807 definition uses power9-specific function.
33808 (vec_any_eq): Likewise.
33809 (vec_all_ne): Change macro definition to use a power9-specific
33810 expansion under #ifdef __POWER9_VECTOR__ control (instead of
33811 _ARCH_PWR9 control).
33812 (vec_any_eq) Likewise.
33813 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
33814 expansion for CMPNEF to remove support for xvcmpnesp instruction.
33815 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
33816 support for xvcmpnedp instruction.
33817 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
33818 macro expansion so that Power9 implementation of vec_all_ne does
33819 not use the AltiVec predicate framework.
33820 (VCMPNEH_P): Likewise.
33821 (VCMPNEW_P): Likewise.
33822 (VCMPNED_P): Likewise.
33823 (VCMPNEFP_P): Likewise.
33824 (VCMPNEDP_P): Likewise.
33825 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
33826 implementation of vec_any_eq to not use AltiVec predicate
33828 (VCMPAEH_P): Likewise.
33829 (VCMPAEW_P): Likewise.
33830 (VCMPAED_P): Likewise.
33831 (VCMPAEFP_P): Likewise.
33832 (VCMPAEDP_P): Likewise.
33833 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
33834 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
33835 not use the AltiVec predicate framework.
33836 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
33837 of vec_any_eq to not use AltiVec predicate framework.
33838 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
33839 support for predefined __POWER9_VECTOR__ macro to indicate that
33840 Power9 instruction selection is enabled.
33841 (altivec_overloaded_builtins): Remove extraneous
33842 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
33843 function argument types RS6000_BTI_bool_V16QI and
33844 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
33845 entry for overloaded function argument types RS6000_BTI_bool_V4SI
33846 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
33847 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
33848 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
33849 Power9 for implementations of vec_cmpne. Change the signature for
33850 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
33851 (representing vec_all_ne) to remove the previously described first
33852 argument of type RS6000_BTI_INTSI, as this was an artifact of
33853 reliance on the AltiVec predicate framework, which is no longer
33854 used in the implementation of these functions. Add
33855 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
33856 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
33857 since, unlike the AltiVec predicate framework implementation, we
33858 do not share function descriptors between vec_alle and vec_anyeq.
33859 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
33860 set of modes that receive special treatment even when
33861 TARGET_P9_VECTOR is true. The special treatment emits code that
33862 does not depend on Power9 instructions.
33863 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
33864 define_expand to not rely on AltiVec predicate framework.
33865 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33867 (vector_ne_v2di_p): Change this define_expand to not rely on
33868 AltiVec predicate framework.
33869 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
33871 (vector_ne_<mode>_p): Change this define_expand to not rely on
33872 AltiVec predicate framework.
33873 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33875 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
33876 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
33877 define_insn pattern.
33878 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
33879 define_insn pattern because the xvcmpne<VSs>. instruction is not
33881 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
33882 instruction is not supported.
33884 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33886 * config/nvptx/nvptx.c: Include intl.h.
33888 2017-03-01 Martin Jambor <mjambor@suse.cz>
33891 * ipa-prop.h (ipa_bits): Removed field known.
33892 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
33893 to pointers. Adjusted their comments to warn about their sharing.
33894 (ipcp_transformation_summary): Change bits to a vector of pointers.
33895 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
33896 (ipa_get_ipa_bits_for_value): Declare.
33897 * tree-vrp.h (value_range): Mark as GTY((for_user)).
33898 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
33899 (ipa_bits_hash_table): Likewise.
33900 (ipa_vr_ggc_hash_traits): Likewise.
33901 (ipa_vr_hash_table): Likewise.
33902 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
33903 being pointers and vr_known being removed.
33904 (ipa_set_jf_unknown): Likewise.
33905 (ipa_get_ipa_bits_for_value): New function.
33906 (ipa_set_jfunc_bits): Likewise.
33907 (ipa_get_value_range): New overloaded functions.
33908 (ipa_set_jfunc_vr): Likewise.
33909 (ipa_compute_jump_functions_for_edge): Use the above functions to
33910 construct bits and vr parts of jump functions.
33911 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
33912 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33914 (ipcp_grow_transformations_if_necessary): Also allocate
33915 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33917 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
33918 them. Fix too long lines.
33919 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
33920 vr_known being removed.
33921 (ipa_read_jump_function): Use new setter functions to construct bits
33922 and vr parts of jump functions or set them to NULL.
33923 (write_ipcp_transformation_info): Adjust for bits being pointers.
33924 (read_ipcp_transformation_info): Likewise.
33925 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
33927 Include gt-ipa-prop.h.
33928 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
33930 (ipcp_store_bits_results): Likewise.
33931 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
33932 Do not write to existing jump functions but use a temporary instead.
33934 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33937 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
33938 attempt to use its first operand as BIT_FIELD_REF base.
33940 2017-03-01 Richard Biener <rguenther@suse.de>
33942 PR middle-end/79721
33943 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
33944 interpolating formula in wrapping arithmetic.
33945 (chrec_apply): Convert chrec_evaluate return value to wanted type.
33947 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33949 PR tree-optimization/79734
33950 * tree-vect-generic.c (expand_vector_condition): Optimize
33951 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
33952 Handle VEC_COND_EXPR where comparison has different inner width from
33953 type's inner width.
33955 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
33957 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
33958 markup, and similar issues. Remove @opindex entries for things
33959 that aren't options. Add missing -mmpy-option entries.
33961 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33963 PR tree-optimization/79737
33964 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
33965 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
33966 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
33967 instead of byte_size. Formatting fix.
33968 (shift_bytes_in_array_right): Formatting fix.
33970 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
33973 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
33974 condition on optimize for the leaf function test.
33976 2017-02-28 Martin Liska <mliska@suse.cz>
33979 * read-rtl-function.c (function_reader::handle_unknown_directive):
33980 Bail out when one uses -flto.
33982 2017-02-28 Martin Liska <mliska@suse.cz>
33984 * common.opt: Replace space with tabular for options of <number>
33986 * config/i386/i386.opt: Show <number> value for
33987 -mlarge-data-threshold.
33988 * opts.c (print_filtered_help): Do not display number in hexadecimal
33991 2017-02-28 Martin Liska <mliska@suse.cz>
33993 * common.opt: Fix --help=option -Q for options which are of
33996 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
33998 * config/i386/i386.c (print_reg): Error out for values
33999 of 8-bit size in invalid integer register.
34001 2017-02-28 Martin Sebor <msebor@redhat.com>
34003 PR tree-optimization/79691
34004 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
34006 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34009 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
34010 gcc_unreachable with output_operand_lossage.
34012 2017-02-28 Richard Biener <rguenther@suse.de>
34014 PR tree-optimization/79740
34015 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
34017 (visit_nary_op): Insert the nary into the hashtable if we
34018 pattern-matched sth.
34019 * tree-ssa-pre.c (eliminate_insert): Robustify.
34021 2017-02-28 Richard Biener <rguenther@suse.de>
34023 PR middle-end/79731
34024 * fold-const.c (decode_field_reference): Reject out-of-bound
34027 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34029 * config/i386/i386.c: Include intl.h.
34030 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
34031 instead of just cond ? "..." : "...".
34032 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
34033 * coverage.c (read_counts_file): Likewise.
34034 * omp-offload.c: Include intl.h.
34035 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
34036 of just cond ? "..." : "...".
34037 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
34038 of just cond ? "..." : "...".
34040 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
34043 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
34045 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
34046 'tune for' CPU name.
34047 * config/arm/arm-cpu-data.h: Regenerated.
34049 2017-02-28 Richard Biener <rguenther@suse.de>
34051 PR tree-optimization/79732
34052 * tree-inline.c (expand_call_inline): Do not shadow var.
34054 2017-02-28 Richard Biener <rguenther@suse.de>
34056 PR tree-optimization/79723
34057 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
34058 address-space properly.
34060 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
34062 * doc/optinfo.texi (Optimization groups): Fix option used for
34064 * doc/invoke.texi (-fopt-info): Document "omp".
34065 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
34066 (OPTGROUP_ALL): Add OPTGROUP_OMP.
34067 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
34068 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
34069 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
34071 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
34073 * dumpfile.c (optgroup_options): Instead of "openmp", associate
34074 OPTGROUP_OMP with "omp".
34076 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
34079 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
34080 for arithmetic shift of unsigned V2DI.
34082 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
34084 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
34085 arc/linux.h headers.
34086 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
34087 (LINK_SPEC): Likewise.
34088 (ARC_TLS_EXTRA_START_SPEC): Likewise.
34089 (EXTRA_SPECS): Likewise.
34090 (STARTFILE_SPEC): Likewise.
34091 (ENDFILE_SPEC): Likewise.
34092 (LIB_SPEC): Likewise.
34093 (TARGET_SDATA_DEFAULT): Likewise.
34094 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
34095 (MULTILIB_DEFAULTS): Likewise.
34096 (DWARF2_UNWIND_INFO): Likewise.
34097 * config/arc/big.h: New file.
34098 * config/arc/elf.h: Likewise.
34099 * config/arc/linux.h: Likewise.
34100 * config/arc/t-uClibc: Remove.
34102 2017-02-27 Bin Cheng <bin.cheng@arm.com>
34104 PR tree-optimization/77536
34105 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
34106 (tree_transform_and_unroll_loop): Use above function to compute the
34107 estimated niter of unrolled loop and use it when scaling profile.
34108 Also use count info rather than frequency if it's non-zero.
34109 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
34110 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
34111 (vect_transform_loop): Call above function.
34113 2017-02-27 Richard Biener <rguenther@suse.de>
34115 PR tree-optimization/45397
34116 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
34117 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
34118 (visit_nary_op): Add pattern matching for CSEing sign-changed
34119 or truncated operations with wider ones.
34121 2017-02-27 Richard Biener <rguenther@suse.de>
34123 PR tree-optimization/79690
34124 * tree-vect-stmts.c (vectorizable_store): Use vector type
34125 built from the DR with address-space.
34127 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
34129 * doc/invoke.texi (Optimize Options): Refine the description
34130 of asan-use-after-return.
34132 2017-02-25 Alan Modra <amodra@gmail.com>
34134 PR rtl-optimization/79584
34135 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
34136 base, not ad->base_term, the reg within base. Remove assertion
34137 that ad->base == ad->base_term. Replace gen_int_mode using
34138 bogus mode with const0_rtx.
34140 2017-02-25 Jakub Jelinek <jakub@redhat.com>
34142 PR middle-end/79396
34143 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
34144 FMA_EXPR like tcc_binary or tcc_unary.
34146 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
34149 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
34151 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
34152 (output_loc_operands): Handle DW_OP_call_ref and
34153 DW_OP_GNU_variable_value.
34154 (struct variable_value_struct): New type.
34155 (struct variable_value_hasher): Likewise.
34156 (variable_value_hash): New variable.
34157 (string_types): Remove.
34158 (copy_loc_descr): New function.
34159 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
34160 (prepend_loc_descr_to_each): New function.
34161 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
34162 instead of add_loc_descr_to_each if the first argument is single
34163 location list and the second has multiple.
34164 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
34165 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
34166 when looking for variable value which doesn't have other location info.
34167 (loc_list_from_tree): Formatting fix.
34168 (gen_array_type_die): Simplify DW_AT_string_length handling.
34169 (adjust_string_types): Remove.
34170 (gen_subprogram_die): Don't call adjust_string_types nor test/set
34171 string_types. Call resolve_variable_values.
34172 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
34173 (resolve_addr_in_expr): Likewise. Add A argument.
34174 (copy_deref_exprloc): Remove deref argument. Adjust for the
34175 original expression being DW_OP_GNU_variable_value with optionally
34176 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
34177 optionally after it.
34178 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
34179 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
34180 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
34181 (variable_value_hasher::hash, variable_value_hasher::equal): New
34183 (resolve_variable_value_in_expr, resolve_variable_value,
34184 resolve_variable_values, note_variable_value_in_expr,
34185 note_variable_value): New functions.
34186 (dwarf2out_early_finish): Call note_variable_value on all toplevel
34189 2017-02-24 Jakub Jelinek <jakub@redhat.com>
34192 * opts.h (handle_generated_option): Add GENERATED_P argument.
34193 * opts-common.c (handle_option): Adjust function comment.
34194 (handle_generated_option): Add GENERATED_P argument, pass it to
34196 (control_warning_option): Pass false to handle_generated_option
34198 * opts.c (maybe_default_option): Pass true to handle_generated_option
34200 * optc-gen.awk: Likewise.
34202 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34204 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
34205 a REG, look at the REG it is a SUBREG of.
34206 (splitter for cmpeqsi_t): Ditto.
34208 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34210 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
34211 the special USEs with the pattern of the insn, not the insn itself.
34213 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
34216 * doc/invoke.texi: Document -mload-store-pairs.
34218 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34219 Sandra Loosemore <sandra@codesourcery.com>
34221 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
34222 argument isn't a CONST_INT.
34223 (nios2_alternate_compare_const): Assert op is a CONST_INT.
34224 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
34225 (nios2_validate_compare): Bypass alternate compare logic if *op2
34226 is not a CONST_INT.
34227 (ldstwm_operation_p): Return false if first_base is not a REG or
34228 if first_offset is not a CONST_INT.
34230 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34232 * config/cris/cris.md: Use correct operand in a define_peephole2.
34234 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34236 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
34238 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34240 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
34241 this_insn if it is an INSN or JUMP_INSN.
34242 (force_offsettable): Look at base, not at addr.
34243 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
34244 on things that aren't necessarily CONST_INTs.
34246 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
34248 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
34249 -mfpmath=sse is the default also for x86-32 targets with SSE2
34250 instruction set when @option{-ffast-math} is enabled
34252 2017-02-24 Jeff Law <law@redhat.com>
34254 PR rtl-optimizatoin/79286
34255 * ira.c (update_equiv_regs): Drop may_trap_p exception to
34258 2017-02-24 Richard Biener <rguenther@suse.de>
34260 PR tree-optimization/79389
34261 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
34264 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
34266 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
34267 function comment to reflect reality.
34268 (loop_exits_before_overflow): Fix typo in function description.
34270 2017-02-24 Richard Biener <rguenther@suse.de>
34272 PR tree-optimization/79389
34273 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
34274 properly that a threading opportunity exists. Detect conditional
34275 copy/constant propagation opportunities.
34277 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
34279 * config/visium/visium.md (type): Add trap.
34280 (b): New mode attribute.
34281 (*btst): Rename into...
34282 (*btst<mode>): ...this and adjust.
34283 (*cbranchsi4_btst_insn): Rename into...
34284 (*cbranch<mode>4_btst_insn): ...this and adjust.
34285 (trap): New define_insn.
34287 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34289 PR tree-optimization/79389
34290 * ifcvt.c (struct noce_if_info): Add rev_cond field.
34291 (noce_reversed_cond_code): New function.
34292 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
34293 reversed_comparison_code. Formatting fix.
34294 (noce_try_store_flag): Test rev_cond != NULL in addition to
34295 reversed_comparison_code.
34296 (noce_try_store_flag_constants): Likewise.
34297 (noce_try_store_flag_mask): Likewise.
34298 (noce_try_addcc): Use rev_cond if non-NULL instead of
34299 reversed_comparison_code.
34300 (noce_try_cmove_arith): Likewise. Formatting fixes.
34301 (noce_try_minmax, noce_try_abs): Clear rev_cond.
34302 (noce_find_if_block): Initialize rev_cond.
34303 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
34304 instead of false as last argument never attempt to reverse it
34307 2017-02-23 Bin Cheng <bin.cheng@arm.com>
34309 PR tree-optimization/79663
34310 * tree-predcom.c (combine_chains): Process refs in reverse order
34311 only for ZERO length chains, and add explaining comment.
34313 2017-02-23 Jeff Law <law@redhat.com>
34315 PR tree-optimization/79578
34316 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
34317 in call to operand_equal_p.
34319 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
34322 * config/i386/cpuid.h: Fix another undefined behavior.
34324 2017-02-23 Richard Biener <rguenther@suse.de>
34326 PR tree-optimization/79683
34327 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
34328 vector types for data-refs.
34330 2017-02-23 Martin Liska <mliska@suse.cz>
34332 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
34334 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34336 PR middle-end/79665
34337 * internal-fn.c (get_range_pos_neg): Moved to ...
34338 * tree.c (get_range_pos_neg): ... here. No longer static.
34339 * tree.h (get_range_pos_neg): New prototype.
34340 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
34341 are known to be in between 0 and signed maximum inclusive, try to
34342 expand both unsigned and signed divmod and use the cheaper one from
34345 2017-02-22 Jeff Law <law@redhat.com>
34347 PR tree-optimization/79578
34348 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
34349 to compare base operands.
34351 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
34354 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
34355 gpc_reg_operand instead of fpr_reg_operand.
34357 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
34359 * config/mips/mips.c (mips_return_in_memory): Force FP
34360 vector types to be returned in memory for o32 ABI.
34362 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34364 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
34365 instead of DW_TAG_member for static data member declarations and don't
34366 set no_linkage_name for static inline data members.
34367 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
34370 2017-02-22 Martin Liska <mliska@suse.cz>
34372 * doc/invoke.texi: Replace inequality signs with square brackets
34375 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
34378 * lra-constraints.c (simplify_operand_subreg): Handle
34379 WORD_REGISTER_OPERATIONS targets.
34381 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34384 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
34385 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
34386 elimination by swapping fld*.
34388 2017-02-22 Richard Biener <rguenther@suse.de>
34390 PR tree-optimization/79673
34391 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
34392 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
34393 irrelevant address-space qualifiers and avoiding a
34394 ADDR_SPACE_CONVERT_EXPR from fold_convert.
34396 2017-02-22 Richard Biener <rguenther@suse.de>
34398 PR tree-optimization/79666
34399 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
34400 to not symbolically negate if that may introduce undefined
34403 2017-02-22 Martin Liska <mliska@suse.cz>
34406 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
34407 * data-streamer-out.c (streamer_write_gcov_count_stream):
34409 * value-prof.c (stream_out_histogram_value): Make assert more
34410 precise based on type of counter.
34412 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
34415 * config/i386/i386.md (standard_x87sse_constant_load splitter):
34416 Use nonimmediate_operand instead of memory_operand for operand 1.
34417 (float-extend standard_x87sse_constant_load splitter): Ditto.
34419 2017-02-21 Jeff Law <law@redhat.com>
34421 PR tree-optimization/79621
34422 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
34423 blocks with edges to themselves.
34425 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34428 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
34429 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
34430 Use gimple_call_builtin_p.
34433 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
34434 on temporarily removed DEBUG_INSNs.
34436 PR tree-optimization/79649
34437 * tree-loop-distribution.c (classify_partition): Give up on
34438 non-generic address space loads/stores.
34440 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
34442 * doc/loop.texi (Loop manipulation): Remove nonexistent
34443 tree_ssa_loop_version from the documentation.
34444 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
34446 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34449 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
34450 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
34451 * config/rs6000/rs6000.c: Include except.h.
34452 (rs6000_expand_split_stack_prologue): Call
34453 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
34455 2017-02-21 Martin Jambor <mjambor@suse.cz>
34458 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
34459 have been analyzed.
34461 2017-02-21 Martin Jambor <mjambor@suse.cz>
34463 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
34464 for backward compatibility only.
34465 * doc/invoke.texi (Option Summary): Remove all references to
34466 -fipa-cp-alignment.
34468 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
34472 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34474 * lra-constraints.c (curr_insn_transform): Handle
34475 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34477 2017-02-21 Martin Liska <mliska@suse.cz>
34479 * config/i386/i386.opt: Replace -masm-dialect with -masm.
34481 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
34483 PR translation/79638
34484 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
34486 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
34489 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
34490 (arm_function_ok_for_sibcall): Return false for an indirect call by
34491 descriptor if all the argument registers are used.
34492 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
34493 alignment of the function.
34495 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34497 PR tree-optimization/61441
34498 * simplify-rtx.c (simplify_const_unary_operation): For
34499 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
34500 the sNaN unmodified.
34502 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34504 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34505 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34506 instead of SYSTEM_HEADER_DIR.
34508 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
34509 Martin Liška <mliska@suse.cz>
34511 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
34512 Fix typos and grammar, use active voice, and clarify.
34514 2017-02-20 Marek Polacek <polacek@redhat.com>
34516 PR middle-end/79537
34517 * gimplify.c (gimplify_expr): Handle unused *&&L;.
34520 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
34522 2017-02-20 Jakub Jelinek <jakub@redhat.com>
34525 * config/i386/i386.c (ix86_expand_builtin): Handle
34526 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
34527 ix86_builtins_isa[fcode].isa as a requirement of those
34528 flags and any other flag in the bitmask.
34529 (ix86_init_mmx_sse_builtins): Use 0 instead of
34530 ~OPTION_MASK_ISA_64BIT as mask.
34531 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
34532 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
34533 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
34534 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
34536 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34539 * lra-constraints.c (split_reg): Check requested split mode
34540 is supported by the register.
34542 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34544 * lra-constraints.c (simplify_operand_subreg): Remove early
34547 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34550 * lra-constraints.c (curr_insn_transform): Tighten condition
34551 for converting SUBREG reloads from OP_OUT to OP_INOUT.
34553 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34556 * lra-constraints.c (curr_insn_transform): Handle
34557 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34559 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
34562 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
34564 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
34566 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
34569 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
34572 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34574 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
34577 2017-02-18 Jakub Jelinek <jakub@redhat.com>
34579 * final.c (last_columnnum, override_columnnum): New variables.
34580 (final_start_function): Set last_columnnum, pass it to begin_prologue
34581 hook and pass 0 to dwarf2out_begin_prologue.
34582 (final_scan_insn): Update override_columnnum. Pass last_columnnum
34583 to source_line debug hook.
34584 (notice_source_line): Compute last_columnnum and for debug_column_info
34585 return true on column changes.
34586 * debug.h (struct gcc_debug_hooks): Add column argument to
34587 source_line and begin_prologue hooks.
34588 (debug_nothing_int_charstar_int_bool): Remove prototype.
34589 (debug_nothing_int_int_charstar,
34590 debug_nothing_int_int_charstar_int_bool): New prototypes.
34591 (dwarf2out_begin_prologue): Add column argument.
34592 * debug.c (do_nothing_debug_hooks): Adjust source_line and
34593 begin_prologue hooks.
34594 (debug_nothing_int_charstar_int_bool): Remove.
34595 (debug_nothing_int_int_charstar,
34596 debug_nothing_int_int_charstar_int_bool): New functions.
34597 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
34598 through to dwarf2out_source_line.
34599 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
34600 (dwarf2out_source_line): Add column argument, emit it if requested.
34601 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
34603 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
34604 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
34605 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
34606 through to dwarf2out_begin_prologue.
34607 (vmsdbgout_source_line): Add column argument, pass it through to
34608 dwarf2out_source_line.
34609 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
34610 dbxout_source_line caller.
34611 (dbxout_source_line): Add column argument.
34613 * common.opt (gno-column-info, gcolumn-info): New options.
34614 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
34615 (check_die): Also test for multiple DW_AT_decl_column attributes.
34616 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
34617 DW_AT_decl_column if requested.
34618 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
34620 (gen_variable_die): Likewise.
34621 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
34622 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
34625 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
34626 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
34627 (ix86_handle_option): Handle OPT_m3dnowa.
34628 * doc/invoke.texi (-m3dnowa): Document.
34629 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
34630 -m3dnowa instead of -m3dnow -march=athlon.
34633 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
34634 instead of gcc_assert for K, r and R code checks. Formatting fixes.
34636 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34639 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
34640 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
34641 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
34642 generator for vsx_xxpermdi_<mode>_be.
34643 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
34644 force big-endian semantics.
34645 (vsx_xxpermdi_<mode>_be): New define_expand with same
34646 implementation as previous version of vsx_xxpermdi_<mode>.
34648 2017-02-17 Jakub Jelinek <jakub@redhat.com>
34650 PR tree-optimization/79327
34651 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
34652 variable, its initialization and use.
34654 2017-02-17 Julia Koval <julia.koval@intel.com>
34656 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
34657 (OPTION_MASK_ISA_PKU_UNSET): New.
34658 (ix86_handle_option): Handle -mrdpid.
34659 * config/i386/cpuid.h (bit_RDPID): New.
34660 * config/i386/driver-i386.c (host_detect_local_cpu):
34661 Detect RDPID feature.
34662 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
34663 * config/i386/i386-c.c (ix86_target_macros_internal):
34665 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
34666 (ix86_valid_target_attribute_inner_p): Add "rdpid".
34667 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
34668 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
34669 * config/i386/i386.md (define_insn "rdpid"): New.
34670 * config/i386/i386.opt Add -mrdpid.
34671 * config/i386/immintrin.h (_rdpid_u32): New.
34673 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
34675 PR rtl-optimization/79541
34676 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
34677 instead of transforming it into USE.
34679 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
34681 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
34682 If HONOR_SNANS (SFmode) force the input to a register.
34683 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
34684 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
34685 an frsp or similar insn.
34687 2017-02-17 Martin Liska <mliska@suse.cz>
34689 PR rtl-optimization/79577
34690 * params.def (selsched-max-sched-times): Increase minimum to 1.
34692 2017-02-17 Martin Liska <mliska@suse.cz>
34694 PR rtl-optimization/79574
34695 * gcse.c (want_to_gcse_p): Prevent integer overflow.
34697 2017-02-17 Martin Liska <mliska@suse.cz>
34699 PR tree-optimization/79529
34700 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
34701 ssa_defined_default_def_p to handle cases which are implicitly
34703 * tree-ssa.c (ssa_defined_default_def_p): New function.
34704 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
34705 which are implicitly defined.
34706 * tree-ssa.h (ssa_defined_default_def_p): Declare.
34708 2017-02-17 Richard Biener <rguenther@suse.de>
34710 PR middle-end/79576
34711 * params.def (max-ssa-name-query-depth): Limit to 10.
34713 2017-02-17 Richard Biener <rguenther@suse.de>
34715 PR tree-optimization/79552
34716 * tree-ssa-structalias.c (visit_loadstore): Properly verify
34719 2017-02-17 Richard Biener <rguenther@suse.de>
34722 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
34724 2017-02-17 Marek Polacek <polacek@redhat.com>
34726 PR middle-end/79536
34727 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
34728 (fold_negate_expr): New wrapper.
34730 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
34732 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
34733 Correct terminology and de-emphasize pre-standard behavior.
34735 2017-02-16 Alan Modra <amodra@gmail.com>
34737 PR rtl-optimization/79286
34738 * ira.c (def_dominates_uses): New function.
34739 (update_equiv_regs): Don't create an equivalence for insns that
34740 may trap where the register def does not dominate the use.
34742 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
34744 PR rtl-optimization/78127
34745 * lra.c (lra): Call lra_eliminate before finish the loop after
34748 2017-02-16 Richard Biener <rguenther@suse.de>
34750 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
34752 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
34753 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
34754 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
34755 (isl_val_int_from_wi): New function.
34756 (extract_affine_gmp): Rename to ...
34757 (extract_affine_wi): ... this, take a widest_int.
34758 (extract_affine_int): Just wrap extract_affine_wi.
34759 (add_param_constraints): Use isl_val_int_from_wi.
34760 (add_loop_constraints): Likewise, and extract_affine_wi.
34762 2017-02-15 Jeff Law <law@redhat.com>
34764 PR middle-end/79521
34765 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
34766 ira_init_register_move_cost_if_necessary.
34768 2017-02-15 Martin Sebor <msebor@redhat.com>
34770 PR middle-end/32003
34771 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
34772 removed in a prior commit.
34774 2017-02-15 Bin Cheng <bin.cheng@arm.com>
34776 PR tree-optimization/79347
34777 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
34778 counters during peeling.
34780 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
34782 * Makefile.in (site.exp): Remove "set ISLVER".
34784 2017-02-15 Jakub Jelinek <jakub@redhat.com>
34787 * real.c (real_from_integer): Call real_convert even for decimal.
34789 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
34792 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
34794 2017-02-14 Andrew Pinski <apinski@cavium.com>
34796 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
34797 cores and change the partno/implementer to be correct.
34798 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
34799 the 'B" as the implementer.
34800 * config/aarch64/aarch64-tune.md: Regenerate.
34802 2017-02-14 Carl Love <cel@us.ibm.com>
34804 * config/rs6000/rs6000.c: Add case statement entry to make the
34805 xvcvuxdsp built-in argument unsigned.
34806 * config/rs6000/vsx.md: Fix the source and return operand types so they
34807 match the instruction definitions from the ISA document. Fix typo
34808 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
34811 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
34814 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
34815 member early_clobber_alts.
34816 * lra-lives.c (reg_early_clobber_p): New.
34817 (process_bb_lives): Use it.
34818 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
34819 (debug_operand_data): Initialize early_clobber_alts.
34820 (setup_operand_alternative): Set up early_clobber_alts.
34821 (collect_non_operand_hard_regs): Ditto. Pass early clobber
34822 alternatives to new_insn_reg.
34823 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
34825 (lra_update_insn_regno_info): Pass the new arg.
34827 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34829 PR middle-end/79505
34830 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
34831 (new_oacc_loop_raw): Don't clear already cleared fields.
34834 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
34835 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
34836 _mm512_prefetch_i64gather_ps): New inline functions and macros.
34838 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
34841 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
34843 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
34846 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
34847 the extra instruction to the right place to store 128-bit constant
34850 2017-02-14 Martin Sebor <msebor@redhat.com>
34852 PR middle-end/79448
34853 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
34854 warning for strings of unknown length.
34856 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
34858 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
34860 2017-02-14 Jeff Law <law@redhat.com>
34863 * ira-costs.c (scan_one_insn): Initialize register move costs
34864 for pseudos seen in USE/CLOBBER insns.
34866 PR tree-optimization/79095
34867 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
34868 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
34869 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
34870 if the operands are known to be not equal, then the resulting range
34872 (intersect_ranges): If the new range is ~[0,0] and the old range is
34873 wide, then prefer ~[0,0].
34874 * tree-vrp.c (overflow_comparison_p_1): New function.
34875 (overflow_comparison_p): New function.
34876 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
34877 if NAME is used in an overflow test.
34878 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
34879 overflow check that can be expressed as an equality test, then adjust
34880 ops to be that equality test.
34882 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34884 * config/s390/s390-builtin-types.def: Remove flags argument.
34885 * config/s390/s390.c (s390_init_builtins): Likewise.
34887 2017-02-14 Martin Liska <mliska@suse.cz>
34889 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
34890 vector. Fix trailing white spaces.
34892 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
34894 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
34897 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34899 PR rtl-optimization/68664
34900 * config/arm/arm.c (arm_sched_can_speculate_insn):
34901 New function. Declare prototype.
34902 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34904 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34906 PR rtl-optimization/68664
34907 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
34909 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34911 2017-02-14 Amit Pawar <amit.pawar@amd.com>
34913 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
34914 max skip bytes for function, loop and jump.
34916 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34918 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
34919 ABS_EXPR for gimple dump.
34921 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34924 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
34926 PR tree-optimization/79408
34927 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
34928 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
34929 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
34930 also if rhs1 is INTEGER_CST.
34932 2017-02-14 Richard Biener <rguenther@suse.de>
34934 PR middle-end/79432
34935 * tree-into-ssa.c (insert_phi_nodes): When the function can
34936 have abnormal edges rewrite SSA names with broken use-def
34937 dominance out of SSA and register them for PHI insertion.
34939 2017-02-13 Martin Sebor <msebor@redhat.com>
34941 PR middle-end/79496
34942 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
34943 clearing info.nowrite flag when snprintf size argument is a range.
34945 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34947 * cprop.c (cprop_jump): Add missing space in string literal.
34948 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
34949 (get_constraint_for_component_ref): Likewise.
34950 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
34951 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
34952 * lra-constraints.c (process_alt_operands): Likewise.
34953 * ipa-inline.c (inline_small_functions): Likewise.
34954 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
34955 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
34956 * trans-mem.c (diagnose_tm_1_op): Likewise.
34957 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
34958 (grid_parallel_clauses_gridifiable): Likewise.
34960 * config/nvptx/mkoffload.c (process): Add space in between
34963 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
34964 "MOD4_SSE_REGS" and "ALL_REGS".
34966 * spellcheck.c (test_data): Add , in between "foo" and "food".
34968 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34971 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
34972 boundary crossing check and subsequent code generation agree.
34974 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34976 * config/aarch64/aarch64.c (has_memory_op): Delete.
34977 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
34980 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34982 PR rtl-optimization/79388
34983 PR rtl-optimization/79450
34984 * combine.c (distribute_notes): When removing TEM_INSN for which
34985 corresponding dest has last value recorded, invalidate that last
34988 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34990 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
34991 of explicit '@'. Add missing assembly comment marker on branch costs
34994 2017-02-13 Nathan Sidwell <nathan@acm.org>
34996 * gengtype-lex.l (<in_struct>): Add '/'.
34998 2017-02-13 Martin Liska <mliska@suse.cz>
35001 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
35003 2017-02-13 Richard Biener <rguenther@suse.de>
35005 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
35007 * configure: Re-generate.
35008 * config.in: Likewise.
35009 * graphite-dependences.c: Simplify as if
35010 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
35011 * graphite-isl-ast-to-gimple.c: Likewise.
35012 * graphite-optimize-isl.c: Likewise.
35013 * graphite-poly.c: Likewise.
35014 * graphite-sese-to-poly.c: Likewise.
35015 * graphite.h: Likewise.
35016 * toplev.c: Include isl/version.h and use isl_version () for
35017 printing the ISL version.
35018 * doc/install.texi: Update ISL requirement.
35020 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
35022 * doc/standards.texi (Standards): Update reference to
35025 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
35027 * doc/extend.texi (Named Address Spaces): sourceware.org now
35029 * doc/install.texi (Binaries): Ditto.
35032 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
35034 * doc/cpp.texi: Replace "stringify"/"stringification" with C
35035 standard terminology "stringize"/"stringizing" throughout.
35036 * doc/cppinternals.texi: Likewise.
35038 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
35040 * doc/extend.texi: Fix some spelling mistakes and typos.
35041 * doc/invoke.texi: Likewise.
35043 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
35046 * params.def (inline-min-speedup) Change from 10 to 8.
35048 2017-02-11 Jakub Jelinek <jakub@redhat.com>
35050 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
35053 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
35056 * ipa-inline-analysis.c (get_minimal_bb): New function.
35057 (record_modified): Use it.
35058 (remap_edge_change_prob): Handle also ancestor functions.
35060 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
35062 * doc/contrib.texi (Contributors): Remove broken link into
35063 the Mauve CVS repository.
35065 2017-02-11 Jakub Jelinek <jakub@redhat.com>
35067 PR middle-end/79454
35068 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
35069 result computation whenever lhs doesn't have vector mode, not
35070 just when it has BLKmode.
35072 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
35074 * doc/makefile.texi (profiledbootstrap): Refer to the
35075 installation instructions only in textual form.
35077 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35080 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
35082 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
35084 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
35085 (Specific): Update mingw-w64 reference.
35087 (Specific): Remove broken link to Renesas RX processor.
35089 2017-02-10 Richard Biener <rguenther@suse.de>
35091 * toplev.c (process_options): Do not mention obsolete graphite
35092 options when printing sorry message about missing graphite support.
35093 Mention -floop-nest-optimize.
35095 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
35097 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
35098 (vtst_p16): Likewise.
35099 (vtstq_p8): Likewise.
35100 (vtstq_p16): Likewise.
35102 (vtstq_p64): Likewise.
35103 * config/arm/arm_neon.h (vgetq_lane_p64): New.
35104 (vset_lane_p64): New.
35105 (vsetq_lane_p64): New.
35107 2017-02-10 Jakub Jelinek <jakub@redhat.com>
35109 PR tree-optimization/79411
35110 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
35111 stmt operands are SSA_NAMEs used in abnormal phis.
35112 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
35115 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35118 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
35121 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35123 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
35125 2017-02-09 Jakub Jelinek <jakub@redhat.com>
35127 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
35131 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
35132 not arbitrary TREE_CONSTANT.
35135 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
35136 "omp declare target link" attribute unless is_global_var.
35137 * omp-offload.c (find_link_var_op): Likewise.
35139 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
35140 Chung-Lin Tang <cltang@codesourcery.com>
35142 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
35144 (gimplify_adjust_omp_clauses): Don't delete TILE.
35145 (gimplify_omp_for): Deal with TILE.
35146 * internal-fn.c (expand_GOACC_TILE): New function.
35147 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
35149 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
35150 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
35152 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
35153 avoid DIV for outermost collapse var.
35154 (expand_oacc_for): Insert tile element loop as needed. Adjust.
35155 Remove out of date comments, fix whitespace.
35156 * omp-general.c (omp_extract_for_data): Deal with tiling.
35157 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
35158 adjust OLF_DIM_BASE value.
35159 (struct omp_for_data): Add tiling field.
35160 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
35161 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
35162 for auto loops. Remove default auto determining, moved to
35163 oacc_loop_fixed_partitions.
35164 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
35165 stmts, add e_mask field.
35166 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
35167 (oacc_thread_numbers): Use oacc_dim_call.
35168 (oacc_xform_tile): New.
35169 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
35170 (finish_oacc_loop): Adjust for ifns vector.
35171 (oacc_loop_discover_walk): Append loop abstraction sites to list,
35172 add case for GOACC_TILE fns.
35173 (oacc_loop_xform_loop): Delete.
35174 (oacc_loop_process): Iterate over call list directly, and add
35175 handling for GOACC_TILE fns.
35176 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
35178 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
35179 vector partitioning to outer loops. Assign 2 partitions to loops
35180 when available. Add TILE handling.
35181 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
35182 (execite_oacc_device_lower): Process GOACC_TILE fns,
35183 ignore unknown specs.
35184 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
35185 * tree.c (omp_clause_num_ops): Adjust TILE ops.
35186 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
35188 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
35190 * configure.ac (ACX_BUGURL): Update.
35191 * configure: Regenerate.
35193 2017-02-09 Richard Biener <rguenther@suse.de>
35195 PR tree-optimization/69823
35196 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
35197 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
35199 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35201 * config/arc/arc-c.def: Add __NPS400__ definition.
35202 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
35203 (TARGET_NPS400): Define.
35205 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35207 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
35209 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
35210 pointer, arch_info.
35211 (arc_cpu_types): Fill the arch_info field with a pointer into the
35212 arc_arch_types table.
35213 (arc_selected_cpu): Declare.
35214 * config/arc/arc.c (arc_selected_cpu): Make global.
35215 (arc_selected_arch): Delete.
35216 (arc_base_cpu): Delete.
35217 (arc_override_options): Remove references to deleted variables,
35218 update access to arch information.
35219 (ARC_OPT): Update access to arch information.
35220 (ARC_OPTX): Likewise.
35221 * config/arc/arc.h (arc_base_cpu): Remove declaration.
35222 (TARGET_ARC600): Update access to arch information.
35223 (TARGET_ARC601): Likewise.
35224 (TARGET_ARC700): Likewise.
35225 (TARGET_EM): Likewise.
35226 (TARGET_HS): Likewise.
35227 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
35230 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
35233 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
35234 condition/operands for integer GE/LE/GEU/LEU operations.
35236 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
35238 PR translation/79397
35239 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
35242 2017-02-08 Martin Jambor <mjambor@suse.cz>
35245 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
35246 whether allocation happened.
35247 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
35248 nothing was allocated.
35250 2017-02-08 Jakub Jelinek <jakub@redhat.com>
35252 PR tree-optimization/79408
35253 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
35254 constant, but SSA_NAME with a known integer range, use the minimum
35255 of that range instead of op1 to determine if modulo can be replaced
35256 with its first operand.
35258 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35260 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
35262 2017-02-08 Richard Biener <rguenther@suse.de>
35264 PR tree-optimization/71824
35265 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35266 Check all loops contained in the merged region.
35268 2017-02-07 Andrew Pinski <apinski@cavium.com>
35270 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
35272 2017-02-07 Andrew Pinski <apinski@cavium.com>
35274 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
35275 (thunderxt88): Likewise.
35276 (thunderxt81): Disable LSE and change v8.1 to v8.
35277 (thunderxt83): Likewise.
35279 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35280 Richard Biener <rguenther@suse.de>
35282 PR middle-end/79399
35283 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
35284 type from int to size_t.
35285 * ira-costs.c (struct_costs_size): Change type from int to size_t.
35287 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35289 PR rtl-optimization/79386
35290 * cprop.c (bypass_conditional_jumps): Initialize
35291 bypass_last_basic_block already before splitting bbs after
35292 unconditional traps...
35293 (bypass_conditional_jumps): ... rather than here.
35296 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
35297 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
35298 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
35299 fix -masm=intel patterns.
35301 2017-02-07 Richard Biener <rguenther@suse.de>
35303 PR tree-optimization/79256
35304 PR middle-end/79278
35305 * builtins.c (get_object_alignment_2): Use min_align_of_type
35306 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
35307 and ADJUST_FIELD_ALIGN.
35309 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
35311 * doc/tm.texi: Regenerate.
35312 * stor-layout.c (layout_decl): Adjust.
35313 (update_alignment_for_field): Likewise.
35314 (place_field): Likewise.
35315 (min_align_of_type): Likewise.
35316 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
35317 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
35318 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
35319 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
35320 * config/frv/frv.c (frv_adjust_field_align): Likewise.
35321 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
35322 * config/i386/i386.c (x86_field_alignment): Likewise.
35323 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
35324 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
35325 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
35326 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
35327 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
35328 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
35332 2017-01-30 Richard Biener <rguenther@suse.de>
35334 PR tree-optimization/79256
35335 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35336 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35339 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
35341 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
35342 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
35343 builtins to SImode and emit a zero-extend, if necessary.
35345 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35347 * docs/invoke.texi (RISC-V Options): Alphabetize.
35349 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35351 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
35354 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35356 * config/riscv/riscv.c: New file.
35357 * common/config/riscv/riscv-common.c: Likewise.
35358 * config.gcc: Likewise.
35359 * config/riscv/constraints.md: Likewise.
35360 * config/riscv/elf.h: Likewise.
35361 * config/riscv/generic.md: Likewise.
35362 * config/riscv/linux.h: Likewise.
35363 * config/riscv/multilib-generator: Likewise.
35364 * config/riscv/peephole.md: Likewise.
35365 * config/riscv/pic.md: Likewise.
35366 * config/riscv/predicates.md: Likewise.
35367 * config/riscv/riscv-builtins.c: Likewise.
35368 * config/riscv/riscv-c.c: Likewise.
35369 * config/riscv/riscv-ftypes.def: Likewise.
35370 * config/riscv/riscv-modes.def: Likewise.
35371 * config/riscv/riscv-opts.h: Likewise.
35372 * config/riscv/riscv-protos.h: Likewise.
35373 * config/riscv/riscv.h: Likewise.
35374 * config/riscv/riscv.md: Likewise.
35375 * config/riscv/riscv.opt: Likewise.
35376 * config/riscv/sync.md: Likewise.
35377 * config/riscv/t-elf-multilib: Likewise.
35378 * config/riscv/t-linux: Likewise.
35379 * config/riscv/t-linux-multilib: Likewise.
35380 * config/riscv/t-riscv: Likewise.
35381 * configure.ac: Likewise.
35382 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
35383 Waterman as RISC-V maintainers.
35384 * doc/install.texi: Add RISC-V entries.
35385 * doc/invoke.texi: Add RISC-V options section.
35386 * doc/md.texi: Add RISC-V constraints section.
35387 * configure: Regenerated.
35389 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
35392 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
35393 false values to be constant vectors with all 0 or all 1 bits set.
35394 (vcondu<mode><mode>): Likewise.
35395 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
35397 (fpmask_comparison_operator): Update comment.
35398 (vecint_comparison_operator): New predicate.
35399 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
35400 vector conditionals when the true and false values are constant
35401 vectors with all 0 bits or all 1 bits set.
35403 2017-02-06 Martin Sebor <msebor@redhat.com>
35405 PR tree-optimization/79376
35406 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
35408 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
35410 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
35411 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
35412 to simplify split condition.
35414 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35416 * omp-expand.c (oxpand_omp_atomic_fetch_op,
35417 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
35420 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
35422 PR rtl-optimization/68664
35423 * target.def (can_speculate_insn): New hook.
35424 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
35425 * doc/tm.texi: Regenerate.
35426 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
35427 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
35428 (rs6000_sched_can_speculate_insn): New function.
35430 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35432 PR tree-optimization/79284
35433 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
35434 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
35435 vectorizable_mask_load_store, vectorizable_operation,
35436 vect_is_simple_cond, get_same_sized_vectype): Use it instead
35437 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
35438 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
35439 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
35440 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
35441 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
35442 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
35443 is_gimple_assign (stmt). Replace another such test with
35444 is_gimple_assign (stmt).
35446 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35449 * config/avr/avr.c (rtl-iter.h): Include it.
35450 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
35451 (avr_legitimate_combined_insn): ...and implementation.
35453 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
35455 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
35456 * config/s390/s390.c (s390_const_operand_ok)
35457 (s390_canonicalize_comparison, s390_extract_part)
35458 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
35459 (s390_contiguous_bitmask_p, s390_rtx_costs)
35460 (legitimize_pic_address): Likewise.
35461 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
35462 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
35463 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
35464 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
35465 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
35467 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35469 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
35470 REGNO($0) == REGNO($1).
35472 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35474 * config/s390/linux.h(SIZE_TYPE): Add comment.
35476 2017-02-06 Julian Brown <julian@codesourcery.com>
35477 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35478 Virendra Pathak <virendra.pathak@broadcom.com>
35480 * config/aarch64/aarch64-cores.def: Change the scheduler
35482 * config/aarch64/aarch64.md: Include thunderx2t99.md.
35483 * config/aarch64/thunderx2t99.md: New file.
35485 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35487 * doc/standards.texi (Go Language): Update link to language
35490 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35492 * tree-eh.c (lower_resx): Sanitize profile.
35493 (cleanup_empty_eh_move_lp): Likewise.
35495 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35498 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
35500 * cfgloopmanip.h (loop_version): Update prototype.
35501 * modulo-sched.c (sms_schedule): Update call of loop_version.
35502 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
35503 * tree-parloops.c (gen_parallel_loop): Likewise.
35504 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
35505 * tree-ssa-loop-split.c (split_loop): Likewise.
35506 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
35507 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
35509 2017-02-05 Martin Liska <mliska@suse.cz>
35512 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
35514 (print_operand_address): Initialize a struct to zero.
35516 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35518 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
35519 garbage collector only in textual form.
35521 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35523 * doc/extend.texi (x86 specific memory model extensions for
35524 transactional memory): Simplify a phrase.
35526 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
35529 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
35530 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
35531 (atomic_storedi_1): Likewise.
35533 2017-02-04 Jakub Jelinek <jakub@redhat.com>
35535 PR tree-optimization/79338
35536 * tree-parloops.c (gather_scalar_reductions): Don't call
35537 vect_analyze_loop_form for loop->inner before destroying loop's
35540 2017-02-03 Martin Sebor <msebor@redhat.com>
35542 PR tree-optimization/79327
35543 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
35544 when precision has resulted in leading zeros.
35545 (format_integer): Adjust the likely counter to assume an unknown
35546 argument that may be zero is non-zero.
35548 2017-02-03 Jason Merrill <jason@redhat.com>
35551 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
35552 avoid copying non-taken branch.
35554 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35556 PR tree-optimization/79340
35557 * tree-vect-loop.c (vectorizable_reduction): Release
35558 vec_defs elements after safe_splicing them into other vectors.
35561 PR tree-optimization/79327
35562 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
35563 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
35565 (format_integer): Use wide_int_to_tree instead of build_int_cst
35566 + to_?hwi. If argmin is NULL, just set argmin and argmax to
35567 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
35568 of shortest and longest sequence.
35570 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
35572 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
35573 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
35575 2017-02-03 Walter Lee <walt@tilera.com>
35578 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
35579 after initial stackframe link reg save.
35580 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
35582 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35585 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
35586 wu for stxssp alternative.
35588 2017-02-03 Martin Sebor <msebor@redhat.com>
35590 PR tree-optimization/79352
35591 * gimple-fold.c (get_range_strlen): Add argument.
35592 (get_range_strlen): Change return type to bool.
35593 (get_maxval_strlen): Pass in a dummy argument.
35594 * gimple-fold.h (get_range_strlen): Change return type to bool.
35595 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
35596 * tree.h (array_at_struct_end_p): Add argument.
35597 * tree.c (array_at_struct_end_p): Handle it.
35599 2017-02-03 Martin Liska <mliska@suse.cz>
35602 * multiple_target.c (create_dispatcher_calls): Redirect edge
35603 from a caller of a dispatcher.
35604 (expand_target_clones): Make the clones local.
35605 (ipa_target_clone): Do both target clones and resolvers.
35606 (ipa_dispatcher_calls): Remove the pass.
35607 (pass_dispatcher_calls::gate): Likewise.
35608 (make_pass_dispatcher_calls): Likewise.
35609 * passes.def (pass_target_clone): Put as very first IPA early
35612 2017-02-03 Martin Liska <mliska@suse.cz>
35614 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
35615 in case of a function with ifunc attribute.
35617 2017-02-03 Martin Liska <mliska@suse.cz>
35619 * cgraph.c (cgraph_node::dump): Dump function version info.
35620 * symtab.c (symtab_node::dump_base): Add missing new line.
35622 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
35624 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
35625 (ifcombine_ifandif): Use it.
35627 2017-02-03 Martin Liska <mliska@suse.cz>
35629 * doc/invoke.texi: Document default value for
35630 use-after-scope-direct-emission-threshold.
35632 2017-02-03 Martin Liska <mliska@suse.cz>
35634 PR tree-optimization/79339
35635 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
35636 (format_floating): Likewise.
35638 2017-02-03 Martin Liska <mliska@suse.cz>
35641 * ipa-prop.c (ipa_node_params_t::insert): Remove current
35643 (ipa_node_params_t::remove): Likewise.
35644 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
35645 initialization from removed ipa_node_params_t::insert.
35646 (ipa_node_params::~ipa_node_params): Move from removed
35647 ipa_node_params_t::release.
35648 * symbol-summary.h (symbol_summary::m_released): New member.
35649 Do not release a summary twice. Do not allow to call finalizer
35650 for types of a summary that live in GGC memory.
35652 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35654 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
35657 2017-02-02 Martin Sebor <msebor@redhat.com>
35659 PR middle-end/79275
35660 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
35661 (format_string): Tighten up the range of output for non-constant
35662 strings and correct the expected range for wide non-constant strings.
35664 2017-02-02 Martin Sebor <msebor@redhat.com>
35666 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
35668 PR middle-end/32003
35669 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
35671 (-fdump-tree-@var): Add to index and document how to come up
35672 with pass-specific option and dump file names.
35673 (-fdump-passes): Clarify where to look for output.
35675 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
35677 PR middle-end/77445
35678 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
35679 statistics of the analyzed path; allow threading for speed when
35680 any of BBs along the path are optimized for speed.
35682 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
35684 PR middle-end/78468
35685 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
35686 settings of the virtual registers.
35689 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
35691 * explow.c (get_dynamic_stack_size): Take known alignment of stack
35692 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
35695 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35697 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
35698 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
35700 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35702 * config/s390/s390.md: Add missing comments with the expanded
35704 * config/s390/vector.md: Likewise.
35705 * config/s390/vx-builtins.md: Likewise.
35707 2017-02-02 Jakub Jelinek <jakub@redhat.com>
35710 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
35711 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
35712 conditions on a single line.
35714 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35716 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35717 __S390_VX__ to __VX__.
35719 2017-02-01 Andrew Pinski <apinski@cavium.com>
35721 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
35722 stmt_info to record_stmt_cost.
35723 (vect_get_known_peeling_cost): Pass stmt_info if known to
35725 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
35726 cpu_vector_cost field into
35727 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
35728 field into vec_int_stmt_cost and vec_fp_stmt_cost.
35729 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
35730 splitting of scalar_stmt_cost and vec_stmt_cost.
35731 (thunderx_vector_cost): Likewise.
35732 (cortexa57_vector_cost): LIkewise.
35733 (exynosm1_vector_cost): Likewise.
35734 (xgene1_vector_cost): Likewise.
35735 (thunderx2t99_vector_cost): Improve after the splitting of the two
35737 (aarch64_builtin_vectorization_cost): Update for the splitting of
35738 scalar_stmt_cost and vec_stmt_cost.
35740 2017-02-01 Torvald Riegel <triegel@redhat.com>
35741 Richard Henderson <rth@redhat.com>
35743 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
35744 conditional on existance of a fast atomic load.
35745 * optabs-query.c (can_atomic_load_p): New function.
35746 * optabs-query.h (can_atomic_load_p): Declare it.
35747 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
35748 no fast atomic load is available for the particular size of access.
35749 (expand_atomic_compare_and_swap): Likewise.
35750 (expand_atomic_load): Likewise.
35751 (expand_atomic_store): Likewise.
35752 (expand_atomic_fetch_op): Likewise.
35753 * testsuite/lib/target-supports.exp
35754 (check_effective_target_sync_int_128): Remove x86 because it provides
35755 no fast atomic load.
35756 (check_effective_target_sync_int_128_runtime): Likewise.
35758 2017-02-01 Richard Biener <rguenther@suse.de>
35760 * graphite.c: Include tree-vectorizer.h for find_loop_location.
35761 (graphite_transform_loops): Provide opt-info for optimized nests.
35762 * tree-parloop.c (parallelize_loops): Provide opt-info for
35763 parallelized loops.
35765 2017-02-01 Richard Biener <rguenther@suse.de>
35767 PR middle-end/79315
35768 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
35769 was not set before.
35771 2017-02-01 Richard Biener <rguenther@suse.de>
35773 PR tree-optimization/71824
35774 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35775 Verify the loops are valid in the merged SESE region.
35776 (scop_detection::can_represent_loop_1): Check analyzing the
35777 evolution of the number of iterations in the region succeeds.
35779 2017-01-31 Ian Lance Taylor <iant@golang.org>
35781 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
35782 REG_ARGS_SIZE note to 32-bit push insns and call insn.
35784 2017-01-31 David Malcolm <dmalcolm@redhat.com>
35786 PR preprocessor/79210
35787 * input.c (get_substring_ranges_for_loc): Replace line_width
35788 assertion with error-handling.
35790 2017-01-31 Richard Biener <rguenther@suse.de>
35792 PR tree-optimization/77318
35793 * graphite-sese-to-poly.c (extract_affine): Fix assert.
35794 (create_pw_aff_from_tree): Take loop parameter.
35795 (add_condition_to_pbb): Pass loop of the condition to
35796 create_pw_aff_from_tree.
35798 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35800 * config/s390/s390.c (s390_asan_shadow_offset): New function.
35801 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
35803 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
35807 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
35809 (convert_int_to_float128): Likewise.
35810 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
35811 (convert_int_to_float128): Likewise.
35812 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
35813 (UNSPEC_IEEE128_CONVERT): Likewise.
35814 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
35815 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
35816 Use local variables for IBM extended format.
35817 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
35818 (fix_trunc<mode>si2_fprs): Likewise.
35819 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
35820 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
35821 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
35822 to know that we can now have integers of all sizes in vector
35824 (fix<uns>_<mode>di2_hw): Likewise.
35825 (float<uns>_<mode>si2_hw): Likewise.
35826 (fix_<mode>si2_hw): Likewise.
35827 (fixuns_<mode>si2_hw): Likewise.
35828 (float<uns>_<mode>di2_hw): Likewise.
35829 (float_<mode>di2_hw): Likewise.
35830 (float_<mode>si2_hw): Likewise.
35831 (floatuns_<mode>di2_hw): Likewise.
35832 (floatuns_<mode>si2_hw): Likewise.
35833 (xscvqp<su>wz_<mode>): Delete, no longer used.
35834 (xscvqp<su>dz_<mode>): Likewise.
35835 (xscv<su>dqp_<mode>): Likewise.
35836 (ieee128_mfvsrd_64bit): Likewise.
35837 (ieee128_mfvsrd_32bit): Likewise.
35838 (ieee128_mfvsrwz): Likewise.
35839 (ieee128_mtvsrw): Likewise.
35840 (ieee128_mtvsrd_64bit): Likewise.
35841 (ieee128_mtvsrd_32bit): Likewise.
35843 2017-01-31 Martin Liska <mliska@suse.cz>
35846 * ipa-prop.c (ipa_free_all_node_params): Call release method
35847 instead of ~sumbol_summary to not to trigger double times
35850 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
35852 PR tree-optimization/71691
35853 * bitmap.h (class auto_bitmap): New.
35854 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
35855 is_maybe_undefined instead of ssa_undefined_value_p.
35857 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35859 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35860 __S390_ARCH_LEVEL__ to __ARCH__.
35862 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35864 PR tree-optimization/79267
35865 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
35866 if should_remove_lhs_p is true.
35868 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
35871 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
35872 (add_alignment_attribute): New.
35873 (base_type_die): Add alignment attribute.
35874 (subrange_type_die): Likewise.
35875 (modified_type_die): Likewise.
35876 (gen_array_type_die): Likewise.
35877 (gen_descr_array_type_die: Likewise.
35878 (gen_enumeration_type_die): Likewise.
35879 (gen_subprogram_die): Likewise.
35880 (gen_variable_die): Likewise.
35881 (gen_field_die): Likewise.
35882 (gen_ptr_to_mbr_type_die): Likewise.
35883 (gen_struct_or_union_type_die): Likewise.
35884 (gen_subroutine_type_die): Likewise.
35885 (gen_typedef_die): Likewise.
35886 (base_type_cmp): Compare alignment attribute.
35888 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35891 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
35892 (setb_unsigned) New pattern for setb with CCUNS.
35893 * config/rs6000/rs6000.c (expand_block_compare): Use a different
35894 subfc./subfe sequence to avoid overflow problems. Generate a
35895 shorter sequence with cmpld/setb for power9.
35896 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
35897 for generating subfc. instruction.
35898 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
35899 now uses this instruction.
35901 2017-01-30 Ian Lance Taylor <iant@google.com>
35904 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
35905 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
35907 2017-01-30 Martin Sebor <msebor@redhat.com>
35909 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
35910 Move constant to the right of a relational operator.
35911 (get_mpfr_format_length, format_character, format_string): Ditto.
35912 (should_warn_p, maybe_warn): Same.
35914 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
35916 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
35919 * asan.c (get_translation_unit_decl): Remove function.
35920 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
35922 2017-01-30 Martin Liska <mliska@suse.cz>
35924 PR gcov-profile/79259
35925 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
35926 -fprofile-generate.
35928 2017-01-30 Martin Liska <mliska@suse.cz>
35931 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
35932 Initialize variables with NULL value.
35934 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
35937 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
35939 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
35941 2017-01-30 Richard Biener <rguenther@suse.de>
35943 PR tree-optimization/79276
35944 * tree-vrp.c (process_assert_insertions): Properly adjust common
35945 when removing a duplicate.
35947 2017-01-30 Richard Biener <rguenther@suse.de>
35949 PR tree-optimization/79256
35950 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35951 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35953 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
35955 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
35958 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
35959 ("*r<noxa>sbg_<mode>_sll_bitmask")
35960 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
35961 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
35962 Use contiguous_bitmask_nowrap_operand.
35964 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35967 * config/rs6000/altivec.h (vec_xl): Revise #define.
35968 (vec_xst): Likewise.
35970 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
35972 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
35974 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
35976 PR rtl-optimization/79194
35977 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
35978 traps before call to bypass_conditional_jumps.
35980 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35982 PR tree-optimization/71374
35983 * lra-constraints.c (check_conflict_input_operands): New.
35984 (match_reload): Use it.
35986 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35989 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
35990 account to calculate conflict_set.
35992 2017-01-27 Bin Cheng <bin.cheng@arm.com>
35994 PR rtl-optimization/78559
35995 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
35996 other_insn in combine.
35998 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
36000 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
36001 uint16_type_node for BT_UINT16.
36003 2017-01-27 David Malcolm <dmalcolm@redhat.com>
36005 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
36006 "RTL Tests" to menu.
36007 (GIMPLE Tests): New node.
36008 (RTL Tests): New node.
36010 2017-01-27 Richard Biener <rguenther@suse.de>
36012 PR tree-optimization/79245
36013 * tree-loop-distribution.c (distribute_loop): Apply cost
36014 modeling also to detected patterns.
36016 2017-01-27 Richard Biener <rguenther@suse.de>
36018 PR tree-optimization/71433
36019 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
36020 (compare_assert_loc): New function.
36021 (process_assert_insertions): Sort and optimize assert locations
36022 to remove duplicates and push down identical assertions on
36023 edges to their destination block.
36025 2017-01-27 Richard Biener <rguenther@suse.de>
36027 PR tree-optimization/79244
36028 * tree-vrp.c (remove_range_assertions): Forcefully propagate
36029 out SSA names even if abnormal.
36031 2017-01-27 Jakub Jelinek <jakub@redhat.com>
36033 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
36034 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
36035 instead of MPFR_RNDN.
36037 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
36040 * arm.c (arm_option_override): Don't call build_target_option_node
36041 until after doing all option overrides.
36042 (arm_valid_target_attribute_tree): Likewise.
36044 2017-01-27 Martin Liska <mliska@suse.cz>
36046 * doc/invoke.texi (-fprofile-arcs): Document profiling support
36047 for {cd}tors and C++ {cd}tors.
36049 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
36051 * config/s390/s390.md ("*setmem_long_and")
36052 ("*setmem_long_and_31z"): Use zero_extend instead of and.
36054 2017-01-26 Martin Sebor <msebor@redhat.com>
36056 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
36059 2017-01-26 Martin Sebor <msebor@redhat.com>
36061 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
36062 HAVE_DFmode before using XFmode or DFmode.
36063 (parse_directive): Avoid using the z length modifier to avoid
36064 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
36066 PR middle-end/78703
36067 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
36068 to accept adjustment as an array.
36069 (get_int_range): New function.
36070 (struct directive): Make width and prec arrays.
36071 (directive::set_width, directive::set_precision): Call get_int_range.
36072 (format_integer, format_floating): Handle width and precision ranges.
36073 (format_string, parse_directive): Same.
36075 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36078 * dwarf2out.c (generate_skeleton_bottom_up): For children with
36079 comdat_type_p set, just clone them, but keep the children in the
36083 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
36084 which have direct callers with -fvar-tracking-assignments enabled
36086 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
36087 inside of type units.
36089 2017-01-26 Martin Sebor <msebor@redhat.com>
36091 PR middle-end/78703
36092 * gimple-ssa-sprintf.c (struct result_range): Add likely and
36094 (struct format_result): Replace number_chars, number_chars_min,
36095 and number_chars_max with a single member of struct result_range.
36097 (format_result::operator+=): Adjust.
36098 (struct fmtresult): Remove bounded. Handle likely and unlikely
36100 (fmtresult::adjust_for_width_or_precision): New function.
36101 (fmtresult:type_max_digits): New function.
36102 (bytes_remaining): Handle likely and unlikely counters.
36103 (min_bytes_remaining): Remove.
36104 (format_percent): Simplify.
36105 (format_integer, format_floating): Set likely and unlikely counters.
36106 (get_string_length, format_character, format_string): Same.
36107 (format_plain, should_warn_p): New function.
36108 (maybe_warn): Call should_warn_p. Update diagnostic messages
36109 and handle those for all directives, including plain strings.
36110 (format_directive): Handle likely and unlikely counters.
36111 Remove unnecessary quoting from diagnostics. Add an informational
36113 (add_bytes): Remove.
36114 (pass_sprintf_length::compute_format_length): Simplify.
36115 (try_substitute_return_value): Handle likely and unlikely counters.
36117 2017-01-26 Carl Love <cel@us.ibm.com>
36119 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
36120 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
36122 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
36125 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
36126 endianess for subregs into account.
36127 * lra-constraints.c (lra_constraints): Do risky transformations
36128 always on the first iteration.
36129 * lra-lives.c (check_pseudos_live_through_calls): Add arg
36130 last_call_used_reg_set.
36131 (process_bb_lives): Define and use last_call_used_reg_set.
36132 * lra.c (lra): Always continue after lra_constraints on the first
36135 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
36137 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
36139 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
36141 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36143 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
36144 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
36145 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
36146 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
36147 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
36148 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
36149 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
36150 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
36151 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
36153 2017-01-26 Marek Polacek <polacek@redhat.com>
36156 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
36157 for the third operand.
36159 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36161 PR middle-end/79236
36162 * omp-low.c (struct omp_context): Add simt_stmt field.
36163 (scan_omp_for): Return omp_context *.
36164 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
36165 context to the _simt_ SIMD stmt.
36166 (lower_omp_for): For combined SIMD with sibling _simt_
36167 SIMD, make sure to use the same decls in _looptemp_
36168 clauses as in the sibling.
36170 2017-01-26 David Sherwood <david.sherwood@arm.com>
36172 PR middle-end/79212
36173 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
36176 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36179 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
36180 emit fld b; fld a; if possible.
36182 * brig-builtins.def: Update copyright years.
36183 * config/arm/arm_acle_builtins.def: Update copyright years.
36185 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
36188 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
36189 constraint instead of o for the stxsd instruction.
36191 2017-01-25 Carl Love <cel@us.ibm.com>
36193 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
36194 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
36196 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
36198 * doc/invoke.texi (C++ Dialect Options): Fix typo.
36200 2017-01-25 Richard Biener <rguenther@suse.de>
36202 PR tree-optimization/69264
36203 * target.def (vector_alignment_reachable): Improve documentation.
36204 * doc/tm.texi: Regenerate.
36205 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
36207 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
36208 earlier changes with respect to TYPE_USER_ALIGN.
36209 (vector_alignment_reachable_p): Likewise. Improve dumping.
36211 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36214 * config/arm/arm.md (xordi3): Force constant operand into a register
36217 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36219 * doc/invoke.texi (-fstore-merging): Correct default optimization
36220 levels at which it is enabled.
36221 (-O): Move -fstore-merging from list to...
36224 2017-01-25 Richard Biener <rguenther@suse.de>
36227 * omp-expand.c: Include debug.h.
36228 (expand_omp_taskreg): Make sure to generate early debug before
36229 outlining anything from a function.
36230 (expand_omp_target): Likewise.
36231 (grid_expand_target_grid_body): Likewise.
36233 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
36236 * asan.c (get_translation_unit_decl): New function.
36237 (asan_add_global): Extract modules file name from globals
36238 TRANSLATION_UNIT_DECL name.
36240 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
36243 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
36244 for long calls with APCS frame and VFP.
36246 2017-01-24 David Malcolm <dmalcolm@redhat.com>
36248 * cfg.c (original_copy_tables_initialized_p): New function.
36249 * cfg.h (original_copy_tables_initialized_p): New decl.
36250 * cfgrtl.c (relink_block_chain): Guard the call to
36251 free_original_copy_tables with a call to
36252 original_copy_tables_initialized_p.
36253 * cgraph.h (symtab_node::native_rtl_p): New decl.
36254 * cgraphunit.c (symtab_node::native_rtl_p): New function.
36255 (symtab_node::needed_p): Don't assert for early assembly output
36256 for __RTL functions.
36257 (cgraph_node::finalize_function): Set "force_output" for __RTL
36259 (cgraph_node::analyze): Bail out early for __RTL functions.
36260 (analyze_functions): Update assertion to support __RTL functions.
36261 (cgraph_node::expand): Bail out early for __RTL functions.
36262 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
36264 * function.h (struct function): Update comment for field
36266 * gimple-expr.c: Include "tree-pass.h".
36267 (gimple_has_body_p): Return false for __RTL functions.
36268 * Makefile.in (OBJS): Add run-rtl-passes.o.
36269 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
36271 (gcc::pass_manager::get_clean_slate): New accessor.
36272 * passes.c: Include "insn-addr.h".
36273 (should_skip_pass_p): Add logging. Update logic for running
36274 "expand" to be compatible with both __GIMPLE and __RTL. Guard
36275 property-provider override so it is only done for gimple passes.
36277 (skip_pass): New function.
36278 (execute_one_pass): Call skip_pass when skipping passes.
36279 * read-md.c (md_reader::read_char): Support filtering
36280 the input to a subset of line numbers.
36281 (md_reader::md_reader): Initialize fields
36282 m_first_line and m_last_line.
36283 (md_reader::read_file_fragment): New function.
36284 * read-md.h (md_reader::read_file_fragment): New decl.
36285 (md_reader::m_first_line): New field.
36286 (md_reader::m_last_line): New field.
36287 * read-rtl-function.c (function_reader::create_function): Only
36288 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
36289 curr_properties. Set DECL_INITIAL to a dummy block.
36290 (read_rtl_function_body_from_file_range): New function.
36291 * read-rtl-function.h (read_rtl_function_body_from_file_range):
36293 * run-rtl-passes.c: New file.
36294 * run-rtl-passes.h: New file.
36296 2017-01-24 Jeff Law <law@redhat.com>
36298 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
36301 2017-01-24 Bin Cheng <bin.cheng@arm.com>
36303 PR tree-optimization/79159
36304 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
36305 (record_nonwrapping_iv): Improve boundary using above function if no
36306 value range information.
36308 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
36309 Martin Jambor <mjambor@suse.cz>
36311 * brig-builtins.def: New file.
36312 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
36313 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
36314 (DEF_HSAIL_SAT_BUILTIN): Likewise.
36315 (DEF_HSAIL_INTR_BUILTIN): Likewise.
36316 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
36317 * builtin-types.def (BT_INT8): New.
36318 (BT_INT16): Likewise.
36319 (BT_UINT8): Likewise.
36320 (BT_UINT16): Likewise.
36321 (BT_FN_ULONG): Likewise.
36322 (BT_FN_UINT_INT): Likewise.
36323 (BT_FN_UINT_ULONG): Likewise.
36324 (BT_FN_UINT_LONG): Likewise.
36325 (BT_FN_UINT_PTR): Likewise.
36326 (BT_FN_ULONG_PTR): Likewise.
36327 (BT_FN_INT8_FLOAT): Likewise.
36328 (BT_FN_INT16_FLOAT): Likewise.
36329 (BT_FN_UINT32_FLOAT): Likewise.
36330 (BT_FN_UINT16_FLOAT): Likewise.
36331 (BT_FN_UINT8_FLOAT): Likewise.
36332 (BT_FN_UINT64_FLOAT): Likewise.
36333 (BT_FN_UINT16_UINT32): Likewise.
36334 (BT_FN_UINT32_UINT16): Likewise.
36335 (BT_FN_UINT16_UINT16_UINT16): Likewise.
36336 (BT_FN_INT_PTR_INT): Likewise.
36337 (BT_FN_UINT_PTR_UINT): Likewise.
36338 (BT_FN_LONG_PTR_LONG): Likewise.
36339 (BT_FN_ULONG_PTR_ULONG): Likewise.
36340 (BT_FN_VOID_UINT64_UINT64): Likewise.
36341 (BT_FN_UINT8_UINT8_UINT8): Likewise.
36342 (BT_FN_INT8_INT8_INT8): Likewise.
36343 (BT_FN_INT16_INT16_INT16): Likewise.
36344 (BT_FN_INT_INT_INT): Likewise.
36345 (BT_FN_UINT_FLOAT_UINT): Likewise.
36346 (BT_FN_FLOAT_UINT_UINT): Likewise.
36347 (BT_FN_ULONG_UINT_UINT): Likewise.
36348 (BT_FN_ULONG_UINT_PTR): Likewise.
36349 (BT_FN_ULONG_ULONG_ULONG): Likewise.
36350 (BT_FN_UINT_UINT_UINT): Likewise.
36351 (BT_FN_VOID_UINT_PTR): Likewise.
36352 (BT_FN_UINT_UINT_PTR: Likewise.
36353 (BT_FN_UINT32_UINT64_PTR): Likewise.
36354 (BT_FN_INT_INT_UINT_UINT): Likewise.
36355 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
36356 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
36357 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
36358 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
36359 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
36360 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
36361 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
36362 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
36363 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
36364 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
36365 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
36366 * doc/frontends.texi: List BRIG FE.
36367 * doc/install.texi (Testing): Add BRIG tesring requirements.
36368 * doc/invoke.texi (Overall Options): Mention BRIG.
36369 * doc/standards.texi (Standards): Doucment BRIG HSA version.
36371 2017-01-24 Richard Biener <rguenther@suse.de>
36373 PR translation/79208
36374 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
36376 2017-01-24 Martin Jambor <mjambor@suse.cz>
36379 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
36380 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
36381 and known_contexts.
36383 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
36385 PR middle-end/79123
36386 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
36387 casts from signed to unsigned really don't have a range.
36389 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
36391 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
36392 GMP_RNDx for compatiblity.
36394 2017-01-24 Martin Liska <mliska@suse.cz>
36397 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
36398 that would prevent us to call alloca with -1 as argument.
36400 2017-01-24 Jakub Jelinek <jakub@redhat.com>
36402 * dwarf2out.c (output_compilation_unit_header, output_file_names):
36403 Avoid -Wformat-security warning.
36405 2017-01-23 Andrew Pinski <apinski@cavium.com>
36407 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
36410 2017-01-23 Martin Sebor <msebor@redhat.com>
36412 PR middle-end/78703
36413 * gimple-ssa-sprintf.c (warn_level): New global.
36414 (format_integer): Use it here and throughout the rest of the file.
36415 Use the same switch to compute sign as base.
36416 (maybe_warn): New function.
36417 (format_directive): Factor out warnings into maybe_warn.
36418 Add debugging output. Use warn_level.
36419 (add_bytes): Use warn_level.
36420 (pass_sprintf_length::compute_format_length): Add debugging output.
36421 (try_substitute_return_value): Same.
36422 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
36424 PR middle-end/78703
36425 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
36426 (struct fmtresult, format_integer, format_floating): Adjust.
36427 (fmtresult::fmtresult): Set max correctly in two argument ctor.
36428 (get_string_length, format_string,format_directive): Same.
36429 (pass_sprintf_length::compute_format_length): Same.
36430 (try_substitute_return_value): Simplify slightly.
36432 PR middle-end/78703
36433 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
36434 (fmtresult::operator+=): Outlined.
36435 (struct fmtresult): Add ctors.
36436 (struct conversion_spec): Rename...
36437 (struct directive): ...to this. Add and remove data members.
36438 (directive::set_width, directive::set_precision): New functions.
36439 (format_percent): Use fmtresult ctor.
36440 (get_width_and_precision): Remove.
36441 (format_integer): Make naming changes. Avoid computing width and
36443 (format_floating): Same. Adjust indentation.
36444 (format_character, format_none): New functions.
36445 (format_string): Moved character handling to format_character.
36446 (format_directive): Remove arguments, change return type.
36447 (parse_directive): New function.
36448 (pass_sprintf_length::compute_format_length): Move directive
36449 parsing to parse_directive.
36451 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36453 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
36454 (assign_assembler_name_if_needed): ... this.
36455 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
36456 (assign_assembler_name_if_needed): ... this.
36457 (free_lang_data_in_cgraph): Adjust callers.
36458 * cgraphunit.c (cgraph_node::analyze): Likewise.
36459 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
36462 2017-01-23 Richard Biener <rguenther@suse.de>
36464 PR tree-optimization/79088
36465 PR tree-optimization/79188
36466 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
36467 resetting loop bounds after last path deletion. Reset loop
36468 bounds of the target loop, make code match the comments.
36469 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
36470 Make sure loops need no fixups.
36472 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
36474 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
36475 exponent support with double type for first argument.
36476 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
36477 type returned by __builtin_vec_extract_sig,
36478 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
36479 functions from "vector int" to "vector unsigned int" or from
36480 "vector long long int" to "vector unsigned long long int".
36481 Changed type returned by __builtin_vec_extract_exp,
36482 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
36483 functions from "vector int" to "vector unsigned int" or from
36484 "vector long long int" to "vector unsigned long long int".
36485 Changed return type of __builtin_vec_test_data_class,
36486 __builtin_vec_test_data_class_sp, and
36487 __builtin_vec_test_data_class_dp from "vector int" to
36488 "vector bool int" or from "vector long long int" to "vector bool
36489 long long int" and changed second argument type from "unsigned
36490 int" to "int". Added new overloaded function forms "vector float
36491 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
36492 "vector float __builtin_vec_insert_exp_sp (vector float, vector
36493 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
36494 double, vector unsigned long long int)" and "vector double
36495 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
36496 long int)". Changed return type of
36497 __builtin_scalar_test_data_class and
36498 __builtin_scalar_test_data_class_sp and
36499 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
36500 int" and changed second argument from "unsigned int" to "int".
36501 Changed type returned by __builtin_scalar_test_neg,
36502 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
36503 from "int" to "bool int". Added new overloaded function form
36504 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
36505 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
36506 exponent double-precision with floating point first argument.
36507 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
36508 documentation of scalar_test_data_class, scalar_test_neg,
36509 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
36510 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
36511 vec_test_data_class built-in functions to reflect refinements in
36512 their type signatures.
36514 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
36516 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
36518 (aarch64_elf_asm_destructor): Likewise.
36520 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
36522 PR rtl-optimization/78634
36523 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
36524 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
36525 * ifcvt.c (noce_try_cmove): Add missing cost check.
36527 PR rtl-optimization/71724
36528 * combine.c (if_then_else_cond): Look for situations where it is
36529 beneficial to undo the work of one of the recursive calls.
36531 2017-01-23 Bin Cheng <bin.cheng@arm.com>
36533 PR tree-optimization/70754
36534 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
36535 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
36536 combined stmt before it if not NULL.
36537 (combine_chains): Process refs reversely and compute dominance point
36540 2017-01-23 Martin Liska <mliska@suse.cz>
36542 PR tree-optimization/79196
36543 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
36544 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
36546 (strlen_optimize_stmt): Call the renamed function.
36548 2017-01-23 Michael Matz <matz@suse.de>
36550 PR tree-optimization/78384
36551 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
36553 2017-01-23 Richard Biener <rguenther@suse.de>
36555 PR tree-optimization/79186
36556 * tree-vrp.c (register_new_assert_for): Make sure we've seen
36557 both incoming edges before moving an assert.
36559 2017-01-23 Martin Jambor <mjambor@suse.cz>
36561 * ipa-prop.c (load_from_param_1): Removed.
36562 (load_from_unmodified_param): Bits from load_from_param_1 put back
36564 (load_from_param): Removed.
36565 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
36566 with stmt. Reverted back to use of load_from_unmodified_param.
36568 2017-01-23 Martin Jambor <mjambor@suse.cz>
36571 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
36572 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
36573 field a pointer to garbage collected vector, mark lattices and
36574 ipcp_orig_node with GTY((skip)).
36575 (ipa_get_param_count): Adjust to descriptors being a pointer.
36576 (ipa_get_param): Likewise.
36577 (ipa_get_type): Likewise.
36578 (ipa_get_param_move_cost): Likewise.
36579 (ipa_set_param_used): Likewise.
36580 (ipa_get_controlled_uses): Likewise.
36581 (ipa_set_controlled_uses): Likewise.
36582 (ipa_is_param_used): Likewise.
36583 (ipa_node_params_t): Move into garbage collector. New methods insert
36585 (ipa_node_params_sum): Annotate wth GTY(()).
36586 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
36588 (ipa_load_from_parm_agg): Adjust declaration.
36589 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
36590 * ipa-profile.c (ipa_profile): Likewise.
36591 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
36592 (ipa_populate_param_decls): Make descriptors parameter garbage
36594 (ipa_dump_param): Adjust to descriptors being a pointer.
36595 (ipa_alloc_node_params): Likewise.
36596 (ipa_initialize_node_params): Likewise.
36597 (load_from_param_1): Make descriptors parameter garbage collected.
36598 (load_from_unmodified_param): Likewise.
36599 (load_from_param): Likewise.
36600 (ipa_load_from_parm_agg): Likewise.
36601 (ipa_node_params::~ipa_node_params): Removed.
36602 (ipa_free_all_node_params): Remove call to delete operator.
36603 (ipa_node_params_t::insert): New.
36604 (ipa_node_params_t::remove): Likewise.
36605 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
36606 copy known_csts and known_contexts vectors.
36607 (ipa_read_node_info): Adjust to descriptors being a pointer.
36608 (ipcp_modif_dom_walker): Make m_descriptors field garbage
36610 (ipcp_transform_function): Make descriptors variable garbage
36613 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
36615 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
36616 * config/i386/avx512dqintrin.h: Ditto.
36617 * config/i386/avx512fintrin.h: Ditto.
36618 * config/i386/i386.c: Handle new builtins.
36619 * config/i386/i386-builtin.def: Add new builtins.
36620 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
36621 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
36623 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36624 Martin Liska <mliska@suse.cz>
36626 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
36627 * asan.c (asan_expand_poison_ifn): Support stores and use
36628 appropriate ASAN report function.
36629 * internal-fn.c (expand_ASAN_POISON_USE): New function.
36630 * internal-fn.def (ASAN_POISON_USE): Declare.
36631 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
36632 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
36633 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
36634 ASAN_POISON calls w/o LHS.
36635 * tree-ssa.c (execute_update_addresses_taken): Create clobber
36636 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
36637 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
36638 * gimplify.c (asan_poison_variables): Add attribute
36639 use_after_scope_memory to variables that really needs to live
36641 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
36642 having the attribute.
36644 2017-01-23 Martin Liska <mliska@suse.cz>
36646 * asan.c (create_asan_shadow_var): New function.
36647 (asan_expand_poison_ifn): Likewise.
36648 * asan.h (asan_expand_poison_ifn): New declaration.
36649 * internal-fn.c (expand_ASAN_POISON): Likewise.
36650 * internal-fn.def (ASAN_POISON): New builtin.
36651 * sanopt.c (pass_sanopt::execute): Expand
36652 asan_expand_poison_ifn.
36653 * tree-inline.c (copy_decl_for_dup_finish): Make function
36655 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
36656 * tree-ssa.c (is_asan_mark_p): New function.
36657 (execute_update_addresses_taken): Rewrite local variables
36658 (identified just by use-after-scope as addressable) into SSA.
36660 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
36662 * doc/install.texi (Specific): opensource.apple.com uses https
36663 now. Remove trailing slash.
36665 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
36667 * README.Portability: Remove note on an Irix compatibility issue.
36669 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
36671 * gcov.c (INCLUDE_ALGORITHM): Define.
36672 (INCLUDE_VECTOR): Define.
36673 No longer include <vector> and <algorithm> directly.
36675 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
36677 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
36679 * doc/invoke.texi (Code Gen Options): Ditto.
36681 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
36684 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
36686 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
36688 rtl-optimization/79125
36689 * cprop.c (local_cprop_pass): Handle cases where we make an
36690 unconditional trap.
36692 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
36696 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
36697 read from, for big endian.
36699 2017-01-20 Jiong Wang <jiong.wang@arm.com>
36701 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
36702 register pauth builtins for LP64 only.
36704 2017-01-20 Marek Polacek <polacek@redhat.com>
36707 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
36710 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
36712 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
36714 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
36715 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
36716 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
36718 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36721 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
36722 in big-endian mode when they are not a single duplicated value.
36724 2017-01-20 Richard Biener <rguenther@suse.de>
36726 * BASE-VER: Bump to 7.0.1.
36728 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
36730 * omp-low.c (omplow_simd_context): New struct. Use it...
36731 (lower_rec_simd_input_clauses): ...here and...
36732 (lower_rec_input_clauses): ...here to hold common data. Adjust all
36733 references to idx, lane, max_vf, is_simt.
36735 2017-01-20 Graham Markall <graham.markall@embecosm.com>
36737 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
36740 2017-01-20 Martin Jambor <mjambor@suse.cz>
36742 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
36743 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
36745 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
36746 (GTFILES): Rename hsa.c to hsa-common.c.
36747 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
36748 * hsa-dump.c: Likewise.
36749 * hsa-gen.c: Likewise.
36750 * hsa-regalloc.c: Likewise.
36751 * ipa-hsa.c: Likewise.
36752 * omp-expand.c: Likewise.
36753 * omp-low.c: Likewise.
36754 * toplev.c: Likewise.
36756 2017-01-20 Marek Polacek <polacek@redhat.com>
36759 * doc/invoke.texi: Document -Wduplicated-branches.
36760 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
36761 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
36762 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
36763 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
36764 return 0 only when not OEP_LEXICOGRAPHIC.
36765 (fold_build_cleanup_point_expr): Use the expression
36766 location when building CLEANUP_POINT_EXPR.
36767 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
36768 * tree.c (add_expr): Handle error_mark_node.
36770 2017-01-20 Martin Liska <mliska@suse.cz>
36773 * tree-profile.c (init_ic_make_global_vars): Do not call
36775 (gimple_init_gcov_profiler): Likewise.
36777 2017-01-20 Martin Liska <mliska@suse.cz>
36780 * cgraph.h (maybe_create_reference): Remove argument and
36782 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
36784 * ipa-cp.c (create_specialized_node): Likewise.
36785 * symtab.c (symtab_node::maybe_create_reference): Handle
36786 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
36788 2017-01-20 Martin Liska <mliska@suse.cz>
36790 * read-rtl-function.c (function_reader::create_function): Use
36791 build_decl instread of build_decl_stat.
36793 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
36795 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
36796 * config/i386/avx512dqintrin.h: Ditto.
36797 * config/i386/avx512fintrin.h: Ditto.
36798 * config/i386/i386-builtin-types.def: Add new types.
36799 * config/i386/i386.c: Handle new types.
36800 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
36801 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
36802 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
36803 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
36804 (__builtin_ia32_kshiftridi): New.
36805 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
36807 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
36811 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
36812 define to rs6000_init_stack_protect_guard.
36813 (rs6000_init_stack_protect_guard): New function.
36815 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36816 Yunqiang Su <yunqiang.su@imgtec.com>
36818 * config.gcc (supported_defaults): Add madd4.
36819 (with_madd4): Add validation.
36820 (all_defaults): Add madd4.
36821 * config/mips/mips.opt (mmadd4): New option.
36822 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
36824 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
36826 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
36827 (ISA_HAS_FUSED_MADD4): Likewise.
36828 * doc/invoke.texi (-mmadd4): Document the new option.
36829 * doc/install.texi (--with-madd4): Document the new option.
36831 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36833 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
36834 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
36835 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
36836 (aarch64_init_pauth_hint_builtins): New.
36837 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
36838 (aarch64_expand_builtin): Expand new builtins.
36840 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36842 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
36843 * combine-stack-adj.c (no_unhandled_cfa): Handle
36844 REG_CFA_TOGGLE_RA_MANGLE.
36845 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
36846 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
36847 info for return address signing.
36848 (aarch64_expand_epilogue): Likewise.
36850 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36852 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
36853 * config/aarch64/aarch64-protos.h
36854 (aarch64_return_address_signing_enabled): New declaration.
36855 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
36857 (aarch64_expand_prologue): Sign return address before it's pushed onto
36859 (aarch64_expand_epilogue): Authenticate return address fetched from
36861 (aarch64_override_options): Sanity check for ILP32 and ISA level.
36862 (aarch64_attributes): New function attributes for "sign-return-address".
36863 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
36864 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
36865 ("*do_return"): Generate combined instructions according to key index.
36866 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
36867 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
36869 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
36870 * config/aarch64/aarch64.opt (msign-return-address=): New.
36871 * doc/extend.texi (AArch64 Function Attributes): Documents
36872 "sign-return-address=".
36873 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
36875 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36877 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
36878 overall option summary.
36880 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36882 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
36883 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
36884 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
36885 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
36887 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
36889 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
36890 -mpower9-minmax by default for -mcpu=power9.
36891 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
36892 128-bit floating point.
36894 2017-01-20 Alan Modra <amodra@gmail.com>
36896 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
36897 optimizing for size.
36899 2017-01-20 Alan Modra <amodra@gmail.com>
36902 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
36903 for strcmp and strncmp from corresponding builtin decl.
36905 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36907 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
36908 instead of i386/rtems-64.h.
36909 * config/i386/rtems-64.h: Remove.
36911 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36915 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
36917 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
36919 2017-01-19 Tamar Christina <tamar.christina@arm.com>
36921 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
36922 Change int to HOST_WIDE_INT.
36923 * config/aarch64/aarch64-protos.h
36924 (aarch64_simd_gen_const_vector_dup): Likewise.
36925 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
36927 2017-01-19 David Malcolm <dmalcolm@redhat.com>
36929 * langhooks-def.h (lhd_type_for_size): New decl.
36930 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
36931 * langhooks.c (lhd_type_for_size): New function, taken from
36934 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
36936 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
36937 define_bypass for CR latency.
36938 (power9-cracked-alu): Update bypass latency and remove power9-branch.
36939 (power9-alu2): Add define_bypass for CR latency.
36941 (power9-mul): Update insn latency.
36942 (power9-mul-compare): Update insn latency, bypass latency and remove
36945 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36947 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
36949 * config/aarch64/aarch64.md
36950 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
36951 aarch64_nopcrelative_literal_loads.
36952 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
36954 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
36956 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
36957 TARGET_LOONGSON_3A.
36958 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
36960 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
36963 * config.gcc (supported_defaults): Add lxc1-sxc1.
36964 (with_lxc1_sxc1): Add validation.
36965 (all_defaults): Add lxc1-sxc1.
36966 * config/mips/mips.opt (mlxc1-sxc1): New option.
36967 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
36969 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
36970 __mips_no_lxc1_sxc1.
36971 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
36972 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
36973 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
36975 2017-01-19 Richard Biener <rguenther@suse.de>
36977 PR tree-optimization/72488
36978 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
36979 sure to restore SSA info.
36980 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
36982 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
36984 PR rtl-optimization/79121
36985 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
36986 of the inner type when shifting an extended value.
36988 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
36991 * symtab.c (symtab_node::equal_address_to): Fix comparing of
36992 interposable aliases.
36994 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
36997 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
36998 Use the evmergelohi instruction.
36999 (mov_si<mode>_e500_subreg4_2_le): Likewise.
37000 (mov_sitf_e500_subreg8_2_be): Likewise.
37001 (mov_sitf_e500_subreg12_2_le): Likewise.
37002 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
37003 (mov_si<mode>_e500_subreg4_2_be): Likewise.
37004 (mov_sitf_e500_subreg8_2_le): Likewise.
37005 (mov_sitf_e500_subreg12_2_be): Likewise.
37007 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37009 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
37010 attribute from vecsimple to vecperm.
37011 (altivec_vbpermq2): Likewise.
37013 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37016 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
37018 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37019 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
37020 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
37021 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
37022 case where N arg is SIZE_MAX.
37023 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
37024 (cmpstrsi): Add pattern.
37026 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
37028 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
37029 __builtin_vec_revb builtins.
37030 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
37031 built-in functions to support generation of the ISA 3.0 XXBR<x>
37032 vector byte reverse instructions.
37033 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
37034 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
37035 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
37036 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
37037 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
37038 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
37039 (P9V_BUILTIN_VEC_REVB): Likewise.
37040 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
37041 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
37042 (p9_xxbrq_v16qi): Likewise.
37043 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
37044 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
37045 (p9_xxbrh_v8hi): Likewise.
37046 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
37047 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
37048 vec_revb built-in functions.
37050 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
37052 PR rtl-optimization/78952
37053 * config/i386/i386.md (any_extract): New code iterator.
37054 (*insvqi_2): Use any_extract for source operand.
37055 (*insvqi_3): Use any_shiftrt for source operand.
37057 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
37059 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
37061 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
37063 2017-01-18 Matthias Klose <doko@ubuntu.com>
37065 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
37067 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37069 * config/rs6000/altivec.h (vec_bperm): Change #define.
37070 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
37071 (altivec_vbpermq2): New define_insn.
37072 (altivec_vbpermd): Likewise.
37073 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
37074 function interface.
37075 (VBPERMD): Likewise.
37076 (VBPERM): New polymorphic function interface.
37077 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
37078 Add entries for P9V_BUILTIN_VEC_VBPERM.
37079 * doc/extend.texi: Add interfaces for vec_bperm.
37081 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37083 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
37084 first letter of error messages.
37085 (s390_resolve_overloaded_builtin): Likewise.
37086 * config/s390/s390.c (s390_expand_builtin): Likewise.
37087 (s390_invalid_arg_for_unprototyped_fn): Likewise.
37088 (s390_valid_target_attribute_inner_p): Likewise.
37089 * config/s390/s390.md ("tabort"): Likewise.
37091 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
37093 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
37094 (ISA_AVOID_DIV_HILO): New macro.
37095 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
37096 (ISA_HAS_DDIV): Likewise.
37098 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37100 * doc/invoke.texi (fabi-version): Correct number of occurrences.
37102 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37104 * doc/invoke.texi (fabi-version): Spelling fix.
37106 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37109 * doc/invoke.texi (fabi-version): Mention mangling fix for
37112 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37115 * doc/invoke.texi (fabi-version): Document discriminator mangling.
37117 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
37120 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
37121 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
37123 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
37125 (stack_protect_test): Ditto.
37126 * config/rs6000/rs6000.opt (mstack-protector-guard=,
37127 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
37129 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
37130 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
37131 -mstack-protector-guard-offset=.
37132 (RS/6000 and PowerPC Options): Ditto.
37134 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37136 * config/i386/i386.h (MASK_CLASS_P): New define.
37137 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
37138 there are no registers from different register sets also when
37139 mask registers are used. Update function comment.
37140 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
37141 to (*k/*r) and (*k/*km) alternatives.
37143 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
37145 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
37146 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
37147 (EH_RETURN_HANDLER_RTX): New define.
37148 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
37149 Force frame pointer in EH return functions.
37150 (aarch64_expand_epilogue): Add barrier for eh_return.
37151 (aarch64_final_eh_return_addr): Remove.
37152 (aarch64_eh_return_handler_rtx): New function.
37153 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
37155 (aarch64_eh_return_handler_rtx): New prototype.
37157 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37159 * config/rs6000/altivec.h (vec_rlmi): New #define.
37160 (vec_vrlnm): Likewise.
37161 (vec_rlnm): Likewise.
37162 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
37163 (UNSPEC_VRLNM): Likewise.
37164 (VIlong): New mode iterator.
37165 (altivec_vrl<VI_char>mi): New define_insn.
37166 (altivec_vrl<VI_char>nm): Likewise.
37167 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
37169 (VRLDNM): Likewise.
37170 (RLNM): New polymorphic function entry.
37171 (VRLWMI): New monomorphic function entry.
37172 (VRLDMI): Likewise.
37173 (RLMI): New polymorphic function entry.
37174 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
37175 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
37176 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
37179 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37182 * dwarf2out.c (field_byte_offset): Restore the
37183 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
37184 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
37185 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
37188 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37191 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
37193 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37196 * dwarf2out.c (add_data_member_location_attribute): For constant
37197 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
37198 instead of DW_AT_data_member_location, DW_AT_bit_offset and
37199 DW_AT_byte_size attributes.
37201 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37203 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
37204 after forcing to constant memory when the code model is medium.
37206 2017-01-17 Julia Koval <julia.koval@intel.com>
37209 * config/i386/avx512fintrin.h
37210 (_mm512_i32gather_ps): Change __addr type to void const*.
37211 (_mm512_mask_i32gather_ps): Ditto.
37212 (_mm512_i32gather_pd): Ditto.
37213 (_mm512_mask_i32gather_pd): Ditto.
37214 (_mm512_i64gather_ps): Ditto.
37215 (_mm512_mask_i64gather_ps): Ditto.
37216 (_mm512_i64gather_pd): Ditto.
37217 (_mm512_mask_i64gather_pd): Ditto.
37218 (_mm512_i32gather_epi32): Ditto.
37219 (_mm512_mask_i32gather_epi32): Ditto.
37220 (_mm512_i32gather_epi64): Ditto.
37221 (_mm512_mask_i32gather_epi64): Ditto.
37222 (_mm512_i64gather_epi32): Ditto.
37223 (_mm512_mask_i64gather_epi32): Ditto.
37224 (_mm512_i64gather_epi64): Ditto.
37225 (_mm512_mask_i64gather_epi64): Ditto.
37226 (_mm512_i32scatter_ps): Change __addr type to void*.
37227 (_mm512_mask_i32scatter_ps): Ditto.
37228 (_mm512_i32scatter_pd): Ditto.
37229 (_mm512_mask_i32scatter_pd): Ditto.
37230 (_mm512_i64scatter_ps): Ditto.
37231 (_mm512_mask_i64scatter_ps): Ditto.
37232 (_mm512_i64scatter_pd): Ditto.
37233 (_mm512_mask_i64scatter_pd): Ditto.
37234 (_mm512_i32scatter_epi32): Ditto.
37235 (_mm512_mask_i32scatter_epi32): Ditto.
37236 (_mm512_i32scatter_epi64): Ditto.
37237 (_mm512_mask_i32scatter_epi64): Ditto.
37238 (_mm512_i64scatter_epi32): Ditto.
37239 (_mm512_mask_i64scatter_epi32): Ditto.
37240 (_mm512_i64scatter_epi64): Ditto.
37241 (_mm512_mask_i64scatter_epi64): Ditto.
37242 * config/i386/avx512pfintrin.h
37243 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
37244 (_mm512_mask_prefetch_i32gather_ps): Ditto.
37245 (_mm512_mask_prefetch_i64gather_pd): Ditto.
37246 (_mm512_mask_prefetch_i64gather_ps): Ditto.
37247 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
37248 (_mm512_prefetch_i32scatter_ps): Ditto.
37249 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
37250 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
37251 (_mm512_prefetch_i64scatter_pd): Ditto.
37252 (_mm512_prefetch_i64scatter_ps): Ditto.
37253 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
37254 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
37255 * config/i386/avx512vlintrin.h
37256 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
37257 (_mm_mmask_i32gather_ps): Ditto.
37258 (_mm256_mmask_i32gather_pd): Ditto.
37259 (_mm_mmask_i32gather_pd): Ditto.
37260 (_mm256_mmask_i64gather_ps): Ditto.
37261 (_mm_mmask_i64gather_ps): Ditto.
37262 (_mm256_mmask_i64gather_pd): Ditto.
37263 (_mm_mmask_i64gather_pd): Ditto.
37264 (_mm256_mmask_i32gather_epi32): Ditto.
37265 (_mm_mmask_i32gather_epi32): Ditto.
37266 (_mm256_mmask_i32gather_epi64): Ditto.
37267 (_mm_mmask_i32gather_epi64): Ditto.
37268 (_mm256_mmask_i64gather_epi32): Ditto.
37269 (_mm_mmask_i64gather_epi32): Ditto.
37270 (_mm256_mmask_i64gather_epi64): Ditto.
37271 (_mm_mmask_i64gather_epi64): Ditto.
37272 (_mm256_i32scatter_ps): Change __addr type to void*.
37273 (_mm256_mask_i32scatter_ps): Ditto.
37274 (_mm_i32scatter_ps): Ditto.
37275 (_mm_mask_i32scatter_ps): Ditto.
37276 (_mm256_i32scatter_pd): Ditto.
37277 (_mm256_mask_i32scatter_pd): Ditto.
37278 (_mm_i32scatter_pd): Ditto.
37279 (_mm_mask_i32scatter_pd): Ditto.
37280 (_mm256_i64scatter_ps): Ditto.
37281 (_mm256_mask_i64scatter_ps): Ditto.
37282 (_mm_i64scatter_ps): Ditto.
37283 (_mm_mask_i64scatter_ps): Ditto.
37284 (_mm256_i64scatter_pd): Ditto.
37285 (_mm256_mask_i64scatter_pd): Ditto.
37286 (_mm_i64scatter_pd): Ditto.
37287 (_mm_mask_i64scatter_pd): Ditto.
37288 (_mm256_i32scatter_epi32): Ditto.
37289 (_mm256_mask_i32scatter_epi32): Ditto.
37290 (_mm_i32scatter_epi32): Ditto.
37291 (_mm_mask_i32scatter_epi32): Ditto.
37292 (_mm256_i32scatter_epi64): Ditto.
37293 (_mm256_mask_i32scatter_epi64): Ditto.
37294 (_mm_i32scatter_epi64): Ditto.
37295 (_mm_mask_i32scatter_epi64): Ditto.
37296 (_mm256_i64scatter_epi32): Ditto.
37297 (_mm256_mask_i64scatter_epi32): Ditto.
37298 (_mm_i64scatter_epi32): Ditto.
37299 (_mm_mask_i64scatter_epi32): Ditto.
37300 (_mm256_i64scatter_epi64): Ditto.
37301 (_mm256_mask_i64scatter_epi64): Ditto.
37302 (_mm_i64scatter_epi64): Ditto.
37303 (_mm_mask_i64scatter_epi64): Ditto.
37304 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
37305 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
37306 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
37307 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
37308 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
37309 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
37310 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
37311 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
37312 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
37313 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
37314 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
37315 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
37316 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
37317 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
37318 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
37319 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
37320 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
37321 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
37322 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
37323 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
37324 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
37325 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
37326 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
37327 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
37328 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
37329 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
37330 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
37331 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
37332 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
37333 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
37334 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
37335 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
37336 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
37337 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
37338 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
37339 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
37340 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
37341 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
37342 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
37343 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
37344 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
37345 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
37346 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
37347 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
37348 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
37349 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
37350 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
37351 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
37352 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
37353 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
37354 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
37355 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
37356 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
37357 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
37358 definitions accordingly.
37360 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
37361 Kuan-Lin Chen <kuanlinchentw@gmail.com>
37364 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
37367 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
37370 * ira-conflicts.c (ira_build_conflicts): Update total conflict
37371 hard regs for inner regno.
37373 2017-01-17 Martin Liska <mliska@suse.cz>
37376 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
37377 assumption and add comment.
37379 2017-01-17 Nathan Sidwell <nathan@acm.org>
37381 * ipa-visibility.c (localize_node): New function, broken out of ...
37382 (function_and_variable_visibility): ... here. Call it.
37384 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
37386 PR middle-end/77445
37387 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
37388 correctly set frequency of oudgoing edge.
37389 (duplicate_thread_path): Fix profile updating.
37391 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37394 * configure.ac: Add GCC_BASE_VER.
37395 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
37396 version from BASE-VER file.
37397 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
37398 (gcc.o): Depend on $(BASEVER).
37399 * common.opt (dumpfullversion): New option.
37400 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
37401 * doc/invoke.texi: Document -dumpfullversion.
37402 * doc/install.texi: Document --with-gcc-major-version-only.
37403 * configure: Regenerated.
37405 2017-01-17 Richard Biener <rguenther@suse.de>
37407 PR tree-optimization/71433
37408 * tree-vrp.c (register_new_assert_for): Merge same asserts
37409 on all incoming edges.
37410 (process_assert_insertions_for): Handle insertions at the
37413 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
37415 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
37416 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
37418 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
37421 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
37424 2017-01-17 Alan Modra <amodra@gmail.com>
37427 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
37428 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
37429 symbolic stack limit when pic.
37431 2017-01-16 Martin Sebor <msebor@redhat.com>
37433 PR tree-optimization/78608
37434 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
37436 2017-01-16 Jeff Law <law@redhat.com>
37439 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
37440 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
37441 for several include directories that may be relative to sysroot.
37442 * config/i386/x-mingw32 (gplus_includedir): Define.
37443 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
37444 (native_system_includedir): Likewise.
37445 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
37446 override if TARGET_SYSTEM_ROOT is defined.
37447 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
37449 PR tree-optimization/79090
37450 PR tree-optimization/33562
37451 PR tree-optimization/61912
37452 PR tree-optimization/77485
37453 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
37454 and computed trims into the dump file.
37456 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37458 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
37460 2017-01-16 Jakub Jelinek <jakub@redhat.com>
37463 * gimplify.c (gimplify_init_constructor): If want_value and
37464 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
37468 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
37469 sequence. Formatting fixes.
37470 (doloop_optimize): Formatting fixes.
37473 * gcc.c (debug_level_greater_than_spec_func): New function.
37474 (static_spec_functions): Add debug-level-gt spec function.
37475 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
37477 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37478 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37479 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
37480 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
37481 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
37482 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
37484 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
37486 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
37487 QImode fixups to general and mask registers only.
37489 2017-01-16 Carl Love <cel@us.ibm.com>
37491 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
37492 for built-in functions
37493 vector signed char vec_nabs (vector signed char)
37494 vector signed short vec_nabs (vector signed short)
37495 vector signed int vec_nabs (vector signed int)
37496 vector signed long long vec_nabs (vector signed long long)
37497 vector float vec_nabs (vector float)
37498 vector double vec_nabs (vector double)
37499 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
37501 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
37502 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
37503 * doc/extend.texi: Update the documentation file for the new built-in
37506 2017-01-16 Martin Sebor <msebor@redhat.com>
37508 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
37511 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37513 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
37514 UNSPEC_VSX__XXSPLTD to require special splat handling.
37516 2017-01-16 David Malcolm <dmalcolm@redhat.com>
37519 * system.h: Poison strndup.
37521 2017-01-16 Alan Modra <amodra@gmail.com>
37524 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
37527 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
37529 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
37531 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
37533 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
37534 call recog here. Assert that INSN_CODE (insn) is non-negative.
37536 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37539 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
37541 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
37542 in the currently scheduled RTL fragment.
37544 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37546 PR rtl-optimization/78751
37547 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
37550 2017-01-14 Jeff Law <law@redhat.com>
37552 PR tree-optimization/79090
37553 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
37554 variable length stores.
37555 (compute_trims): Delete dead assignment to *trim_tail.
37556 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
37559 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
37561 PR rtl-optimization/78626
37562 PR rtl-optimization/78727
37563 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
37564 of a block, and split such blocks after everything else is finished.
37566 2017-01-14 Alan Modra <amodra@gmail.com>
37569 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
37570 target legitimate_combined_insn.
37571 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
37572 (rs6000_legitimate_combined_insn): New function.
37573 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
37575 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
37576 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
37577 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
37579 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
37581 * doc/frontends.texi (G++ and GCC): Remove references to Java.
37583 2017-01-13 Jeff Law <law@redhat.com>
37585 PR tree-optimization/33562
37586 PR tree-optimization/61912
37587 PR tree-optimization/77485
37588 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
37590 (delete_dead_assignment): Likewise.
37591 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
37592 statement to delete_dead_call and delete_dead_assignment.
37594 2017-01-13 David Malcolm <dmalcolm@redhat.com>
37597 * substring-locations.c (format_warning_va): Strengthen case 1 so
37598 that both endpoints of the substring must be within the format
37599 range for just the substring to be printed.
37601 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
37603 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
37604 * config/i386/i386.c (ix86_target_string): Add missing options
37605 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
37606 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
37607 flags_other and ix86_target_other to flags2_other. Display unknown
37609 (ix86_valid_target_attribute_inner_p): Add missing options and
37610 reorder options by implied ISAs, as in ix86_target_string.
37612 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
37614 * hash-table.h (hash_table::too_empty_p): New function.
37615 (hash_table::expand): Use it.
37616 (hash_table::traverse): Likewise.
37617 (hash_table::empty_slot): Use sizeof (value_type) instead of
37618 sizeof (PTR) to convert bytes to elements. Shrink the table
37619 if the current size is excessive for the current number of
37622 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
37624 * ira-costs.c (record_reg_classes): Break from the inner loop
37625 early once alt_fail is known to be true. Update outer loop
37626 handling accordingly.
37628 2017-01-13 Jeff Law <law@redhat.com>
37630 * tree-ssa-dse.c (decrement_count): New function.
37631 (increment_start_addr, maybe_trim_memstar_call): Likewise.
37632 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
37633 when we know the partially dead statement is a mem* function.
37635 PR tree-optimization/61912
37636 PR tree-optimization/77485
37637 * tree-ssa-dse.c: Include expr.h.
37638 (maybe_trim_constructor_store): New function.
37639 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
37641 PR tree-optimization/33562
37642 PR tree-optimization/61912
37643 PR tree-optimization/77485
37644 * doc/invoke.texi: Document new dse-max-object-size param.
37645 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
37646 * tree-ssa-dse.c: Include params.h.
37647 (dse_store_status): New enum.
37648 (initialize_ao_ref_for_dse): New, partially extracted from
37650 (valid_ao_ref_for_dse, normalize_ref): New.
37651 (setup_live_bytes_from_ref, compute_trims): Likewise.
37652 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
37653 (maybe_trim_partially_dead_store): Likewise.
37654 (maybe_trim_complex_store): Likewise.
37655 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
37656 Track what bytes live from the original store. Return tri-state
37657 for dead, partially dead or live.
37658 (dse_dom_walker): Add constructor, destructor and new private members.
37659 (delete_dead_call, delete_dead_assignment): New extracted from
37661 (dse_optimize_stmt): Make a member of dse_dom_walker.
37662 Use initialize_ao_ref_for_dse.
37664 PR tree-optimization/33562
37665 PR tree-optimization/61912
37666 PR tree-optimization/77485
37667 * sbitmap.h (bitmap_count_bits): Prototype.
37668 (bitmap_clear_range, bitmap_set_range): Likewise.
37669 * sbitmap.c (bitmap_clear_range): New function.
37670 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
37672 2017-01-13 Martin Liska <mliska@suse.cz>
37675 * function.c (set_cfun): Add new argument force.
37676 * function.h (set_cfun): Likewise.
37677 * ipa-inline-transform.c (inline_call): Use the function when
37678 strict alising from is dropped for function we inline to.
37680 2017-01-13 Richard Biener <rguenther@suse.de>
37682 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
37683 for dumping GIMPLE INTEGER_CSTs.
37685 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
37687 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
37688 to 201112L since C++17.
37690 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
37693 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
37694 if -fsanitize=kernel-address is present.
37696 2017-01-13 Richard Biener <rguenther@suse.de>
37698 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
37699 as _Literal ( type ) number in case usual suffixes do not
37700 preserve all information.
37702 2017-01-13 Richard Biener <rguenther@suse.de>
37704 PR tree-optimization/77283
37705 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
37706 and ssa-iterators.h.
37707 (is_feasible_trace): Implement a cost model based on joiner
37710 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
37713 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
37714 char or short to __float128/_Float128 directly.
37716 2017-01-12 Martin Sebor <msebor@redhat.com>
37718 to -Wformat-overflow.
37719 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
37720 (min_bytes_remaining): Same.
37721 (get_string_length): Same.
37722 (format_string): Same.
37723 (format_directive): Same.
37725 (pass_sprintf_length::handle_gimple_call): Same.
37727 2017-01-12 Jakub Jelinek <jakub@redhat.com>
37729 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
37730 info.nowrite calls with no lhs that can't throw. Return bool
37731 whether gsi_remove has been called or not.
37732 (pass_sprintf_length::handle_gimple_call): Return bool whether
37733 try_substitute_return_value called gsi_remove. Formatting fix.
37734 (pass_sprintf_length::execute): Don't use gsi_remove if
37735 handle_gimple_call returned true.
37738 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
37739 be removed due to side-effects, don't remove following barrier nor
37740 turn the successor edge into fallthru edge.
37742 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37745 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
37746 element-reversing loads and stores as not swappable.
37748 2017-01-12 Nathan Sidwell <nathan@acm.org>
37749 Nicolai Stange <nicstange@gmail.com>
37751 * combine.c (try_combine): Don't ignore result of overlap checking
37752 loop. Combine overlap & asm check into single loop.
37754 2017-01-12 Richard Biener <rguenther@suse.de>
37756 * tree-pretty-print.c (dump_generic_node): Provide -gimple
37757 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
37759 2017-01-12 Richard Biener <rguenther@suse.de>
37761 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
37762 and TS_TARGET_OPTION directly derive from TS_BASE.
37763 * tree-core.h (tree_optimization_option): Derive from tree_base.
37764 (tree_target_option): Likewise.
37766 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
37768 * config/i386/i386.c (memory_address_length): Increase len
37769 only when rip_relative_addr_p returns false.
37771 2017-01-11 Julia Koval <julia.koval@intel.com>
37773 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
37774 (OPTION_MASK_ISA_SGX_SET): New.
37775 (ix86_handle_option): Handle OPT_msgx.
37776 * config.gcc: Added sgxintrin.h.
37777 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
37778 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
37779 * config/i386/i386.c (ix86_target_string): Add -msgx.
37781 (ix86_option_override_internal): Handle new options.
37782 (ix86_valid_target_attribute_inner_p): Add sgx.
37783 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
37784 * config/i386/i386.opt: Add msgx.
37785 * config/i386/sgxintrin.h: New file.
37786 * config/i386/x86intrin.h: Add sgxintrin.h.
37788 2017-01-11 Jakub Jelinek <jakub@redhat.com>
37791 * fold-const.c (maybe_nonzero_address): Return 1 for function
37793 (tree_single_nonzero_warnv_p): Don't handle function local objects
37797 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
37800 2017-01-11 David Malcolm <dmalcolm@redhat.com>
37803 * opts.c: Include "spellcheck.h"
37804 (struct string_fragment): New struct.
37805 (struct edit_distance_traits<const string_fragment &>): New
37807 (get_closest_sanitizer_option): New function.
37808 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
37810 2017-01-11 Jakub Jelinek <jakub@redhat.com>
37812 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
37814 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
37815 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
37816 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
37817 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
37818 for initial die_offset if dwarf_split_debug_info.
37819 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
37820 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
37822 (output_skeleton_debug_sections): Formatting fix. Use
37823 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
37824 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
37826 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
37828 * config/arm/cortex-a53.md: Add bypasses for
37829 cortex_a53_r2f_cvt.
37830 (cortex_a53_r2f): Only use for transfers.
37831 (cortex_a53_f2r): Likewise.
37832 (cortex_a53_r2f_cvt): Add reservation for conversions.
37833 (cortex_a53_f2r_cvt): Likewise.
37835 2017-01-11 Tamar Christina <tamar.christina@arm.com>
37837 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
37838 to all inlined functions, change static to extern.
37840 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
37843 * config/arm/arm.c (legitimize_pic_address): Handle reference to
37845 (arm_assemble_integer): Likewise.
37847 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
37849 * config.gcc: Use new awk script to check CPU, FPU and architecture
37850 parameters for --with-... options.
37851 * config/arm/parsecpu.awk: New file
37852 * config/arm/arm-cpus.in: New file.
37853 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
37855 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
37857 * config/arm/t-arm: Update dependency rules.
37858 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
37859 of processing .def files.
37860 * config/arm/genopt.sh: Deleted.
37861 * config/arm/gentune.sh: Deleted.
37862 * config/arm/arm-cores.def: Deleted.
37863 * config/arm/arm-arches.def: Deleted.
37864 * config/arm/arm-fpus.def: Deleted.
37865 * config/arm/arm-tune.md: Regenerated.
37866 * config/arm/arm-tables.opt: Regenerated.
37867 * config/arm/arm-cpu.h: New generated file.
37868 * config/arm/arm-cpu-data.h: New generated file.
37869 * config/arm/arm-cpu-cdata.h: New generated file.
37871 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
37874 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
37876 (input_varpool_node): Unpack dynamically_initialized bit.
37878 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
37880 PR rtl-optimization/79032
37881 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
37882 the alignment of the adjusted memory reference against that of MODE,
37883 instead of the alignment of the original memory reference.
37885 2017-01-11 Martin Jambor <mjambor@suse.cz>
37887 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
37889 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
37890 decorated functions.
37892 2017-01-11 Richard Biener <rguenther@suse.de>
37894 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
37895 set range/nonnull info for PHI results. Do not set it on
37896 stmts marked for removal.
37898 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
37900 * expr.c (store_field): In the bitfield case, fetch the return value
37901 from the registers before applying a single big-endian adjustment.
37902 Always do a final load for a BLKmode value not larger than a word.
37904 2017-01-10 David Malcolm <dmalcolm@redhat.com>
37907 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
37908 that we correctly handle column numbers greater than
37909 LINE_MAP_MAX_COLUMN_NUMBER.
37911 2017-01-10 Martin Sebor <msebor@redhat.com>
37913 PR middle-end/78245
37914 * gimple-ssa-sprintf.c (get_destination_size): Call
37915 {init,fini}object_sizes.
37916 * tree-object-size.c (addr_object_size): Adjust.
37917 (pass_through_call): Adjust.
37918 (pass_object_sizes::execute): Adjust.
37919 * tree-object-size.h (fini_object_sizes): Declare.
37921 2017-01-10 Martin Sebor <msebor@redhat.com>
37923 PR tree-optimization/78775
37924 * builtins.c (get_size_range): Move...
37925 * calls.c: ...to here.
37926 (alloc_max_size): Accept zero argument.
37927 (operand_signed_p): Remove.
37928 (maybe_warn_alloc_args_overflow): Call get_size_range.
37929 * calls.h (get_size_range): Declare.
37931 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
37933 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
37934 from TI's devices.csv file as of September 2016.
37935 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
37937 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
37939 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
37940 * doc/invoke.texi: Likewise.
37941 * doc/md.texi: Likewise.
37942 * doc/objc.texi: Likewise.
37944 2017-01-10 Joshua Conner <joshconner@google.com>
37946 * config/arm/fuchsia-elf.h: New file.
37947 * config/fuchsia.h: New file.
37948 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
37949 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
37951 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
37953 2016-01-10 Richard Biener <rguenther@suse.de>
37955 PR tree-optimization/79034
37956 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
37957 Propagate out degenerate PHIs in the joiner.
37959 2017-01-10 Martin Liska <mliska@suse.cz>
37961 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
37962 (sort_congruence_classes_by_decl_uid): Likewise.
37963 (sort_congruence_class_groups_by_decl_uid): Likewise.
37964 (sem_item_optimizer::merge_classes): Sort class, groups in these
37965 classes and members in the groups by DECL_UID of declarations.
37966 This would make merge operations stable.
37968 2017-01-10 Martin Liska <mliska@suse.cz>
37970 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
37971 usage of m_classes_vec.
37972 (sem_item_optimizer::~sem_item_optimizer): Likewise.
37973 (sem_item_optimizer::get_group_by_hash): Likewise.
37974 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
37975 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
37976 (sem_item_optimizer::verify_classes): Likewise.
37977 (sem_item_optimizer::process_cong_reduction): Likewise.
37978 (sem_item_optimizer::dump_cong_classes): Likewise.
37979 (sem_item_optimizer::merge_classes): Likewise.
37980 * ipa-icf.h (congruence_class_hash): Rename from
37981 congruence_class_group_hash. Remove declaration of m_classes_vec.
37983 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
37985 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
37986 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
37987 * config.gcc: Add avx512vpopcntdqintrin.h.
37988 * config/i386/avx512vpopcntdqintrin.h: New.
37989 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
37990 * config/i386/i386-builtin-types.def: Add new types.
37991 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
37992 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
37993 __builtin_ia32_vpopcountq_v8di_mask): New.
37994 * config/i386/i386-c.c (ix86_target_macros_internal): Define
37995 __AVX512VPOPCNTDQ__.
37996 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
37997 (PTA_AVX512VPOPCNTDQ): Define.
37998 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
37999 TARGET_AVX512VPOPCNTDQ_P): Define.
38000 * config/i386/i386.opt: Add mavx512vpopcntdq.
38001 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
38002 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
38004 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38006 PR middle-end/77484
38007 * predict.def (PRED_CALL): Set to 67.
38009 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
38011 * expr.c (store_field): In the bitfield case, if the value comes from
38012 a function call and is of an aggregate type returned in registers, do
38013 not modify the field mode; extract the value in all cases if the mode
38014 is BLKmode and the size is not larger than a word.
38016 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
38019 * config/i386/cpuid.h: Fix undefined behavior.
38021 2017-01-04 Jeff Law <law@redhat.com>
38023 PR tree-optimization/79007
38024 PR tree-optimization/67955
38025 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
38026 conservative for pt.null when flag_non_call_exceptions is on.
38028 2017-01-09 Jakub Jelinek <jakub@redhat.com>
38030 PR translation/79019
38031 PR translation/79020
38032 * params.def (PARAM_INLINE_MIN_SPEEDUP,
38033 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
38034 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
38036 * config/avr/avr.opt (maccumulate-args): Likewise.
38037 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
38038 * common.opt (freport-bug): Likewise.
38039 * cif-code.def (CIF_FINAL_ERROR): Likewise.
38040 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
38041 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
38042 translatable string.
38043 * config/i386/i386.c (function_value_32): Likewise.
38044 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
38045 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
38047 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
38048 * common/config/msp430/msp430-common.c (msp430_handle_option):
38050 * symtab.c (symtab_node::verify_base): Likewise.
38051 * opts.c (set_debug_level): Likewise.
38052 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
38053 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
38054 missing whitespace to translatable strings.
38055 * config/avr/avr.md (bswapsi2): Fix typo in comment.
38056 * config/sh/superh.h: Likewise.
38057 * config/i386/xopintrin.h: Likewise.
38058 * config/i386/znver1.md: Likewise.
38059 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
38060 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
38061 * double-int.h (struct double_int): Likewise.
38062 * double-int.c (div_and_round_double): Likewise.
38063 * wide-int.cc: Likewise.
38064 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
38065 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
38066 * cfgcleanup.c (crossjumps_occured): Renamed to ...
38067 (crossjumps_occurred): ... this.
38068 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
38071 PR tree-optimization/78899
38072 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
38073 returning bool return struct loop *, NULL for failure and the new
38075 (versionable_outer_loop_p): Don't version outer loop if it has
38076 dont_vectorized bit set.
38077 (tree_if_conversion): When versioning outer loop, ensure
38078 tree_if_conversion is performed also on the inner loop of the
38079 non-vectorizable outer loop copy.
38080 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
38081 LOOP_VECTORIZED in inner loop of the scalar outer loop and
38082 prevent vectorization of it.
38083 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
38084 the outer loop vectorization of the non-scalar version is attempted
38085 before vectorization of the inner loop in scalar version. If
38086 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
38087 vectorization of its inner loop.
38088 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
38089 has 2 inner loops, rename also on edges from bb whose single pred
38090 is outer_loop->header. Fix typo in function comment.
38092 2017-01-09 Martin Sebor <msebor@redhat.com>
38095 * asan.c (asan_emit_stack_protection): Increase local buffer size
38096 to avoid snprintf truncation warning.
38098 2017-01-09 Andrew Pinski <apinski@cavium.com>
38100 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
38101 to reference thunderx2t99 for the tuning structure
38102 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
38104 (thunderx2t99_extra_costs): This.
38105 * config/aarch64/aarch64-tune.md: Regenerate.
38106 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
38107 (vulcan_addrcost_table): This.
38108 (vulcan_regmove_cost): Rename to ...
38109 (thunderx2t99_regmove_cost): This.
38110 (vulcan_vector_cost): Rename to ...
38111 (thunderx2t99_vector_cost): this.
38112 (vulcan_branch_cost): Rename to ...
38113 (thunderx2t99_branch_cost): This.
38114 (vulcan_tunings): Rename to ...
38115 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
38116 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
38118 2017-01-09 Martin Jambor <mjambor@suse.cz>
38122 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
38123 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
38124 (propagate_vr_accross_jump_function): Use the above function for all
38125 value range computations for pass-through jump functions and type
38126 converasion from explicit value range values.
38127 (ipcp_propagate_stage): Do not attempt to deduce types of formal
38128 parameters from TYPE_ARG_TYPES.
38129 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
38130 (ipa_write_node_info): Stream type of the actual argument.
38131 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
38133 2017-01-09 Martin Liska <mliska@suse.cz>
38136 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
38137 (lookup_compiler): Do not show error message with have_E.
38139 2017-01-09 Jakub Jelinek <jakub@redhat.com>
38141 PR tree-optimization/78938
38142 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
38143 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
38144 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
38145 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
38148 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38150 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
38153 2017-01-09 Richard Biener <rguenther@suse.de>
38155 PR tree-optimization/78997
38156 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
38157 name condition properly.
38159 2017-01-09 Richard Biener <rguenther@suse.de>
38162 * dwarf2out.c (is_cxx): New overload with context.
38163 (is_naming_typedef_decl): Use it.
38165 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38167 * invoke.texi (Option Summary): Correct spacing in option lists
38168 and add line breaks to fix over-long lines.
38170 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38172 PR middle-end/17660
38174 * extend.texi (Common Variable Attributes): Add xref to GCC
38175 Internals manual to explain mode attribute keywords.
38177 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38180 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
38181 and Preprocessor Options.
38182 (Options for Linking): Document -pthread here....
38183 (RS/6000 and PowerPC Options): ...not here.
38184 (Solaris 2 Options): ...or here.
38185 * doc/cppopts.texi: Document -pthread.
38187 2017-01-08 Martin Sebor <msebor@redhat.com>
38189 PR middle-end/77708
38190 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
38191 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
38192 New member functions.
38193 (format_directive): Used them.
38195 (pass_sprintf_length::handle_gimple_call): Same.
38196 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
38197 to avoid truncation for any argument.
38198 (extract_affine_mul): Same.
38199 * tree.c (get_file_function_name): Same.
38201 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38203 PR middle-end/77484
38204 * predict.def (PRED_INDIR_CALL): Set to 86.
38206 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38208 PR preprocessor/54124
38209 * doc/cppopts.texi: Reformat -d subtable to list the full name
38210 of the options. Add cross-reference to the docs for the general
38211 compiler -d options.
38212 * doc/invoke.texi (Developer Options): Add cross-reference to the
38213 preprocessor-specific -d option documentation.
38215 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38217 PR preprocessor/13498
38218 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
38219 redudant material, and reflect new command-line options.
38220 (System Headers): Likewise.
38222 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38224 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
38225 -isystem, and -idirafter. Copy-edit.
38226 * doc/cppopts.texi: Copy-edit. Remove contradiction about
38227 default for -ftrack-macro-expansion. Delete obsolete and
38228 badly-formatted implementation details about -fdebug-cpp output.
38229 * doc/cppwarnopts.texi: Copy-edit.
38231 2017-01-07 David Malcolm <dmalcolm@redhat.com>
38234 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
38235 that the transition from a max line width >= 1<<10 to narrower
38236 lines works correctly.
38238 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
38240 * doc/options.texi (PerFunction): New.
38241 * opt-functions.awk (switch_flags): Map both Optimization and
38242 PerFunction to CL_OPTIMIZATION.
38243 * opth-gen.awk: Test for PerFunction flag along with
38245 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
38246 it only when the latter is present. Skip those that don't in
38247 the hash function generator.
38248 * common.opt (fvar-tracking): Mark as PerFunction instead of
38250 (fvar-tracking-assignments): Likewise.
38251 (fvar-tracking-assignments-toggle): Likewise.
38252 (fvar-tracking-uninit): Likewise.
38254 2017-01-07 Jakub Jelinek <jakub@redhat.com>
38256 PR translation/79018
38257 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
38260 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
38263 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
38264 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
38265 TARGET_LONG_JUMP_TABLE_OFFSETS.
38266 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
38267 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
38268 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
38269 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
38270 * config/m68k/m68k.md (tablejump expander): Likewise.
38271 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
38272 TARGET_LONG_JUMP_TABLE_OFFSETS.
38273 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
38274 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
38276 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
38277 David Holsgrove <david.holsgrove@xilinx.com>
38279 * common/config/microblaze/microblaze-common.c
38280 (TARGET_EXCEPT_UNWIND_INFO): Remove.
38281 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
38283 * config/microblaze/microblaze.c (microblaze_must_save_register)
38284 (microblaze_expand_epilogue, microblaze_return_addr): Handle
38286 (microblaze_eh_return): New function.
38287 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
38288 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
38289 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
38290 * config/microblaze/microblaze.md (eh_return): New pattern.
38292 2017-01-06 Jakub Jelinek <jakub@redhat.com>
38294 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
38295 GCC_DIAGNOSTIC_STRINGIFY): Define.
38297 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
38299 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38301 * config/arm/arm.md (<mcrr>): New.
38303 * config/arm/arm.c (arm_arch5te): New.
38304 (arm_option_override): Set arm_arch5te.
38305 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
38307 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
38308 (arm_mcrr_qualifiers): ... this. New.
38309 (MRRC_QUALIFIERS): Define to...
38310 (arm_mrrc_qualifiers): ... this. New.
38311 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
38313 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
38314 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
38315 (MRRCI, mrrc, MRRC): New.
38316 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
38317 VUNSPEC_MRRC2): New.
38319 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38321 * config/arm/arm.md (<mcr>): New.
38323 * config/arm/arm.c (arm_coproc_builtin_available): Add
38324 support for mcr, mrc, mcr2 and mrc2.
38325 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
38326 (arm_mcr_qualifiers): ... this. New.
38327 (MRC_QUALIFIERS): Define to ...
38328 (arm_mrc_qualifiers): ... this. New.
38329 (MCR_QUALIFIERS): Define to ...
38330 (arm_mcr_qualifiers): ... this. New.
38331 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
38333 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
38334 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
38335 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
38336 VUNSPEC_MRC2): New.
38338 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38340 * config/arm/arm.md (*ldc): New.
38344 * config/arm/arm.c (arm_coproc_builtin_available): Add
38345 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
38346 (arm_coproc_ldc_stc_legitimate_address): New.
38347 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
38348 'qualifier_const_pointer'.
38349 (LDC_QUALIFIERS): Define to...
38350 (arm_ldc_qualifiers): ... this. New.
38351 (STC_QUALIFIERS): Define to...
38352 (arm_stc_qualifiers): ... this. New.
38353 * config/arm/arm-protos.h
38354 (arm_coproc_ldc_stc_legitimate_address): New.
38355 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
38356 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
38357 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
38358 stc2, stcl, stc2l): New.
38359 * config/arm/constraints.md (Uz): New.
38360 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
38361 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
38362 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
38363 VUNSPEC_STC2L): New.
38365 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38367 * config/arm/arm.md (<cdp>): New.
38368 * config/arm/arm.c (neon_const_bounds): Rename this ...
38369 (arm_const_bounds): ... this.
38370 (arm_coproc_builtin_available): New.
38371 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
38372 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
38373 (CDP_QUALIFIERS): Define to...
38374 (arm_cdp_qualifiers): ... this. New.
38376 (arm_expand_builtin_args): Add case for 6 arguments.
38377 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
38378 (arm_const_bounds): ... this.
38379 (arm_coproc_builtin_available): New.
38380 * config/arm/arm_acle.h (__arm_cdp): New.
38382 * config/arm/arm_acle_builtins.def (cdp): New.
38384 * config/arm/iterators.md (CDPI,CDP,cdp): New.
38385 * config/arm/neon.md: Rename all 'neon_const_bounds' to
38386 'arm_const_bounds'.
38387 * config/arm/types.md (coproc): New.
38388 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
38389 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
38390 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
38391 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
38393 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38395 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
38396 (UBINOP_QUALIFIERS): New.
38398 (acle_builtin_data): New. Change comment.
38399 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
38400 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
38401 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
38402 arm_acle_builtins.def.
38403 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
38404 (arm_init_acle_builtins): New.
38405 (CRC32_BUILTIN): Remove.
38406 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
38407 crc32cb, crc32ch and crc32cw.
38408 (arm_init_crc32_builtins): Remove.
38409 (arm_init_builtins): Use arm_init_acle_builtins rather
38410 than arm_init_crc32_builtins.
38411 (arm_expand_acle_builtin): New.
38412 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
38413 * config/arm/arm_acle_builtins.def: New.
38415 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38417 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
38418 (arm_builtin_datum): ... this.
38419 (arm_init_neon_builtin): Rename to ...
38420 (arm_init_builtin): ... this. Add a new parameters PREFIX
38421 and USE_SIG_IN_NAME.
38422 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
38423 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
38424 'arm_builtin_datum'.
38425 (arm_init_vfp_builtins): Likewise.
38426 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
38427 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
38428 (arm_expand_neon_args): Rename to ...
38429 (arm_expand_builtin_args): ... this. Rename builtin_arg
38430 enum values and differentiate between ARG_BUILTIN_MEMORY
38431 and ARG_BUILTIN_NEON_MEMORY.
38432 (arm_expand_neon_builtin_1): Rename to ...
38433 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
38434 values, arm_expand_builtin_args and add bool parameter NEON.
38435 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
38436 (arm_expand_vfp_builtin): Likewise.
38437 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
38439 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38441 PR middle-end/77484
38442 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
38443 * predict.c (tree_estimate_probability_bb): Reverse direction of
38444 polymorphic call predictor.
38446 2017-01-06 David Malcolm <dmalcolm@redhat.com>
38448 * passes.c (execute_one_pass): Split out pass-skipping logic into...
38449 (determine_pass_name_match): ...this new function and...
38450 (should_skip_pass_p): ...this new function.
38452 2017-01-06 Nathan Sidwell <nathan@acm.org>
38454 * ipa-visibility.c (function_and_variable_visibility): Reformat
38455 comments and long lines. Remove extrneous if.
38456 * symtab.c (symtab_node::make_decl_local): Fix code format.
38457 (symtab_node::set_section_for_node): Fix comment typo.
38459 2017-01-06 Martin Liska <mliska@suse.cz>
38462 * lra-constraints.c: Rename invariant to lra_invariant.
38463 * predict.c (set_even_probabilities): Initialize e to NULL.
38465 2017-01-05 Martin Sebor <msebor@redhat.com>
38467 PR tree-optimization/78910
38468 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
38469 (format_integer): Correct off-by-one error in the handling
38470 of precision with negative numbers in signed conversions..
38472 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
38474 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
38476 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38478 PR tree-optimization/71016
38479 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
38480 factor_out_conditional_conversion. Formatting fix.
38481 (factor_out_conditional_conversion): Add cond_stmt argument.
38482 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
38483 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
38486 2017-01-05 David Malcolm <dmalcolm@redhat.com>
38488 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
38489 read-rtl-function.o, and selftest-rtl.o.
38490 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
38491 (selftest::aarch64_test_loading_full_dump): New function.
38492 (selftest::aarch64_run_selftests): New function.
38493 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
38494 selftest::aarch64_run_selftests.
38495 * config/i386/i386.c
38496 (selftest::ix86_test_loading_dump_fragment_1): New function.
38497 (selftest::ix86_test_loading_call_insn): New function.
38498 (selftest::ix86_test_loading_full_dump): New function.
38499 (selftest::ix86_test_loading_unspec): New function.
38500 (selftest::ix86_run_selftests): Call the new functions.
38501 * emit-rtl.c (maybe_set_max_label_num): New function.
38502 * emit-rtl.h (maybe_set_max_label_num): New decl.
38503 * function.c (instantiate_decls): Guard call to
38504 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
38505 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
38507 * gensupport.c (gen_reader::gen_reader): Pass "false"
38508 for new "compact" param of rtx_reader.
38509 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
38510 rather than an empty string for NULL strings.
38511 * read-md.c: Potentially include config.h rather than bconfig.h.
38512 Wrap include of errors.h with #ifdef GENERATOR_FILE.
38513 (have_error): New global, copied from errors.c.
38514 (md_reader::read_name): Rename to...
38515 (md_reader::read_name_1): ...this, adding "out_loc" param,
38516 and converting "missing name or number" to returning false, rather
38518 (md_reader::read_name): Reimplement in terms of read_name_1.
38519 (md_reader::read_name_or_nil): New function.
38520 (md_reader::read_string): Handle "(nil)" by returning NULL.
38521 (md_reader::md_reader): Add new param "compact".
38522 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
38523 (md_reader::read_file): New method.
38524 * read-md.h (md_reader::md_reader): Add new param "compact".
38525 (md_reader::read_file): New method.
38526 (md_reader::is_compact): New accessor.
38527 (md_reader::read_name): Convert return type from void to file_location.
38528 (md_reader::read_name_or_nil): New decl.
38529 (md_reader::read_name_1): New decl.
38530 (md_reader::m_compact): New field.
38531 (noop_reader::noop_reader): Pass "false" for new "compact" param
38533 (rtx_reader::rtx_reader): Add new "compact" param.
38534 (rtx_reader::read_rtx_operand): Make virtual and convert return
38535 type from void to rtx.
38536 (rtx_reader::read_until): New decl.
38537 (rtx_reader::handle_any_trailing_information): New virtual function.
38538 (rtx_reader::postprocess): New virtual function.
38539 (rtx_reader::finalize_string): New virtual function.
38540 (rtx_reader::m_in_call_function_usage): New field.
38541 (rtx_reader::m_reuse_rtx_by_id): New field.
38542 * read-rtl-function.c: New file.
38543 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
38544 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
38545 (selftest::verify_three_block_rtl_cfg): New decl.
38546 * read-rtl-function.h: New file.
38547 * read-rtl.c: Potentially include config.h rather than bconfig.h.
38548 For host, include function.h, memmodel.h, and emit-rtl.h.
38549 (one_time_initialization): New function.
38550 (struct compact_insn_name): New struct.
38551 (compact_insn_names): New array.
38552 (find_code): Handle insn codes in compact dumps.
38553 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
38554 (bind_subst_iter_and_attr): Likewise.
38555 (add_condition_to_string): Likewise.
38556 (add_condition_to_rtx): Likewise.
38557 (apply_attribute_uses): Likewise.
38558 (add_current_iterators): Likewise.
38559 (apply_iterators): Likewise.
38560 (initialize_iterators): Guard usage of apply_subst_iterator with
38561 #ifdef GENERATOR_FILE.
38562 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
38563 (md_reader::read_mapping): Likewise.
38564 (add_define_attr_for_define_subst): Likewise.
38565 (add_define_subst_attr): Likewise.
38566 (read_subst_mapping): Likewise.
38567 (check_code_iterator): Likewise.
38568 (rtx_reader::read_rtx): Likewise. Move one-time initialization
38570 (one_time_initialization): New function.
38571 (rtx_reader::read_until): New method.
38572 (read_flags): New function.
38573 (parse_reg_note_name): New function.
38574 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
38575 Handle reuse_rtx ids.
38576 Wrap iterator lookup within #ifdef GENERATOR_FILE.
38577 Add parsing support for RTL dumps, mirroring the special-cases in
38578 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
38579 values, and calling handle_any_trailing_information.
38580 (rtx_reader::read_rtx_operand): Convert return type from void
38581 to rtx, returning return_rtx. Handle case 'e'. Call
38582 finalize_string on XSTR and XTMPL fields.
38583 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
38584 "(nil)" values were omitted. Call the postprocess vfunc on the
38586 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
38587 class ctor. Initialize m_in_call_function_usage. Call
38588 one_time_initialization.
38589 * rtl-tests.c (selftest::test_uncond_jump): Call
38590 set_new_first_and_last_insn.
38591 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
38592 * selftest-rtl.c: New file.
38593 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
38594 (selftest::get_insn_by_uid): New decl.
38595 * selftest-run-tests.c (selftest::run_tests): Call
38596 read_rtl_function_c_tests.
38597 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
38598 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
38601 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
38603 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
38604 operands in a special way. Assert that pos+len <= mode precision.
38606 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38608 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
38609 3 argument Alias with unlimited for the negative form.
38610 (fno-vect-cost-model): Removed.
38612 2017-01-05 Martin Liska <mliska@suse.cz>
38614 * hsa-gen.c (gen_hsa_divmod): New function.
38615 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
38617 2017-01-05 Martin Liska <mliska@suse.cz>
38620 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
38623 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38625 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
38626 small constant length operands.
38628 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38630 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
38631 between loop iterations.
38633 2017-01-05 Martin Liska <mliska@suse.cz>
38636 * gimplify.c (gimplify_decl_expr): Compare to
38637 asan_poisoned_variables instread of checking flags.
38638 (gimplify_target_expr): Likewise.
38639 (gimplify_expr): Likewise.
38640 (gimplify_function_tree): Conditionally initialize
38641 asan_poisoned_variables.
38643 2017-01-04 Jeff Law <law@redhat.com>
38645 PR tree-optimizatin/78812
38646 * rtl.h (contains_mem_rtx_p): Prototype.
38647 * ifcvt.c (containts_mem_rtx_p): Move from here to...
38648 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
38649 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
38650 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
38651 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
38653 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38655 * input.c (assert_char_at_range): Default-initialize actual_range.
38657 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38659 * df-scan.c (df_ref_create_structure): Make regno unsigned,
38660 to match the caller.
38662 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38664 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
38665 insns after final jump in test to emit dummy move.
38667 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38669 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
38670 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
38672 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38674 * multiple_target.c (create_dispatcher_calls): Init e_next.
38675 * tree-ssa-loop-split.c (split_loop): Init border.
38676 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
38679 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
38684 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
38685 (altivec_register_operand): Do not return true if the operand
38686 contains a SUBREG mixing SImode and SFmode.
38687 (vsx_register_operand): Likewise.
38688 (vsx_reg_sfsubreg_ok): New predicate.
38689 (vfloat_operand): Do not return true if the operand contains a
38690 SUBREG mixing SImode and SFmode.
38691 (vint_operand): Likewise.
38692 (vlogical_operand): Likewise.
38693 (gpc_reg_operand): Likewise.
38694 (int_reg_operand): Likewise.
38695 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
38696 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
38697 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
38699 (rs6000_emit_move_si_sf_subreg): New helper function.
38700 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
38701 fixup SUBREGs involving SImode and SFmode.
38702 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
38703 numbers for the new peephole2 optimization.
38704 (peephole2 for SFmode unions): New peephole2 to optimize cases in
38705 the GLIBC math library that do AND/IOR/XOR operations on single
38706 precision floating point.
38707 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
38708 target macros to say whether we need to avoid SUBREGs mixing
38710 (TARGET_ALLOW_SF_SUBREG): Likewise.
38711 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
38712 (UNSPEC_SI_FROM_SF): Likewise.
38713 (iorxor): Change spacing.
38714 (and_ior_xor): New iterator for AND, IOR, and XOR.
38715 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
38716 (movdi_from_sf_zero_ext): Likewise.
38717 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
38718 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
38719 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
38720 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
38721 (fms<mode>4): Likewise.
38722 (fnma<mode>4): Likewise.
38723 (fnms<mode>4): Likewise.
38724 (nfma<mode>4): Likewise.
38725 (nfms<mode>4): Likewise.
38727 2017-01-04 Marek Polacek <polacek@redhat.com>
38730 * doc/invoke.texi: Document -Wpointer-compare.
38732 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38734 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
38737 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
38738 descriptions for -gdwarf-5 and emit them as uleb128 instead of
38741 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
38744 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
38745 documentation of the powerpc_popcntb_ok attribute.
38746 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
38747 code to issue warning messages if a requested CPU configuration is
38748 not supported by the binary (assembler and loader) toolchain.
38749 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
38750 made to define a built-in function that has been disabled.
38751 (paired_init_builtins): Add assertion to prevent ICE if attempt is
38752 made to define a built-in function that has been disabled.
38753 (altivec_init_builtins): Add comment explaining why definition
38754 of the DST built-in functions is not preceded by an assertion
38755 check. Add assertions to prevent ICE if attempts are made to
38756 define an altivec predicate or an abs* built-in function that has
38758 (htm_init_builtins): Add comment explaining why definition of the
38759 htm built-in functions is not preceded by an assertion check.
38761 2017-01-04 Jeff Law <law@redhat.com>
38763 PR tree-optimizatin/67955
38764 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
38765 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
38766 the points-to solution does not include pt_null. Use DECL_PT_UID
38769 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
38771 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
38772 Use gen_int_mode instead of gen_lopwart for const_int operands.
38774 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38776 PR tree-optimization/71563
38777 * match.pd: Simplify X << Y into X if Y is known to be 0 or
38778 out of range value - has low bits known to be zero.
38780 2017-01-04 Alan Modra <amodra@gmail.com>
38782 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
38783 * configure: Regenerate.
38784 * config.in: Regenerate.
38786 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38789 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
38790 a substring of the message, but strcmp with the whole message. Ifdef
38791 ENABLE_NLS, translate the message first using dgettext.
38793 2017-01-03 Jeff Law <law@redhat.com>
38795 PR tree-optimizatin/78856
38796 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
38797 (mark_threaded_blocks): Remove code to truncate thread paths that
38798 cross multiple loop headers. Instead invalidate the cached loop
38799 iteration information and handle case of a thread path walking
38800 into an irreducible region.
38802 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
38805 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
38806 assertions. Add support for doing the signbit if the IEEE 128-bit
38807 floating point value is in a GPR.
38808 * config/rs6000/rs6000.md (Fsignbit): Delete.
38809 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
38810 Update the length attribute if the value is in a GPR.
38811 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
38812 the sign or zero extension instruction, since the value is always 0/1.
38813 (signbit<mode>2_dm2): Delete using <Fsignbit>.
38816 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
38817 extracting SImode to a GPR register so that we can generate a
38818 store, limit the vector to be in a traditional Altivec register
38819 for the vextuwrx instruction.
38821 2017-01-03 Ian Lance Taylor <iant@google.com>
38823 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
38825 2017-01-03 Martin Sebor <msebor@redhat.com>
38827 PR tree-optimization/78696
38828 * gimple-ssa-sprintf.c (format_floating): Correct handling of
38829 precision. Use MPFR for %f for greater fidelity. Correct handling
38831 (pass_sprintf_length::compute_format_length): Set width and precision
38832 specified by asrerisk to void_node for vararg functions.
38833 (try_substitute_return_value): Adjust dump output.
38835 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
38837 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
38839 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
38841 * doc/invoke.texi (SPARC options): Document -mlra as the default.
38842 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
38843 -mlra/-mno-lra was passed to the compiler.
38845 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
38847 PR rtl-optimization/65618
38848 * emit-rtl.c (try_split): Move initialization of "before" and
38849 "after" to just before the call to emit_insn_after_setloc.
38851 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
38853 * doc/md.texi (Standard Names): Remove reference to Java frontend.
38855 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
38857 * dwarf2out.c (gen_enumeration_type_die): When
38858 -gno-strict-dwarf, add a DW_AT_encoding attribute.
38860 2017-01-03 Jakub Jelinek <jakub@redhat.com>
38862 PR tree-optimization/78965
38863 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
38864 Change first argument from const call_info & to call_info &. For %n
38865 set info.nowrite to false.
38867 PR middle-end/78901
38868 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
38869 possibly throwing calls.
38871 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
38872 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
38873 and fns handling, rather than in a separate case SSA_NAME.
38875 2017-01-02 Jeff Law <law@redhat.com>
38877 * config/darwin-driver.c (darwin_driver_init): Const-correctness
38878 fixes for first_period and second_period variables.
38880 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
38883 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
38884 (*insvqi_1): New insn pattern.
38885 (*insvqi_1_mem_rex64): Ditto.
38886 (*insvqi_2): Ditto.
38887 (*insvqi_3): Rename from *insvqi.
38889 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
38891 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
38893 * doc/cfg.texi (Edges): Remove reference to Java.
38894 (Maintaining the CFG): Ditto.
38896 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38898 PR middle-end/77674
38899 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
38900 transparent aliases.
38902 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38904 PR middle-end/77484
38905 * predict.def (PRED_CALL): Update hitrate.
38906 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
38907 * predict.c (tree_estimate_probability_bb): Split CALL predictor
38908 into direct/indirect/polymorphic variants.
38910 2017-01-01 Jakub Jelinek <jakub@redhat.com>
38912 Update copyright years.
38914 * gcc.c (process_command): Update copyright notice dates.
38915 * gcov-dump.c (print_version): Ditto.
38916 * gcov.c (print_version): Ditto.
38917 * gcov-tool.c (print_version): Ditto.
38918 * gengtype.c (create_file): Ditto.
38919 * doc/cpp.texi: Bump @copying's copyright year.
38920 * doc/cppinternals.texi: Ditto.
38921 * doc/gcc.texi: Ditto.
38922 * doc/gccint.texi: Ditto.
38923 * doc/gcov.texi: Ditto.
38924 * doc/install.texi: Ditto.
38925 * doc/invoke.texi: Ditto.
38927 Copyright (C) 2017 Free Software Foundation, Inc.
38929 Copying and distribution of this file, with or without modification,
38930 are permitted in any medium without royalty provided the copyright
38931 notice and this notice are preserved.