1 2017-12-14 Sudakshina Das <sudi.das@arm.com>
2 Bin Cheng <bin.cheng@arm.com>
5 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
7 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
10 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
13 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
15 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
16 Alan Hayward <alan.hayward@arm.com>
17 David Sherwood <david.sherwood@arm.com>
19 * poly-int.h: New file.
20 * poly-int-types.h: Likewise.
21 * coretypes.h: Include them.
22 (POLY_INT_CONVERSION): Define.
23 * target.def (estimated_poly_value): New hook.
24 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
25 * doc/tm.texi: Regenerate.
26 * doc/poly-int.texi: New file.
27 * doc/gccint.texi: Include it.
28 * doc/rtl.texi: Describe restrictions on subreg modes.
29 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
30 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
31 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
32 * targhooks.h (default_estimated_poly_value): Declare.
33 * targhooks.c (default_estimated_poly_value): New function.
34 * target.h (estimated_poly_value): Likewise.
35 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
36 (wi::unary_traits): Delete.
37 (wi::binary_traits::signed_shift_result_type): Define for
38 offset_int << HOST_WIDE_INT, etc.
39 (generic_wide_int::operator <<=): Define for all types and use
40 wi::lshift instead of <<.
41 (wi::hwi_with_prec): Add a default constructor.
42 (wi::ints_for): New class.
43 (operator <<): Define for all wide-int types.
44 (operator /): New function.
45 (operator %): Likewise.
46 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
47 (ASSERT_MAYBE_NE_AT): New macros.
49 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
50 Dominik Vogt <vogt@linux.vnet.ibm.com>
53 * emit-rtl.c (init_emit): Remove ??? comment.
54 * explow.c (get_dynamic_stack_size): Take known alignment of stack
55 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
56 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
57 alignment of 3 virtual registers to BITS_PER_WORD.
59 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
61 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
63 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
64 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
66 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
69 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
71 2017-12-13 Jakub Jelinek <jakub@redhat.com>
73 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
74 with the exception of debug begin stmt markers appear before
78 * final.c (rest_of_handle_final): Call variable_tracking_main only
79 if !flag_var_tracking.
81 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
82 Jakub Jelinek <jakub@redhat.com>
86 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
87 labels debug stmts that can only appear after labels.
89 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
91 PR rtl-optimization/82398
92 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
93 EXPR_USEFULNESS in priority comparison.
95 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
97 PR rtl-optimization/83393
98 * combine.c (move_deaths): If reg_stat points to a too new insn in
99 last_death, do not use it: find the proper insn instead.
101 2017-12-12 Jeff Law <law@redhat.com>
103 PR tree-optimization/83298
104 PR tree-optimization/83362
105 PR tree-optimization/83383
106 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
107 push_value_range a public interface. Add new argument to
108 record_ranges_from_stmt.
109 * gimple-ssa-evrp-analyze.c
110 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
111 Update comments. Handle recording temporary equivalences.
112 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
113 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
114 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
115 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
116 gimple-ssa-evrp-analyze.h.
117 (record_temporary_equivalences_from_phis): Add new argument. When
118 the PHI arg is an SSA_NAME, set the result's range to the range
120 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
122 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
123 Pass it down to children as needed.
124 (thread_outgoing_edges): Likewise.
125 (thread_across_edge): Likewise. Push/pop range state as needed.
126 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
128 2017-12-12 Julia Koval <julia.koval@intel.com>
130 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
131 (PTA_CANNONLAKE): Remove PTA_CLWB.
133 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
136 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
137 (vcondv2div2df): Likewise.
138 (vconduv2dfv2di): Likewise.
140 2017-12-12 Jakub Jelinek <jakub@redhat.com>
142 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
143 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
144 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
145 ATTR_MATHFN_FPROUNDING_ERRNO.
147 2017-12-12 Richard Biener <rguenther@suse.de>
149 PR tree-optimization/83385
150 * graphite-scop-detection.c (get_order, order): Remove.
151 (bb_to_rpo): New global.
153 (build_scops): Sort pbbs in RPO order.
155 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
157 * combine.c (simplify_set): Do not transform subregs to zero_extends
158 if the destination mode is a vector mode.
160 2017-12-12 Jakub Jelinek <jakub@redhat.com>
162 PR tree-optimization/83359
163 * tree-cfg.h (fold_loop_internal_call): Declare.
164 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
165 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
166 (find_loop_dist_alias): New function.
167 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
168 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
171 PR tree-optimization/80631
172 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
173 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
174 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
175 hardcoding zero as the value if COND_EXPR is never true. For
176 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
177 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
178 hardcoding MAX_EXPR as the reduction operation.
179 (is_nonwrapping_integer_induction): Allow negative step.
180 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
181 vect_create_epilog_for_reduction, if no value is suitable, don't
182 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
184 2017-12-12 Richard Biener <rguenther@suse.de>
186 PR tree-optimization/81889
187 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
188 range info from the non-wrapping IV instead of just the range
191 2017-12-12 Julia Koval <julia.koval@intel.com>
193 * config.gcc: Add vaesintrin.h.
194 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
196 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
197 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
199 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
200 * config/i386/immintrin.h: Include vaesintrin.h.
201 * config/i386/sse.md (vaesdec_<mode>): New pattern.
202 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
203 _mm_aesdec_epi128): New intrinsics.
205 2017-12-12 Julia Koval <julia.koval@intel.com>
207 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
208 OPTION_MASK_ISA_VAES_UNSET): New.
209 (ix86_handle_option): Handle -mvaes.
210 * config/i386/cpuid.h: Define bit_VAES.
211 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
212 * config/i386/i386-c.c (__VAES__): New.
213 * config/i386/i386.c (ix86_target_string): Add -mvaes.
214 (ix86_valid_target_attribute_inner_p): Ditto.
215 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
216 * config/i386/i386.opt: Add -mvaes.
217 * doc/invoke.texi: Ditto.
219 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
221 * debug.h (gcc_debug_hooks): Add inline_entry.
222 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
223 * debug.c (do_nothing_debug_hooks): Likewise.
224 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
225 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
226 (dwarf2_lineno_debug_hooks): Likewise.
228 * common.opt (gstatement-frontiers): New, setting
229 debug_nonbind_markers_p.
230 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
231 * toplev.c (process_options): Autodetect value for debug statement
233 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
234 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
236 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
237 markers. Integrate source bind into debug stmt expand loop.
238 (pass_expand::execute): Check debug marker limit. Avoid deep
239 TER and expand debug locations for debug bind insns only.
240 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
242 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
243 * final.c (reemit_insn_block_notes): Take current block from
244 nonbind markers. Declare note where it's first set.
245 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
246 begin stmt markers if enabled.
247 (notice_source_line): Handle nonbind markers. Fail if their
248 location is unknown or that of builtins.
249 (rest_of_handle_final): Convert begin stmt markers to notes if
250 var-tracking didn't run.
251 (rest_of_clean_state): Skip begin stmt markers.
252 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
254 * function.c (allocate_struct_function): Set begin_stmt_markers.
255 * function.h (struct function): Add debug_marker_count counter
256 and debug_nonbind_markers flag.
257 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
258 * gimple-low.c (lower_function_body): Adjust
259 debug_nonbind_markers.
260 (lower_stmt): Drop or skip gimple debug stmts.
261 (lower_try_catch): Skip debug stmts.
262 * gimple.c (gimple_build_debug_begin_stmt): New.
263 (gimple_copy): Increment debug_marker_count if copying one.
264 * gimple.h (gimple_build_debug_begin_stmt): Declare.
265 * gimplify.c (rexpr_location): New.
266 (rexpr_has_location): New.
267 (warn_switch_unreachable_r): Handle gimple debug stmts.
268 (shortcut_cond_r): Call expr_location.
270 (find_goto_label): New.
271 (shortcut_cond_expr): Call expr_has_location, expr_location, and
273 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
275 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
276 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
277 (LANG_HOOKS_INITIALIZER): ... this.
278 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
279 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
281 (update_ebb_live_info): Skip debug insn markers.
282 * lra.c (debug_insn_static_data): Rename to...
283 (debug_bind_static_data): ... this.
284 (debug_marker_static_data): New.
285 (lra_set_insn_recog_data): Select one of the above depending
287 (lra_update_isn_regno_info): Don't assume debug insns have
289 (push_insns): Skip debug insns.
290 * lto-streamer-in.c (input_function): Drop debug stmts
291 depending on active options. Adjust debug_nonbind_markers.
292 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
293 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
294 begin stmt marker notes.
295 (print_insn): Likewise.
296 * recog.c (extract_insn): Recognize rtl for debug markers.
297 * rtl.def (DEBUG_MARKER): New.
298 * tree-inline.c: Include params.h.
299 (remap_gimple_stmt): Handle nonbind markers.
300 (maybe_move_debug_stmts_to_successors): Likewise.
301 (copy_debug_stmt): Likewise.
302 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
303 markers regardless of no side effects.
304 (tsi_link_before): Don't update container's side effects when adding
306 (tsi_link_after): Likewise.
307 (expr_first): Skip begin stmt markers.
308 (expr_last): Likewise.
309 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
310 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
311 Disregard nonbind markers.
312 * tree.c (make_node_stat): Don't set side effects for begin stmt
314 (build1_stat): Likewise.
315 * tree.def (DEBUG_BEGIN_STMT): New.
316 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
317 * var-tracking.c (delete_debug_insns): Renamed to...
318 (delete_vta_debug_insns): ... this.
319 (reemit_marker_as_note): New.
320 (vt_initialize): Reemit markers.
321 (delete_vta_debug_insns): Likewise.
322 (vt_debug_insns_local): Reemit or delete markers.
323 (variable_tracking_main_1): Likewise.
324 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
325 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
326 (gimple_debug_nonbind_marker_p): New.
327 (gimple_build_debug_bind): Adjust.
328 (gimple_build_debug_begin_stmt): New.
329 * doc/invoke.texi (max-debug-marker-count): New param.
330 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
331 (debug_parameter_ref, debug_marker): New.
332 (NOTE_INSN_BEGIN_STMT): New.
333 (DEBUG_INSN): Describe begin stmt markers.
335 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
336 without debug insns, we wouldn't, but clean up debug insns
337 after a control flow insn nevertheless.
339 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
340 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
341 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
342 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
343 (rtl_tidy_fallthru_edge): Likewise.
344 (rtl_verify_fallthru): Likewise.
345 (rtl_verify_bb_layout): Likewise.
346 (skip_insns_after_block): Likewise.
347 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
348 * dwarf2out.c: Include print-rtl.h.
349 (dwarf2out_next_real_insn): New.
350 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
351 Dump debug binds in asm comments.
352 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
353 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
354 callers to use gsi_start_nondebug_bb instead.
355 (gsi_after_labels): Skip gimple debug stmts.
356 (gsi_start_nondebug): New.
357 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
358 (proper_loop_form_for_interchange): Adjust.
359 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
360 * gimple.h (gimple_seq_last_nondebug_stmt): New.
361 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
362 (collect_fallthrough_labels): Likewise.
363 (should_warn_for_implicit_fallthrough): Likewise.
364 (warn_implicit_fallthrough_r): Likewise.
365 (expand_FALLTHROUGH_r): Likewise.
366 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
367 (graphite_copy_stmts_from_block): Skip nonbind markers.
368 * haifa-sched.c (sched_extend_bb): Skip debug insns.
369 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
370 * jump.c (clean_barriers): Skip debug insns.
371 * omp-expand.c (expand_parallel_call): Skip debug insns.
372 (expand_task_call): Likewise.
373 (remove_exit_barrier): Likewise.
374 (expand_omp_taskreg): Likewise.
375 (expand_omp_for_init_counts): Likewise.
376 (expand_omp_for_generic): Likewise.
377 (expand_omp_for_static_nochunk): Likewise.
378 (expand_omp_for_static_chunk): Likewise.
379 (expand_omp_simd): Likewise.
380 (expand_omp_taskloop_for_outer): Likewise.
381 (expand_omp_taskloop_for_inner): Likewise.
382 (expand_oacc_for): Likewise.
383 (expand_omp_sections): Likewise.
384 (expand_omp_single): Likewise.
385 (expand_omp_synch): Likewise.
386 (expand_omp_atomic_load): Likewise.
387 (expand_omp_atomic_store): Likewise.
388 (expand_omp_atomic_fetch_op): Likewise.
389 (expand_omp_atomic_pipeline): Likewise.
390 (expand_omp_atomic_mutex): Likewise.
391 (expand_omp_target): Likewise.
392 (grid_expand_omp_for_loop): Likewise.
393 (grid_expand_target_grid_body): Likewise.
394 (build_omp_regions_1): Likewise.
395 * omp-low.c (check_combined_parallel): Skip debug stmts.
396 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
397 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
398 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
400 * tree-cfg.c (make_blobs_1): Skip debug stmts.
401 (make_edges): Likewise.
402 (cleanup_dead_labels): Likewise.
403 (gimple_can_merge_blocks_p): Likewise.
404 (stmt_starts_bb_p): Likewise.
405 (gimple_block_label): Likewise.
406 (gimple_redirect_edge_and_branch): Likewise.
407 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
409 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
411 * tree-pretty-print (print_declaration): Omit initializer in slim
413 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
415 (eliminate_unnecessary_stmts): Stabilize block removal order.
416 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
417 * var-tracking.c (get_first_insn): New.
418 (vt_emit_notes): Call it.
419 (vt_initialize): Walk any insns before the first BB.
420 (delete_debug_insns): Likewise.
422 * gimple.h (enum gimple_debug_subcode): Add
423 GIMPLE_DEBUG_BEGIN_STMT.
424 (gimple_debug_begin_stmt_p): New.
425 (gimple_debug_nonbind_marker_p): New.
426 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
427 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
428 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
429 * insn-notes.def (BEGIN_STMT): New.
430 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
431 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
432 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
433 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
434 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
435 (INSN_DEBUG_MARKER_KIND): New.
436 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
437 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
438 (INSN_VAR_LOCATION_PTR): New.
439 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
441 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
442 deps only if debug bind insns are enabled.
443 (pass_expand::execute): Avoid deep TER and expand
444 debug locations for debug bind insns only.
445 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
446 debug stmts special handling down to debug bind stmts.
447 * combine.c (try_combine): Narrow debug insns special handling
448 down to debug bind insns.
449 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
450 Narrow debug insns preexisting special handling down to debug
452 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
453 handling down to debug bind insns.
454 * function.c (instantiate_virtual_regs): Skip debug markers,
455 adjust handling of debug binds.
456 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
457 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
458 * haifa-sched.c (schedule_insn): Narrow special handling of debug
459 insns to debug bind insns.
460 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
461 special handling of debug stmts to debug bind stmts.
462 * ipa-split.c (split_function): Likewise.
463 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
464 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
466 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
467 * regrename.c (build_def_use): Likewise.
468 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
469 (pass_cprop_hardreg): Narrow special casing of debug insns to
471 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
472 * reload1.c (reload): Likewise.
473 * sese.c (sese_insert_phis_for_liveouts): Narrow special
474 casing of debug stmts to debug bind stmts.
475 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
476 * ssa-iterators.h (num_imm_uses): Likewise.
477 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
478 debug stmts to debug bind stmts.
479 * tree-inline.c (tree_function_versioning): Narrow special casing
480 of debug stmts to debug bind stmts.
481 * tree-loop-distribution.c (generate_loops_for_partition):
482 Narrow special casing of debug stmts to debug bind stmts.
483 * tree-sra.c (analyze_access_subtree): Narrow special casing
484 of debug stmts to debug bind stmts.
485 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
486 stmts to debug bind stmts.
487 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
488 casing of debug stmts to debug bind stmts.
489 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
490 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
491 casing of debug stmts to debug bind stmts.
492 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
494 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
495 debug stmts to debug bind stmts.
496 (gimple_replace_ssa_lhs): Likewise.
497 (insert_debug_temp_for_var_def): Likewise.
498 (insert_debug_temps_for_defs): Likewise.
499 (reset_debug_uses): Likewise.
500 * tree-ssanames.c (release_ssa_name_fn): Likewise.
501 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
502 (adjust_debug_stmts): Likewise.
503 (adjust_phi_and_debug_stmts): Likewise.
504 (vect_do_peeling): Likewise.
505 * tree-vect-loop.c (vect_transform_loop): Likewise.
506 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
507 * var-tracking.c (adjust_mems): Narrow special casing of debug
508 insns to debug bind insns.
509 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
510 (compute_bb_dataflow, vt_find_locations): Likewise.
511 (vt_expand_loc, emit_notes_for_changes): Likewise.
512 (vt_init_cfa_base): Likewise.
513 (vt_emit_notes): Likewise.
514 (vt_initialize): Likewise.
515 (vt_finalize): Likewise.
517 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
518 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
519 (next_nonnote_nondebug_insn_bb): New.
520 (prev_nonnote_nondebug_insn_bb): New.
521 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
522 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
523 (prev_nonnote_nondebug_insn_bb): Declare.
524 (next_nonnote_nondebug_insn_bb): Declare.
525 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
526 * cfgrtl.c (get_last_bb_insn): Likewise.
527 * lra.c (push_insns): Likewise.
529 2017-12-11 David Malcolm <dmalcolm@redhat.com>
532 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
534 * selftest.c (selftest::test_runner::test_runner): New ctor.
535 (selftest::test_runner::~test_runner): New dtor.
536 * selftest.h (class selftest::test_runner): New class.
538 2017-12-11 Carl Love <cel@us.ibm.com>
540 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
541 vec_extract_fp32_from_shortl]): Add #defines.
542 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
543 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
544 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
545 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
546 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
547 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
548 * doc/extend.texi: Add documentation for the added builtins.
550 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
552 PR rtl-optimization/80693
553 PR rtl-optimization/81019
554 PR rtl-optimization/81020
555 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
556 are not mentioned in i3. Place the REG_UNUSED note on i2,
557 possibly modified to REG_DEAD, if it did not originate in i3.
559 2017-12-11 Jakub Jelinek <jakub@redhat.com>
561 * recog.c (store_data_bypass_p_1): New function.
562 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
563 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
565 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
567 PR rtl-optimization/83361
568 * ifcvt.c (if_convert): Call fixup_partitions.
570 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
572 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
573 early folding of splat_u{8,16,32}.
575 2017-12-11 Jakub Jelinek <jakub@redhat.com>
577 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
578 output_operand_lossage first argument with capital letter.
579 (aarch64_override_options): Don't start error and sorry first argument
582 2017-12-11 Andi Kleen <ak@linux.intel.com>
584 PR gcov-profile/83355
585 * auto-profile.c (string_table::get_index_by_decl): Don't
586 recurse when abstract origin points to itself.
588 2017-12-11 Bin Cheng <bin.cheng@arm.com>
590 PR tree-optimization/83320
591 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
592 (prune_datarefs_not_in_loop): Ditto.
594 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
596 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
598 2017-12-10 Jakub Jelinek <jakub@redhat.com>
600 PR tree-optimization/83337
601 * gimple-loop-interchange.cc (compute_access_stride): Handle
602 bitfield DRs properly.
604 2017-12-09 Jakub Jelinek <jakub@redhat.com>
606 PR tree-optimization/83338
607 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
608 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
609 vector of unsigned integers to vector of signed integers.
611 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
613 PR rtl-optimization/83317
614 * lra-constraints.c (process_address_1): Add insn code check.
616 2017-12-08 Michael Matz <matz@suse.de>
618 Fix PR tree-optimization/83323
619 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
620 head-controlled loops and loop BBs.
621 * common.opt (funroll-and-jam): Remove, instead ...
622 (floop-unroll-and-jam): ... reuse this option.
623 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
624 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
625 (-floop-unroll-and-jam): ... this option.
627 2017-12-08 Jakub Jelinek <jakub@redhat.com>
629 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
630 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
631 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
633 2017-12-08 Richard Biener <rguenther@suse.de>
636 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
637 handle accesses outside of zero-sized vars.
639 2017-12-08 Martin Jambor <mjambor@suse.cz>
641 PR tree-optimization/83141
642 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
643 test for MEM_REFs implicitely changing types with padding. Remove
645 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
647 2017-12-08 Jakub Jelinek <jakub@redhat.com>
649 * config/arc/arc.c (arc_attribute_table): Add exclusions to
651 * config/avr/avr.c (avr_attribute_table): Likewise.
652 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
653 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
654 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
655 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
656 of struct attribute_spec.
657 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
659 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
662 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
664 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
667 * profile-count.c (profile_count::from_gcov_type): Move from
668 profile-count.h; handle overflow.
669 * profile-count.h (profile_count::from_gcov_type): Move offline.
671 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
673 PR rtl-optimization/83304
674 * combine.c (move_deaths): If we do not know where a register died,
677 2017-12-08 Richard Biener <rguenther@suse.de>
679 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
680 Provide -fopt-info-loop feedback when we interchange in a nest.
682 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
684 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
685 for armv6 ARM CPU IDs.
687 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
689 * common/config/arm/arm-common.c: Include <algorithm>.
690 (INCLUDE_VECTOR): Define.
691 (compare_opt_names): New function.
692 (arm_rewrite_selected_arch): Only strip out extensions that can be
693 expressed through -mfpu. Sort the remaining extensions
696 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
698 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
699 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
700 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
701 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
703 2017-06-08 Tristan Gingold <gindold@adacore.com>
706 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
707 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
709 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
711 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
712 of struct attribute_spec.
714 2017-12-08 Julia Koval <julia.koval@intel.com>
716 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
717 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
718 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
719 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
720 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
721 _mm_maskz_dpwssds_epi32): Ditto.
723 2017-12-08 Richard Biener <rguenther@suse.de>
725 PR tree-optimization/81303
726 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
727 conditions try to create a comparison vector type matching
728 the data vector type.
729 (vectorizable_condition): Adjust.
730 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
731 Leave invariant conditions alone in case we can vectorize those.
733 2017-12-08 Julia Koval <julia.koval@intel.com>
735 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
736 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
737 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
738 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
739 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
741 2017-12-08 Julia Koval <julia.koval@intel.com>
743 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
744 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
745 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
746 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
747 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
748 _mm_maskz_dpbusds_epi32): New intrinsics.
750 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
752 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
754 (ld<bh>io_signed): Likewise.
755 (st<bhw>io): Likewise.
756 * config/nios2/predicates.md (ldstio_memory_operand): Allow
757 SMALL_INT12 constant integer operand.
759 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
761 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
763 (nios2_large_constant_allowed): ...this. Adjust uses.
764 (nios2_plus_symbolic_constant_p): Rename to...
765 (nios2_plus_large_constant_p): ...this. Adjust uses.
766 (nios2_legitimate_address_p): Correct CONST_INT handling.
767 (nios2_symbolic_memory_operand_p): Rename to...
768 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
769 (nios2_large_constant_p): Check for large constant integers too.
770 (nios2_split_large_constant): Handle constant integers.
771 (nios2_split_symbolic_memory_operand): Rename to...
772 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
773 (nios2_legitimize_constant_address): Handle constant integers.
774 (r0rel_constant_p): Handle small constant integers.
775 (nios2_print_operand_address): Handle r0-relative integer addresses.
776 * config/nios2/nios2-protos.h: Adjust for renamed functions.
777 * config/nios2/nios2.md: Adjust for renamed functions.
779 2017-12-07 Andrew Waterman <andrew@sifive.com>
781 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
782 (TARGET_HAVE_SRODATA_SECTION): New define.
783 (riscv_select_section): New function.
785 2017-12-08 Joseph Myers <joseph@codesourcery.com>
786 Alexander Monakov <amonakov@ispras.ru>
787 Jakub Jelinek <jakub@redhat.com>
790 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
792 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
794 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
795 a bool success value. Don't call output_operand_lossage here.
796 (aarch64_print_ldpstp_address): Return a bool success value.
797 (aarch64_print_operand_address): Call output_addr_const if
798 aarch64_print_address_internal fails.
799 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
800 'y'; call output_operand_lossage instead. Call output_operand_lossage
801 if aarch64_print_ldpstp_address fails.
803 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
805 * tree-vector-builder.h
806 (tree_vector_builder::binary_encoded_nelts): Declare.
807 * tree-vector-builder.c
808 (tree_vector_builder::binary_encoded_nelts): New function.
809 * fold-const.c (negate_expr_p): Likewise.
810 (operand_equal_p, fold_checksum_tree): Likewise.
811 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
812 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
813 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
814 (uniform_vector_p): Likewise.
815 * varasm.c (const_hash_1, compare_constant): Likewise.
816 * tree-ssa-ccp.c: Include tree-vector-builder.h.
817 (valid_lattice_transition): Operate directly on the VECTOR_CST
819 * ipa-icf.c: Include tree-vector-builder.h.
820 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
821 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
823 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
825 * tree.c (build_vector): Delete.
826 * tree.h (build_vector): Make static and move into the self-testing
829 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
831 * vector-builder.h (vector_builder::derived): New const overload.
832 (vector_builder::elt): New function.
833 * tree-vector-builder.h (tree_vector_builder::type): New function.
834 (tree_vector_builder::apply_step): Declare.
835 * tree-vector-builder.c (tree_vector_builder::apply_step): New
837 * gimple-fold.h (tree_vector_builder): Declare.
838 (gimple_build_vector): Take a tree_vector_builder instead of a
839 type and vector of elements.
840 * gimple-fold.c (gimple_build_vector): Likewise.
841 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
843 (get_initial_defs_for_reduction): Likewise.
844 (vectorizable_induction): Likewise.
846 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
848 * tree-vector-builder.h
849 (tree_vector_builder::new_binary_operation): Declare.
850 * tree-vector-builder.c
851 (tree_vector_builder::new_binary_operation): New function.
852 * fold-const.c (fold_relational_const): Use it.
853 (const_binop): Likewise. Check that both input vectors have
854 the same number of elements, thus excluding things like WIDEN_SUM.
855 Check whether it is possible to operate directly on the encodings
858 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
860 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
861 new_unary_operation, operating only on the encoded elements.
862 (const_unop): Likewise.
863 (exact_inverse): Likewise.
864 (distributes_over_addition_p): New function.
865 (const_binop): Use tree_vector_builder and new_unary_operation
866 for combinations of VECTOR_CST and INTEGER_CST. Operate only
867 on the encoded elements unless the encoding is strided and the
868 operation does not distribute over addition.
869 (fold_convert_const): Use tree_vector_builder and
870 new_unary_operation. Operate only on the encoded elements
871 for truncating integer conversions, or for non-stepped encodings.
873 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
875 * config/sparc/sparc.c: Include tree-vector-builder.h.
876 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
877 * expmed.c: Include tree-vector-builder.h.
878 (make_tree): Use tree_vector_builder instead of build_vector.
879 * fold-const.c: Include tree-vector-builder.h.
880 (const_binop): Use tree_vector_builder instead of build_vector.
881 (const_unop): Likewise.
882 (native_interpret_vector): Likewise.
883 (fold_vec_perm): Likewise.
884 (fold_ternary_loc): Likewise.
885 * gimple-fold.c: Include tree-vector-builder.h.
886 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
888 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
889 (simplify_vector_constructor): Use tree_vector_builder instead
891 * tree-vect-generic.c: Include tree-vector-builder.h.
892 (add_rshift): Use tree_vector_builder instead of build_vector.
893 (expand_vector_divmod): Likewise.
894 (optimize_vector_constructor): Likewise.
895 * tree-vect-loop.c: Include tree-vector-builder.h.
896 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
897 of build_vector. Explicitly use a stepped encoding for
899 * tree-vect-slp.c: Include tree-vector-builder.h.
900 (vect_get_constant_vectors): Use tree_vector_builder instead
902 (vect_transform_slp_perm_load): Likewise.
903 (vect_schedule_slp_instance): Likewise.
904 * tree-vect-stmts.c: Include tree-vector-builder.h.
905 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
906 (vect_gen_perm_mask_any): Likewise.
907 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
908 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
910 (build_vector_from_val): Likewise. Explicitly use a duplicate
913 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
915 * doc/generic.texi (VECTOR_CST): Describe new representation of
917 * vector-builder.h: New file.
918 * tree-vector-builder.h: Likewise.
919 * tree-vector-builder.c: Likewise.
920 * Makefile.in (OBJS): Add tree-vector-builder.o.
921 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
922 * tree-core.h (tree_base): Add a vector_cst field to the u union.
923 (tree_vector): Change the number of elements to
924 vector_cst_encoded_nelts.
925 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
926 (VECTOR_CST_ELTS): Delete.
927 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
928 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
929 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
930 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
931 (VECTOR_CST_ENCODED_ELT): Likewise.
932 (vector_cst_encoded_nelts): New function.
933 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
934 VECTOR_CST_NELTS_PER_PATTERN as arguments.
935 (vector_cst_int_elt, vector_cst_elt): Declare.
936 * tree.c: Include tree-vector-builder.h.
937 (tree_code_size): Abort if passed VECTOR_CST.
938 (tree_size): Update for new VECTOR_CST layout.
939 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
940 VECTOR_CST_NELTS_PER_PATTERN as arguments.
941 (build_vector): Use tree_vector_builder.
942 (vector_cst_int_elt, vector_cst_elt): New functions.
943 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
944 encoded elements and then create the vector in the canonical form.
945 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
946 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
947 (tree_c_tests): Call test_vector_cst_patterns.
948 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
950 (hash_tree): Likewise.
951 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
952 (streamer_write_tree_header): Likewise.
953 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
954 (streamer_alloc_tree): Likewise. Update call to make_vector.
955 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
957 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
959 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
960 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
961 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
963 2017-12-07 Bin Cheng <bin.cheng@arm.com>
964 Richard Biener <rguenther@suse.de>
966 PR tree-optimization/81303
967 * Makefile.in (gimple-loop-interchange.o): New object file.
968 * common.opt (floop-interchange): Reuse the option from graphite.
969 * doc/invoke.texi (-floop-interchange): Ditto. New document for
970 -floop-interchange and mention it for -O3.
971 * opts.c (default_options_table): Enable -floop-interchange at -O3.
972 * gimple-loop-interchange.cc: New file.
973 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
974 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
975 * passes.def (pass_linterchange): New pass.
976 * timevar.def (TV_LINTERCHANGE): New time var.
977 * tree-pass.h (make_pass_linterchange): New declaration.
978 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
979 interchange. Record IV before/after increment in new parameters.
980 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
981 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
983 (check_reduction_path): ...New function here.
984 * tree-vectorizer.h (check_reduction_path): New declaration.
986 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
989 PR rtl-optimization/80818
990 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
991 always early clobbered.
992 * lra-lives.c (process_bb_lives): Check input hard regs for early
993 clobbered non-operand hard reg.
995 2017-12-07 Jakub Jelinek <jakub@redhat.com>
998 * tree-cfg.c (verify_gimple_assign_binary): Don't require
999 types_compatible_p, just that TYPE_MODE is the same.
1001 2017-12-07 Martin Sebor <msebor@redhat.com>
1004 * attribs.c (empty_attribute_table): Initialize new member of
1005 struct attribute_spec.
1006 (decl_attributes): Add argument. Handle mutually exclusive
1007 combinations of attributes.
1008 (selftests::test_attribute_exclusions): New function.
1009 (selftests::attribute_c_tests): Ditto.
1010 * attribs.h (decl_attributes): Add default argument.
1011 * selftest.h (attribute_c_tests): Declare.
1012 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
1013 * tree-core.h (attribute_spec::exclusions, exclude): New type and
1015 * doc/extend.texi (Common Function Attributes): Update const and pure.
1016 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
1017 of struct attribute_spec.
1018 * config/arc/arc.c (arc_attribute_table): Same.
1019 * config/arm/arm.c (arm_attribute_table): Same.
1020 * config/avr/avr.c ( avr_attribute_table): Same.
1021 * config/bfin/bfin.c (bfin_attribute_table): Same.
1022 * config/cr16/cr16.c (cr16_attribute_table): Same.
1023 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
1024 * config/h8300/h8300.c (h8300_attribute_table): Same.
1025 * config/i386/i386.c (ix86_attribute_table): Same.
1026 * config/ia64/ia64.c (ia64_attribute_table): Same.
1027 * config/m32c/m32c.c (m32c_attribute_table): Same.
1028 * config/m32r/m32r.c (m32r_attribute_table): Same.
1029 * config/m68k/m68k.c (m68k_attribute_table): Same.
1030 * config/mcore/mcore.c (mcore_attribute_table): Same.
1031 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
1032 * config/mips/mips.c (mips_attribute_table): Same.
1033 * config/msp430/msp430.c (msp430_attribute_table): Same.
1034 * config/nds32/nds32.c (nds32_attribute_table): Same.
1035 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
1036 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
1037 * config/rl78/rl78.c (rl78__attribute_table): Same.
1038 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
1039 * onfig/rx/rx.c (rx_attribute_table): Same.
1040 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
1041 * config/sh/sh.c (sh_attribute_table): Same.
1042 * config/sparc/sparc.c (sparc_attribute_table): Same.
1043 * config/spu/spu.c (spu_attribute_table): Same.
1044 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
1045 * config/v850/v850.c (v850_attribute_table): Same.
1046 * config/visium/visium.c (visium_attribute_table): Same.
1048 2017-12-07 Tamar Christina <tamar.christina@arm.com>
1051 * config/arm/arm.c (INCLUDE_STRING): Define.
1052 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
1053 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
1056 2017-12-07 Michael Matz <matz@suse.de>
1058 Add unroll and jam pass
1060 * gimple-loop-jam.c: New file.
1061 * Makefile.in (OBJS): Add gimple-loop-jam.o.
1062 * common.opt (funroll-and-jam): New option.
1063 * opts.c (default_options_table): Add unroll-and-jam at -O3.
1064 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
1065 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
1066 * passes.def: Add pass_loop_jam.
1067 * timevar.def (TV_LOOP_JAM): Add.
1068 * tree-pass.h (make_pass_loop_jam): Declare.
1069 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
1070 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
1071 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
1072 to flow_loop_tree_node_add.
1073 (duplicate_subloops, copy_loops_to): Append to sibling list.
1074 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
1075 * doc/invoke.texi (-funroll-and-jam): Document new option.
1076 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
1078 2017-12-07 Richard Biener <rguenther@suse.de>
1080 PR tree-optimization/83296
1081 PR tree-optimization/67769
1082 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
1083 flow sensitive info in an unrelated BB.
1084 (value_replacement): Use reset_flow_sensitive_info.
1085 (minmax_replacement): Reset flow sensitive info on the def
1086 we move. Do not reset flow sensitive info in the whole BB
1087 we move the stmt to.
1088 (abs_replacement): Likewise.
1090 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1093 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1094 rs6000_cpu to the given -mcpu=, or to the default processor.
1096 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1098 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
1099 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
1100 instead of rs6000_cpu_attr.
1101 (rs6000_adjust_cost): Ditto.
1102 (is_microcoded_insn): Ditto.
1103 (rs6000_adjust_priority): Ditto.
1104 (rs6000_issue_rate): Ditto.
1105 (rs6000_use_sched_lookahead): Ditto.
1106 (rs6000_use_sched_lookahead_guard): Ditto.
1107 (rs6000_sched_reorder): Ditto.
1108 (force_new_group): Ditto.
1109 * config/rs6000/rs6000.md (cpu attribute): Ditto.
1110 (group_ending_nop): Ditto.
1112 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1114 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
1115 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
1116 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
1117 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
1119 (rs6000_reassociation_width): Ditto.
1120 (rs6000_emit_epilogue): Ditto.
1121 (rs6000_adjust_cost): Ditto.
1122 (is_microcoded_insn): Ditto.
1123 (is_cracked_insn): Ditto.
1124 (rs6000_adjust_priority): Ditto.
1125 (rs6000_sched_reorder): Ditto.
1126 (rs6000_sched_reorder2): Ditto.
1127 (insn_must_be_first_in_group): Ditto.
1128 (insn_must_be_last_in_group): Ditto.
1129 (rs6000_register_move_cost): Ditto.
1130 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
1133 2017-12-07 Julia Koval <julia.koval@intel.com>
1135 * config.gcc: Add vaesintrin.h.
1136 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
1138 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
1139 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
1141 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1142 * config/i386/immintrin.h: Include vaesintrin.h.
1143 * config/i386/sse.md (vaesdec_<mode>): New pattern.
1144 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
1145 _mm_aesdec_epi128): New intrinsics.
1147 2017-12-06 David Malcolm <dmalcolm@redhat.com>
1149 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
1150 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
1151 c-family/c-spellcheck.cc.
1152 (best_macro_match::best_macro_match): Likewise.
1154 (struct edit_distance_traits<cpp_hashnode *>): Move to
1155 c-family/c-spellcheck.h.
1156 (class best_macro_match): Likewise.
1158 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1160 PR tree-optimization/83293
1161 * gimple-ssa-strength-reduction.c (insert_initializers): Use
1162 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
1163 might insert into empty bb.
1166 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
1167 simplify for plus with :c added, and pointer_plus without that.
1168 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
1169 with undefined overflow and the conversion is not widening,
1170 perform negation in utype and only convert to type afterwards.
1171 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
1172 simplify for plus with :c added, and pointer_plus without that.
1173 If type is integral with undefined overflow and the conversion is
1174 not widening, perform minus in utype and only convert to type
1175 afterwards. Move the last pointer_diff_expr simplify into the
1178 2017-12-06 Martin Sebor <msebor@redhat.com>
1180 PR tree-optimization/82646
1181 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
1182 strncpy, not maxlen.
1184 2017-12-06 Martin Sebor <msebor@redhat.com>
1186 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
1189 PR tree-optimization/83075
1190 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
1191 strncat/strncpy don't change length of source string.
1193 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
1196 2017-11-29 Martin Aberg <maberg@gaisler.com>
1198 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1199 to prevent b2bst errata sequence.
1200 (sqrtdf2_fix): Likewise.
1202 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1204 PR tree-optimization/81945
1205 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
1206 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
1207 to dest_cfun has orig_loop_num set, either remap it to the new
1208 loop number if the loop got moved too, or clear it.
1210 2017-12-05 Steve Ellcey <sellcey@cavium.com>
1212 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
1214 (thunderx2t99_nothing): New insn reservation.
1215 (thunderx2t99_mrs): New insn reservation.
1216 (thunderx2t99_multiple): New insn reservation.
1217 (thunderx2t99_alu_basi): Add bfx to reservation.
1218 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
1220 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1223 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
1225 2017-12-05 Bin Cheng <bin.cheng@arm.com>
1227 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
1228 tree-ssa-pre.c::remove_dead_inserted_code.
1229 * tree-ssa-dce.h: New file.
1230 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
1231 (remove_dead_inserted_code): Move and rename to function
1232 tree-ssa-dce.c::simple_dce_from_worklist.
1233 (pass_pre::execute): Update use.
1235 2017-12-05 Richard Biener <rguenther@suse.de>
1237 PR tree-optimization/83277
1238 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
1239 to code-gen liveout vars.
1241 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
1243 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
1244 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
1245 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
1248 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
1250 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
1252 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
1253 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
1256 2017-12-05 Richard Biener <rguenther@suse.de>
1258 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
1259 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
1260 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
1261 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
1263 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1265 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
1267 2017-12-05 Martin Liska <mliska@suse.cz>
1268 Jakub Jelinek <jakub@redhat.com>
1270 * doc/invoke.texi: Document the options.
1271 * flag-types.h (enum sanitize_code): Add
1272 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1273 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
1274 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1275 * opts.c: Define new sanitizer options.
1276 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
1277 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
1279 2017-12-05 Julia Koval <julia.koval@intel.com>
1281 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
1282 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
1283 (ix86_handle_option): Handle -mavx512vnni.
1284 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
1285 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
1286 * config/i386/i386-c (__AVX512VNNI__): New.
1287 * config/i386/i386.c (ix86_target_string): Handle new option.
1288 (ix86_valid_target_attribute_inner_p): Handle new option.
1289 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
1290 * config/i386/i386.opt (mavx512vnni): New option.
1292 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1295 * athlon.md: Disable for generic.
1296 * haswell.md: Enable for generic.
1297 * i386.c (ix86_sched_init_global): Add core hooks for generic.
1298 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
1300 (ix86_adjust_cost): Move generic to haswell path.
1302 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
1304 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
1305 instead of MEM_P in a couple more places. Fix formatting issues.
1307 2017-12-04 Jim Wilson <jimw@sifive.com>
1309 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
1310 instead of GP_REG_LAST-1.
1311 (riscv_adjust_libcall_cfi_prologue): Likewise.
1312 (riscv_adjust_libcall_cri_epilogue): Likewise.
1313 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
1316 2017-12-04 Luis Machado <luis.machado@linaro.org>
1318 * ipa-pure-const.c (check_decl): Add missing newline.
1319 (state_from_flags): Likewise.
1321 2017-12-04 Jeff Law <law@redhat.com>
1323 PR tree-optimizatin/78496
1324 * gimple-ssa-evrp-analyze.h
1325 (evrp_range_analyzer::get_vr_values): Simplify.
1326 * gimple-ssa-evrp-analyze.c: Corresponding changes.
1327 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
1328 and gimple-ssa-evrp-analyze.h.
1329 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
1330 (simplify_stmt_for_jump_threading): Copy a blob of code from
1331 tree-vrp.c to use ranges to simplify statements.
1332 (dom_opt_dom_walker::before_dom_children): Call
1333 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
1334 (dom_opt_dom_walker::after_dom_children): Similarly for
1335 evrp_range_analyzer::leave.
1336 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
1339 * gimple-ssa-evrp-analyze.c
1340 (evrp_range_analyzer::extract_range_from_stmt): Always use
1341 vr_values::update_value_range so preexisting range info is
1342 medged with new range info, even if the new range is VR_VARYING.
1344 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1346 * combine.c: Adjust comment.
1347 (use_crosses_set_p): Delete.
1348 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
1349 (try_combine): Ditto.
1351 2017-12-04 Richard Biener <rguenther@suse.de>
1353 PR tree-optimization/83255
1354 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
1355 Re-add zero-iteration check.
1357 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1359 PR rtl-optimization/83245
1360 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
1361 hard registers as earlyclobber, also if not in an asm.
1363 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1367 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1370 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1371 rs6000_cpu based on cpu_index, not tune_index.
1373 2017-12-04 Richard Biener <rguenther@suse.de>
1375 PR tree-optimization/83238
1376 * graphite-scop-detection.c (scop_detection::merge_sese): Make
1377 code match comment, rejecting invalid SESE regions.
1379 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
1381 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
1382 require base operand is a REG_POINTER prior to reload on targets
1383 with non-equivalent space registers.
1385 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1387 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
1388 (update_specialized_profile): Fix updating of counts.
1389 (perhaps_add_new_callers): Likewise.
1391 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1394 * x86-tune.def: Remove obsolette FIXMEs.
1395 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
1396 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
1397 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
1399 (X86_TUNE_PAD_RETURNS): Disable for generic.
1401 2017-12-02 Jakub Jelinek <jakub@redhat.com>
1403 PR tree-optimization/83170
1404 PR tree-optimization/83241
1405 * gimple-ssa-store-merging.c
1406 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
1407 gimple_vuse (ins_stmt) in case it has changed.
1408 (imm_store_chain_info::output_merged_store): Likewise.
1410 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
1414 * tree-cfg.c (pass_warn_function_return::execute): Handle
1415 __builtin_ubsan_handle_missing_return like __builtin_unreachable
1416 with BUILTINS_LOCATION.
1420 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
1421 is BLKmode for vector field with vector raw mode, use TYPE_MODE
1422 instead of DECL_MODE.
1424 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
1425 last argument to rtx pointer.
1426 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
1427 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
1428 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
1429 depending on the chosen ISAs.
1430 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
1431 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
1432 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
1434 * config/i386/sse.md (mov<mode>_internal): Likewise.
1435 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
1437 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1439 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
1441 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1443 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
1444 parameters from prototype.
1445 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
1446 parameters. Don't print a comment.
1447 (emit_fusion_gpr_load): Adjust.
1448 (emit_fusion_load_store): Adjust.
1449 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
1450 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
1451 comment on the second line.
1453 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1456 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1457 rs6000_cpu based on cpu_index, not tune_index.
1459 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1461 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
1462 which_alternative instead of which_alternative + 1.
1463 (output_asm_insn): Print an extra tab if the template is short.
1465 2017-12-01 Jim Wilson <jimw@sifive.com>
1467 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
1469 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
1470 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
1472 * doc/tm.texi: Regenerate.
1474 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
1477 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
1478 whether we can allocate pseudos before trying to fix an address.
1479 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
1480 memory address is indexed or indirect.
1481 (floatuns_<mode>si2_hw2): Likewise.
1483 2017-12-01 Jason Merrill <jason@redhat.com>
1485 * Makefile.in (TAGS): Add c-family/*.cc.
1487 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
1489 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
1490 (call_value_insn): Likewise.
1491 (sibcall_insn): Likewise.
1492 (sibcall_value_insn): Likewise.
1493 (movsi_aarch64): Likewise.
1494 (movdi_aarch64): Likewise.
1495 (add_losym_): Likewise.
1496 (ldr_got_small_): Likewise.
1497 (ldr_got_small_sidi): Likewise.
1498 (ldr_got_small_28k_): Likewise.
1499 (ldr_got_small_28k_sidi): Likewise.
1500 * config/aarch64/aarch64.c (aarch64_print_address_internal):
1501 Move output_addr_const to symbolic case. Add error check.
1503 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1505 * config/s390/predicates.md (plus16_Q_operand): New predicate.
1506 * config/s390/s390.md: Disable MVC merging peephole if it would
1507 disable operand forwarding.
1508 (new peephole2): Split MVCs if it would turn them into up to 2
1511 2017-12-01 Richard Biener <rguenther@suse.de>
1513 PR tree-optimization/83232
1514 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
1515 detection of same access. Instead of breaking the group here
1516 do not consider the duplicate. Add comment explaining real fix.
1518 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
1520 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
1522 2017-12-01 Sudakshina Das <sudi.das@arm.com>
1524 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
1526 2017-12-01 Jakub Jelinek <jakub@redhat.com>
1528 * function.h (struct function): Remove cilk_frame_decl,
1529 is_cilk_function and calls_cilk_spawn fields.
1530 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
1532 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
1533 cilk_elemental field.
1534 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
1535 * target.def: Adjust comment.
1536 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
1537 Don't test cilk_elemental.
1539 PR tree-optimization/83233
1540 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
1541 bswap_stat name for the struct.
1544 * tree.h (SWITCH_BREAK_LABEL_P): Define.
1545 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
1546 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
1547 SWITCH_BREAK_LABEL_P set on the label.
1548 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
1549 added for default case if it was missing and not all cases covered.
1550 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
1551 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
1553 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
1554 to avoid -Wimplicit-fallthrough warning.
1555 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
1556 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
1558 PR tree-optimization/83221
1559 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
1561 (init_reassoc): Formatting fix.
1564 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
1565 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
1569 * tree-cfg.c: Include asan.h.
1570 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
1572 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1574 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
1576 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1578 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
1579 vpcmpeqd instruction.
1581 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1583 * config/i386/i386.c (standard_sse_constant_opcode): Fix
1584 registers type for 128bit mode.
1586 2017-11-30 Jakub Jelinek <jakub@redhat.com>
1588 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
1590 * gengtype-state.c (read_state_pair): Likewise.
1591 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
1592 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
1593 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
1594 * ifcvt.c (noce_try_store_flag_constants): Likewise.
1595 * tree-ssa-ccp.c (ccp_finalize): Likewise.
1596 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
1597 * builtins.c (fold_builtin_3): Likewise.
1598 * graphite-scop-detection.c
1599 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
1600 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
1603 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
1606 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
1607 parameter. Return true if ignore_decl_rtl_set_p is true and other
1608 conditions are satisfied.
1609 * asan.h (asan_protect_global): Add new parameter.
1610 * varasm.c (categorize_decl_for_section): Pass true as second parameter
1611 to asan_protect_global calls.
1613 2017-11-30 Jim Wilson <jimw@sifive.com>
1615 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
1616 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
1617 -mstrict-align, add info on default value. Delete redundant lines for
1618 -mabi. Add missing -mexplicit-relocs docs.
1620 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1622 * config/arc/arc.md (trap): New pattern.
1624 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1626 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
1627 instruction to end into a delay slot.
1628 * config/arc/arc.md (cond_delay_insn): Check if the instruction
1629 can be placed into a delay slot against reg_note.
1631 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1633 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
1634 labels number of usages.
1636 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1638 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
1640 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
1642 2017-11-30 Jakub Jelinek <jakub@redhat.com>
1645 * internal-fn.c (expand_mul_overflow): Optimize unsigned
1646 multiplication by power of 2 constant into two shifts + comparison.
1648 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
1651 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
1653 2017-11-30 Richard Biener <rguenther@suse.de>
1655 PR tree-optimization/83202
1656 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
1657 allow_peel argument and guard peeling.
1658 (canonicalize_loop_induction_variables): Likewise.
1659 (canonicalize_induction_variables): Pass false.
1660 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
1661 peeling from cunrolli.
1663 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1665 * combine.c (try_combine): Print a message to dump file whenever
1666 I0, I1, or I2 cannot be combined into I3.
1668 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1670 PR rtl-optimization/83156
1671 PR rtl-optimization/82621
1672 * combine.c (try_combine): Don't split an I2 if one of the dests is
1673 set again before I3. Allow unused dests.
1675 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1677 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
1679 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
1681 PR rtl-optimization/80818
1682 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
1683 recursively. Use insn code for clobber.
1684 (lra_set_insn_recog_data): Pass the new arg to
1685 collect_non_operand_hard_regs.
1686 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
1688 (lra_update_insn_regno_info): Pass insn to
1689 add_regs_to_insn_regno_info.
1691 2017-11-29 Jim Wilson <jimw@sifive.com>
1692 Andrew Waterman <andrew@sifive.com>
1694 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
1695 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
1696 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
1697 use SINGLE_SHIFT_COST.
1698 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
1699 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
1701 2017-11-29 Julia Koval <julia.koval@intel.com>
1703 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
1704 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
1705 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
1706 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
1707 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
1708 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
1709 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
1710 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
1711 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
1712 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
1713 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
1714 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
1715 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
1716 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
1717 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
1718 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
1719 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
1720 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
1721 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
1722 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
1723 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
1724 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
1725 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
1726 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
1727 __builtin_ia32_vpshldv_v2di_mask,
1728 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
1729 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
1730 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
1732 2017-11-29 Julia Koval <julia.koval@intel.com>
1734 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
1735 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
1736 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
1737 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
1738 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
1739 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
1740 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
1741 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
1742 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
1743 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
1744 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
1745 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
1746 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
1747 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
1748 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
1749 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
1750 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
1751 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
1752 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1753 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
1754 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
1756 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1758 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
1759 movsi_pic_gotdata_op instruction as a load for the UT699 errata
1762 2017-11-29 Martin Aberg <maberg@gaisler.com>
1764 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1765 to prevent b2bst errata sequence.
1766 (sqrtdf2_fix): Likewise.
1768 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1770 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
1771 (div_sqrt_insn_p): New function.
1772 (sparc_do_work_around_errata): Insert NOP instructions to
1773 prevent sequences that could trigger the TN-0013 errata for
1774 certain LEON3 processors.
1775 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
1776 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
1777 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
1778 (in_branch_delay): Prevent div and sqrt in delay slot if
1780 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
1782 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1784 * config/sparc/sparc.c (atomic_insn_p): New function.
1785 (sparc_do_work_around_errata): Insert NOP instructions to
1786 prevent sequences that could trigger the TN-0010 errata for
1788 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
1789 instruction referable in atomic_insns_p.
1791 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1793 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
1794 (atomic_compare_and_swap_leon3_1): Likewise.
1797 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1799 * config/sparc/sparc.c (fpop_insn_p): New function.
1800 (sparc_do_work_around_errata): Insert NOP instructions to
1801 prevent sequences that could trigger the TN-0012 errata for
1803 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
1804 * config/sparc/sparc.md (fix_gr712rc): New attribute.
1805 (in_branch_annul_delay): Prevent floating-point instructions
1806 in delay slot of annulled integer branch.
1808 2017-11-29 Richard Biener <rguenther@suse.de>
1810 PR tree-optimization/83202
1811 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
1813 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
1814 nodes vectorized to the same stmts multiple times.
1815 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
1816 (vect_analyze_slp_instance): Adjust.
1817 (scalar_stmts_to_slp_tree_map_t): New typedef.
1818 (vect_schedule_slp_instance): Add a map recording the SLP node
1819 representing the vectorized stmts for a set of scalar stmts.
1820 Avoid code-generating redundancies.
1821 (vect_schedule_slp): Allocate map and pass it down.
1823 2017-11-29 Nathan Sidwell <nathan@acm.org>
1826 * tree.c (build_complex_type): Fix canonicalization. Only fill in
1829 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
1831 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
1832 cases for printing LDP/STP memory addresses.
1833 (aarch64_print_address_internal): Renamed from
1834 aarch64_print_operand_address, added parameter, add Pmode check.
1835 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
1836 (aarch64_print_operand_address): Indirect to
1837 aarch64_print_address_internal.
1838 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
1839 'y' operand output specifier.
1841 2017-11-29 Jakub Jelinek <jakub@redhat.com>
1844 * tree.c (build_simple_mem_ref_loc): Handle
1845 get_addr_base_and_unit_offset returning a MEM_REF.
1848 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
1851 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
1854 2017-11-28 David Malcolm <dmalcolm@redhat.com>
1856 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
1858 (layout::move_to_column): Likewise.
1860 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1863 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
1864 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
1866 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1867 Martin Jambor <mjambor@suse.cz>
1870 * tree.h (expr_type_first_operand_type_p): Declare
1871 * tree.c (expr_type_first_operand_type_p): New function.
1872 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
1873 (ipa_value_from_jfunc): Adjust declaration.
1874 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
1875 Use it as result type for arithmetics, unless it is NULL in which case
1876 be more conservative.
1877 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
1878 ipa_get_jf_pass_through_result.
1879 (propagate_vals_across_pass_through): Likewise.
1880 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
1881 is to propagate_vals_across_pass_through.
1882 (propagate_constants_across_call): Pass PARM_TYPE to
1883 propagate_scalar_across_jump_function.
1884 (find_more_scalar_values_for_callers_subset): Pass parameter type to
1885 ipa_value_from_jfunc.
1886 (cgraph_edge_brings_all_scalars_for_node): Likewise.
1887 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
1888 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
1889 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
1890 target_type, pass it to ipa_value_from_jfunc.
1891 (update_indirect_edges_after_inlining): Pass parameter type to
1892 try_make_edge_direct_simple_call.
1894 2017-11-28 Jeff Law <law@redhat.com>
1896 * gimple-ssa-evrp-analyze.c
1897 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
1898 refine ranges if scev_initialized_p returns true.
1899 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
1901 2017-11-28 Julia Koval <julia.koval@intel.com>
1903 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
1904 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
1905 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
1906 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
1907 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
1908 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
1909 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
1910 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
1911 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
1912 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
1913 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
1914 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
1915 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
1916 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
1917 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
1918 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
1919 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
1920 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
1921 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
1922 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
1923 __builtin_ia32_vpshrd_v2di_mask): New builtins.
1924 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
1926 2017-11-28 Julia Koval <julia.koval@intel.com>
1928 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
1929 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
1930 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
1931 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
1932 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
1933 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
1934 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
1935 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
1936 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
1937 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
1938 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
1939 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
1940 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
1941 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
1942 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
1943 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
1944 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
1945 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
1946 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
1947 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
1948 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
1949 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
1950 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
1951 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
1952 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
1953 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
1954 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
1955 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
1956 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
1957 __builtin_ia32_vpshld_v2di_mask): New builtins.
1958 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1959 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
1961 2017-11-28 Richard Biener <rguenther@suse.de>
1963 PR tree-optimization/80776
1964 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
1966 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
1968 (evrp_range_analyzer::record_ranges_from_incoming_edges):
1969 If the incoming edge is an effective fallthru because the other
1970 edge only reaches a __builtin_unreachable () then record ranges
1971 derived from the controlling condition in SSA info.
1972 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
1973 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
1975 2017-11-28 Olivier Hainque <hainque@adacore.com>
1977 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
1978 on mingw build hosts.
1980 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1983 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
1985 2017-11-28 Richard Biener <rguenther@suse.de>
1988 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
1989 copies generated from memcpy use a character array as reference
1992 2017-11-28 Julia Koval <julia.koval@intel.com>
1993 Sebastian Peryt <sebastian.peryt@intel.com>
1995 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
1996 c-family/c-cilkplus.o, c-family/array-notation-common.o,
1997 cilk-common.o, cilk.h, cilk-common.c): Remove.
1999 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
2000 * builtins.c (is_builtin_name): Remove cilkplus condition.
2001 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
2002 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
2003 cilk-builtins.def, cilkplus.def): Remove.
2004 * cif-code.def (CILK_SPAWN): Remove.
2005 * cilk-builtins.def: Delete.
2006 * cilk-common.c: Ditto.
2008 * cilkplus.def: Ditto.
2009 * config/darwin.h (fcilkplus): Delete.
2010 * cppbuiltin.c: Ditto.
2011 * doc/extend.texi: Remove cilkplus doc.
2012 * doc/generic.texi: Ditto.
2013 * doc/invoke.texi: Ditto.
2014 * doc/passes.texi: Ditto.
2015 * gcc.c (fcilkplus): Remove.
2016 * gengtype.c (cilk.h): Remove.
2017 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
2019 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
2021 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
2023 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
2024 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
2025 cilkplus conditions.
2026 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
2027 inline_read_section): Ditto.
2028 * ipa-inline-analysis.c (cilk.h): Remove.
2029 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
2030 * lto-wrapper.c (merge_and_complain, append_compiler_options,
2031 append_linker_options): Remove condition for fcilkplus.
2032 * lto/lto-lang.c (cilk.h): Remove.
2033 (lto_init): Remove condition for fcilkplus.
2034 * omp-expand.c (expand_cilk_for_call): Delete.
2035 (expand_omp_taskreg, expand_omp_for_static_chunk,
2036 expand_omp_for): Remove cilkplus
2038 (expand_cilk_for): Delete.
2039 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
2040 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
2041 execute_lower_omp, diagnose_sb_0): Ditto.
2042 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
2043 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2044 * tree-nested.c: Ditto.
2045 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
2046 (dump_generic_node): Ditto.
2047 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2048 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
2050 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
2052 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2054 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
2055 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
2056 for vector mode and !TARGET_SIMD.
2058 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2060 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
2062 * tree.h (SWITCH_LABELS): Remove.
2063 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
2064 assert SWITCH_BODY is non-NULL.
2065 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
2067 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
2069 PR tree-optimization/80788
2070 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
2071 has TREE_OVERFLOW set, call drop_tree_overflow.
2073 2017-11-28 Richard Biener <rguenther@suse.de>
2075 PR tree-optimization/83158
2076 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
2078 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
2081 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
2082 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
2084 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
2086 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
2087 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
2089 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2092 * varasm.c (compare_constant): Take the mode of the constants into
2093 account when comparing floating point constants.
2095 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
2097 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
2098 from explicit instantiation of debug_helper.
2099 * vec.h (DEFINE_DEBUG_VEC): Ditto.
2101 2017-11-27 Richard Biener <rguenther@suse.de>
2103 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
2106 2017-11-27 Richard Biener <rguenther@suse.de>
2108 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
2109 (type_hash_canon): Also clear int_cst_hash_table entry for
2111 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
2113 2017-11-27 Tamar Christina <tamar.christina@arm.com>
2115 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2116 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2117 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2119 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2121 * hash-map.h (gt_cleare_cache): Avoid UB.
2123 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
2125 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
2126 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
2127 (decide_unroll_runtime_iterations): Likewise.
2128 (decide_unroll_stupid): Likewise.
2130 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2133 * config/i386/i386.md: Add a loop with incssp.
2135 2017-11-27 Martin Jambor <mjambor@suse.cz>
2137 PR tree-optimization/81248
2138 * tree-sra.c (splice_param_accesses): Remove size check.
2139 (decide_one_param_reduction): Fix size check.
2140 * gimple-pretty-print.c (dump_profile): Silence warning.
2141 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
2143 2017-11-27 Jakub Jelinek <jakub@redhat.com>
2146 * dbxout.c (lastlineno): New variable.
2147 (dbx_debug_hooks): Use dbxout_switch_text_section as
2148 switch_text_section debug hook.
2149 (dbxout_function_end): Switch to current_function_section
2150 rather than function_section. If crtl->has_bb_partition,
2151 output just one N_FUN, depending on in_cold_section_p.
2152 (dbxout_source_line): Remember last lineno in lastlineno.
2153 (dbxout_switch_text_section): New function.
2154 (dbxout_function_decl): Adjust dbxout_block caller.
2155 (dbx_block_with_cold_children): New function.
2156 (dbxout_block): Return true if any LBRAC/RBRAC have been
2157 emitted. Use dbx_block_with_cold_children at depth == 0
2158 in second partition. Add PARENT_BLOCKNUM argument, pass
2159 it optionally adjusted to children. Output LBRAC/RBRAC
2160 around recursive call only if the block is in the current
2161 partition, if not and anything was output, emit empty
2163 * final.c (final_scan_insn): Compute cold_function_name
2164 before calling switch_text_section debug hook. Call
2165 that hook even if dwarf2out_do_frame if not emitting
2169 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
2170 TREE_READONLY decls.
2172 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2174 PR rtl-optimization/82488
2175 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
2177 2017-11-26 Julia Koval <julia.koval@intel.com>
2179 * config/i386/i386.c (processor_target_table): Add skylake_cost for
2181 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
2184 2017-11-26 Julia Koval <julia.koval@intel.com>
2186 * config/i386/driver-i386.c (host_detect_local_cpu):
2187 Detect skylake-avx512.
2189 2017-11-26 Julia Koval <julia.koval@intel.com>
2191 * config.gcc: Add -march=cannonlake.
2192 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
2193 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
2194 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
2195 (PTA_CANNONLAKE): New.
2196 (processor_target_table): Add cannonlake.
2197 (ix86_option_override_internal): Ditto.
2198 (fold_builtin_cpu): Ditto.
2199 (get_builtin_code_for_version): Handle cannonlake.
2200 (M_INTEL_COREI7_CANNONLAKE): New.
2201 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
2202 * doc/invoke.texi: Add -march=cannonlake.
2204 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
2206 * plugin.c (add_new_plugin): Use platform-specific library extensions.
2207 (try_init_one_plugin): Alternative implementation for MinGW.
2208 * Makefile.in (plugin_implib): New.
2209 (gengtype-lex.c): Fix broken AIX workaround.
2210 * configure: Regenerate.
2211 * doc/plugins.texi: Document support for MinGW.
2213 2017-11-25 Jakub Jelinek <jakub@redhat.com>
2215 PR rtl-optimization/81553
2216 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
2217 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
2218 is a shift where C1 has different mode than the whole shift, use C1's
2219 mode for MULT rather than the shift's mode.
2222 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
2223 builtins not enabled in the currently selected ISA.
2225 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
2227 PR tree-optimization/71026
2228 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
2229 (insert_reciprocals): Change to insert reciprocals before a division
2230 by a square and to insert the square of a reciprocal.
2231 (execute_cse_reciprocals_1): Change to consider division by a square.
2232 (register_division_in): Add importance parameter.
2234 2017-11-24 Richard Biener <rguenther@suse.de>
2236 PR tree-optimization/82402
2237 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
2238 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
2240 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
2242 * match.pd (0-ptr): New transformation.
2244 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
2247 * ipa-inline.c (inline_small_functions): Set current badnes correctly
2248 when skipping checking.
2250 2017-11-24 Richard Biener <rguenther@suse.de>
2252 PR tree-optimization/83128
2253 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
2254 (vn_reference_lookup_3): Likewise.
2256 2017-11-24 Jakub Jelinek <jakub@redhat.com>
2259 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
2260 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
2263 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
2264 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
2265 check for the latter with gimple_call_builtin_p. Do not handle
2266 BUILT_IN_STPNCPY_CHK which is not a pass through call.
2268 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
2270 * config/arm/arm_neon.h: Fix pragma GCC push_options before
2273 2017-11-23 Julia Koval <julia.koval@intel.com>
2275 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
2276 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
2277 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
2278 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
2279 _mm512_maskz_expandloadu_epi16): New intrinsics.
2280 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
2281 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
2282 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
2283 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
2284 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
2285 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
2286 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
2287 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
2288 _mm256_maskz_expandloadu_epi8): New intrinsics.
2289 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
2290 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
2291 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
2292 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
2293 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
2294 * config/i386/sse.md (VI248_VLBW): New iterator.
2295 (expand<mode>_mask, expand<mode>_maskz): New patterns.
2297 2017-11-23 Julia Koval <julia.koval@intel.com>
2299 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
2300 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
2301 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
2302 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
2303 _mm512_mask_compressstoreu_epi16): New.
2304 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
2305 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
2306 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
2307 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
2308 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
2309 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
2310 _mm256_mask_compressstoreu_epi8): New.
2311 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
2312 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
2313 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
2314 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
2315 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
2316 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
2317 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
2318 __builtin_ia32_compresshi128_mask,
2319 __builtin_ia32_compressstoreuqi512_mask,
2320 __builtin_ia32_compressstoreuhi512_mask,
2321 __builtin_ia32_compressstoreuqi256_mask,
2322 __builtin_ia32_compressstoreuqi128_mask,
2323 __builtin_ia32_compressstoreuhi256_mask,
2324 __builtin_ia32_compressstoreuhi128_mask): New builtins.
2325 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
2327 (ix86_expand_special_args_builtin): Handle new types.
2328 (s4fma_expand): Handle new builtin array.
2329 * config/i386/immintrin.h: Include new headers.
2330 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
2331 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
2333 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2335 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
2336 fixes. Declare temp and g variables at the top in order to avoid
2337 {} in most of the cases.
2339 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
2341 * match.pd (ptr-0): New transformation.
2343 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2345 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
2346 (struct addr_mode_cost_table): New.
2347 (struct tune_params): Add field addr_mode_costs.
2348 * config/arm/arm.c (generic_addr_mode_costs): New.
2349 (arm_slowmul_tune): Initialise addr_mode_costs field.
2350 (arm_fastmul_tune): Likewise.
2351 (arm_strongarm_tune): Likewise.
2352 (arm_xscale_tune): Likewise.
2353 (arm_9e_tune): Likewise.
2354 (arm_marvell_pj4_tune): Likewise.
2355 (arm_v6t2_tune): Likewise.
2356 (arm_cortex_tune): Likewise.
2357 (arm_cortex_a8_tune): Likewise.
2358 (arm_cortex_a7_tune): Likewise.
2359 (arm_cortex_a15_tune): Likewise.
2360 (arm_cortex_a35_tune): Likewise.
2361 (arm_cortex_a53_tune): Likewise.
2362 (arm_cortex_a57_tune): Likewise.
2363 (arm_exynosm1_tune): Likewise.
2364 (arm_xgene1_tune): Likewise.
2365 (arm_cortex_a5_tune): Likewise.
2366 (arm_cortex_a9_tune): Likewise.
2367 (arm_cortex_a12_tune): Likewise.
2368 (arm_cortex_a73_tune): Likewise.
2369 (arm_v7m_tune): Likewise.
2370 (arm_cortex_m7_tune): Likewise.
2371 (arm_v6m_tune): Likewise.
2372 (arm_fa726te_tune): Likewise.
2373 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
2375 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2377 * config/arm/arm.c (arm_mem_costs): New function.
2378 (arm_rtx_costs_internal): Use arm_mem_costs.
2380 2017-11-23 Mark Wielaard <mark@klomp.org>
2382 * dwarf2out.c (init_sections_and_labels): Use generation to create
2383 unique ranges_section_label and ranges_base_label. Return generation.
2384 (output_rnglists): Add generation argument. Use generation to create
2385 unique ranges labels.
2386 (dwarf2out_finish): Get generation from init_sections_and_labels
2387 and pass generation to output_rnglists.
2389 2017-11-23 Mike Stump <mikestump@comcast.net>
2390 Eric Botcazou <ebotcazou@adacore.com>
2392 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
2393 * cfgloop.h (struct loop): Add unroll field.
2394 * function.h (struct function): Add has_unroll bitfield.
2395 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
2396 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
2397 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
2398 (pass_rtl_unroll_loops::gate): Likewise.
2399 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
2400 for which loop->unroll==1.
2401 (decide_unroll_constant_iterations): Use note for consistency and
2402 take loop->unroll into account. Return early if loop->unroll is set.
2403 Fix thinko in existing test.
2404 (decide_unroll_runtime_iterations): Use note for consistency and
2405 take loop->unroll into account.
2406 (decide_unroll_stupid): Likewise.
2407 * lto-streamer-in.c (input_cfg): Read loop->unroll.
2408 * lto-streamer-out.c (output_cfg): Write loop->unroll.
2409 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
2411 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
2412 (print_loop): Print loop->unroll if set.
2413 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
2414 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
2415 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
2417 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
2418 loop->unroll is set and smaller than the trip count. Otherwise bypass
2419 entirely the heuristics if loop->unroll is set. Remove dead note.
2420 Fix off-by-one bug in other note.
2421 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
2422 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
2424 (tree_unroll_loops_completely): Make static.
2425 (pass_complete_unroll::execute): Use correct type for variable.
2426 (pass_complete_unrolli::execute): Fix formatting.
2427 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
2429 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
2431 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
2432 enable when TARGET_PREFER_AVX128 is set.
2434 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2436 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
2438 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
2439 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
2441 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2443 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
2444 frequencies. Fix estimation of aggregate parameters.
2446 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2448 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
2451 2017-11-23 Tom de Vries <tom@codesourcery.com>
2453 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
2455 2017-11-23 Tom de Vries <tom@codesourcery.com>
2457 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
2460 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
2463 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
2464 sibcall_value_pcrel_fdpic): Use local variable instead of
2466 (calli_tbr_rel): Add missing operand 2.
2467 (call_valuei_tbr_rel): Add missing operand 3.
2469 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2472 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
2473 bitpos/bitsize covering the whole destination, use store_expr only if
2474 the complex mode is the same. Otherwise, use expand_normal and if
2475 it returns CONCAT, subreg each part separately instead of trying to
2476 subreg the whole result.
2478 2017-11-23 Richard Biener <rguenther@suse.de>
2480 PR tree-optimization/23094
2481 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
2482 come here from walking over backedges in the first iteration.
2483 (vn_reference_lookup_3): Skip clobbers that store the same value.
2485 2017-11-23 Richard Biener <rguenther@suse.de>
2487 PR tree-optimization/81403
2488 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
2489 a block we need a leader relative to.
2490 (phi_translate_1): For nary processing require a leader from
2491 get_representative_for given we run expression simplification
2492 using match-and-simplify. Remove previous fix.
2494 2017-11-22 Jeff Law <law@redhat.com>
2496 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
2497 Use new method allocate_value_range rather than accessing the
2498 vrp_value_range_pool data member directly.
2499 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
2500 to use extract_range_from_stmt method to avoid need for
2501 extract_range_from_assignment method.
2502 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
2503 method rather than setting values_propgated data member directly.
2504 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
2505 and values propagated data members and extract_range_from_assignment
2506 method. Reorder private data members to conform to standards.
2507 Add new methods set_lattice_propagation_complete and
2508 allocate_value_range.
2510 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
2512 PR rtl-optimization/83030
2513 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
2514 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
2515 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
2516 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
2517 CROSSING_JUMP_P flag is already set before setting it.
2518 * cfgrtl.c (fixup_partition_crossing): Likewise.
2519 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
2522 2017-11-22 Jakub Jelinek <jakub@redhat.com>
2524 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
2525 Handle the case where both arguments are using gen_const_vec_series.
2527 2017-11-22 David Malcolm <dmalcolm@redhat.com>
2530 * pretty-print.c (pp_format): Move quoting implementation to
2531 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
2532 to pass address of "quote" local.
2533 (pp_begin_quote): New function.
2534 (pp_end_quote): New function.
2535 * pretty-print.h (printer_fn): Convert penultimate param from bool
2537 (pp_begin_quote): New decl.
2538 (pp_end_quote): New decl.
2539 * tree-diagnostic.c (default_tree_printer): Convert penultimate
2540 param from bool to bool *.
2541 * tree-diagnostic.h (default_tree_printer): Likewise.
2543 2017-11-22 Jeff Law <law@redhat.com>
2545 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
2546 Initialize vr_values.
2547 (evrp_range_analyzer::try_find_new_range): Call methods attached to
2548 vr_values via vr_values class instance rather than delegators.
2549 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
2550 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
2551 (evrp_range_analyzer::push_value_range): Likewise.
2552 (evrp_range_analyzer::pop_value_range): Likewise.
2553 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
2554 most delegators. Those remaining are exposed as public interfaces.
2555 Make vr_values a pointer and private.
2556 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
2558 (evrp_range_analyzer::get_vr_value): New method.
2559 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
2560 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
2561 (class evrp_dom_walker): Attach evrp_folder class, initialize
2562 it in the ctor. Remove temporary delegators.
2563 (evrp_dom_walker::before_dom_children): Call methods in attached
2564 evrp_range_analyzer class via class instance pointer. Use
2565 free value_range_constant_singleton to remove need for
2566 op_with_constant_singleton_value delegator method. Do not
2567 create a vrp_prop class instance for every call! Narrow
2568 scope of a couple variables.
2569 (evrp_dom_walker::cleanup): Call methods in attached
2570 evrp_range_analyzer class via class instance pointer.
2571 * vr-values.h (class vr_values): Privatize many methods and
2574 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2576 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
2577 * cfgexpand.c (expand_debug_expr): Remove handling for them.
2578 * expr.c (expand_expr_real_2): Likewise.
2579 * fold-const.c (const_unop): Likewise.
2580 * optabs-tree.c (optab_for_tree_code): Likewise.
2581 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2582 * tree-inline.c (estimate_operator_cost): Likewise.
2583 * tree-pretty-print.c (dump_generic_node): Likewise.
2584 (op_code_prio): Likewise.
2585 (op_symbol_code): Likewise.
2586 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
2587 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
2588 * internal-fn.c (direct_internal_fn_optab): New function.
2589 (direct_internal_fn_array, direct_internal_fn_supported_p
2590 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
2591 * fold-const-call.c (fold_const_reduction): New function.
2592 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
2594 * tree-vect-loop.c: Include internal-fn.h.
2595 (reduction_code_for_scalar_code): Rename to...
2596 (reduction_fn_for_scalar_code): ...this and return an internal
2598 (vect_model_reduction_cost): Take an internal_fn rather than
2600 (vect_create_epilog_for_reduction): Likewise. Build calls rather
2602 (vectorizable_reduction): Use internal functions rather than tree
2603 codes for the reduction operation. Update calls to the functions
2605 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
2606 Use calls to internal functions rather than REDUC tree codes.
2607 * config/aarch64/aarch64-simd.md: Update comment accordingly.
2609 2017-11-22 Olivier Hainque <hainque@adacore.com>
2611 * config/vxworks.c (vxworks_override_options): Pick default
2612 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
2613 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
2614 DWARF_GNAT_ENCODINGS_DEFAULT.
2615 * config/vxworksae.h: Likewise.
2617 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
2619 PR tree-optimization/83104
2620 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
2623 2017-11-22 Marek Polacek <polacek@redhat.com>
2624 H.J. Lu <hongjiu.lu@intel.com>
2625 Jason Merrill <jason@redhat.com>
2630 * calls.c (initialize_argument_information): Call
2631 warn_parameter_passing_abi target hook.
2632 (store_one_arg): Use 0 for empty record size. Don't push 0 size
2633 argument onto stack.
2634 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
2635 * common.opt: Update -fabi-version description.
2636 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
2637 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
2639 (ix86_is_empty_record): New function.
2640 (ix86_warn_parameter_passing_abi): New function.
2641 (TARGET_EMPTY_RECORD_P): Redefine.
2642 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
2643 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
2644 * doc/tm.texi: Regenerated.
2645 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
2646 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
2647 * dwarf2out.c (get_ultimate_context): Move to tree.c.
2648 * explow.c (hard_function_value): Call arg_int_size_in_bytes
2649 instead of int_size_in_bytes.
2650 * expr.c (copy_blkmode_to_reg): Likewise.
2651 * function.c (aggregate_value_p): Return 0 for empty types.
2652 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
2654 (locate_and_pad_parm): Call arg size_in_bytes instead
2656 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
2657 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
2658 * target.def (empty_record_p, warn_parameter_passing_abi): New target
2660 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
2661 (std_gimplify_va_arg_expr): Skip empty records. Call
2662 arg_size_in_bytes instead size_in_bytes.
2663 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
2664 * tree-core.h (tree_type_common): Add empty_flag.
2665 (tree_decl_common): Update comments.
2666 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
2668 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
2669 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
2671 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
2672 * tree.c (default_is_empty_type): New function.
2673 (default_is_empty_record): New function.
2674 (arg_int_size_in_bytes): New function.
2675 (arg_size_in_bytes): New function.
2676 (get_ultimate_context): New function.
2677 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
2678 TRANSLATION_UNIT_WARN_EMPTY_P.
2679 (default_is_empty_record, arg_int_size_in_bytes,
2680 arg_size_in_bytes, get_ultimate_context): Declare.
2682 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2684 * config/arm/arm.c (cmse_clear_registers): New function.
2685 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
2686 code by call to cmse_clear_registers.
2687 (cmse_nonsecure_entry_clear_before_return): Likewise.
2689 2017-11-22 Tamar Christina <tamar.christina@arm.com>
2691 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
2692 (vdot_s32, vdotq_s32): New.
2693 (vdot_lane_u32, vdotq_lane_u32): New.
2694 (vdot_lane_s32, vdotq_lane_s32): New.
2697 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2700 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
2701 for unsigned values with fewer HWIs than the precision.
2702 (test_overflow): New function.
2703 (wide_int_cc_tests): Call it.
2705 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2707 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
2708 has the same mode class as Pmode.
2710 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2712 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
2713 padding_bits_to_clear_ptr.
2714 (cmse_nonsecure_entry_clear_before_return): Likewise.
2716 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2718 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
2719 auto_sbitap instead of integer bitfield to control register needing
2722 2017-11-22 Jakub Jelinek <jakub@redhat.com>
2724 PR tree-optimization/83044
2725 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
2726 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
2727 that need the upper bound. Subtract offset from
2728 get_addr_base_and_unit_offset only if positive and subtract it
2729 before division by eltsize rather than after it.
2732 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
2733 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
2734 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
2738 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
2739 expand_binop with *mul_widen_optab, make sure at least one of the
2740 operands doesn't have VOIDmode.
2743 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
2745 PR rtl-optimization/82044
2746 PR tree-optimization/82042
2747 * dse.c (record_store): Check for overflow.
2748 (check_mem_read_rtx): Properly check for overflow if width == -1, call
2749 add_wild_read instead of clear_rhs_from_active_local_stores on
2750 overflow and log it into dump_file.
2752 2017-11-22 Richard Biener <rguenther@suse.de>
2754 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
2755 fake edges to exit when looking for a place to insert.
2756 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
2758 (insert_into_preds_of_block): Commit edge insertion immediately,
2759 assert that doesn't require new BBs.
2760 (fini_pre): Release expressions.
2761 (pass_pre::execute): Shuffle things around a bit, if the fn
2762 is too large do not compute AVAIL either as this is really the
2765 2017-11-22 Richard Biener <rguenther@suse.de>
2767 PR tree-optimization/83089
2768 * tree-if-conv.c (pass_if_conversion::execute): If anything
2769 changed reset SCEV and free the number of iteration estimates.
2771 2017-11-21 Martin Sebor <msebor@redhat.com>
2773 PR tree-optimization/82945
2774 * calls.h (warn_nonstring_bound): Remove unused function.
2776 2017-11-21 Martin Sebor <msebor@redhat.com>
2778 PR tree-optimization/82945
2779 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
2780 * calls.h (maybe_warn_nonstring_arg): Declare new function.
2781 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
2783 (initialize_argument_information): Call maybe_warn_nonstring_arg.
2784 * calls.h (get_attr_nonstring_decl): Declare new function.
2785 * doc/extend.texi (attribute nonstring): Update.
2786 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
2787 get_attr_nonstring_decl and handle it.
2788 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
2789 detection of nul-termination.
2790 (strlen_to_stridx): Change to a pointer.
2791 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
2792 (pass_strlen::execute): Same.
2794 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
2796 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
2797 for the new option -mprefer-vector-width=[none|128|256|512].
2798 * config/i386/i386.c (ix86_target_string): remove old style options
2799 -mprefer-avx256 and make -mprefer-avx128 as alias.
2800 (ix86_option_override_internal): Apply defaults for the
2801 -mprefer-vector-width=[128|256] option.
2802 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
2803 Implement macros to work with -mprefer-vector-width=.
2804 * config/i386/i386.opt: Implemented option
2805 -mprefer-vector-width=[none|128|256|512].
2806 * doc/invoke.texi: Documentation for
2807 -mprefer-vector-width=[none|128|256|512].
2809 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
2811 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
2813 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
2815 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
2817 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
2819 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
2821 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
2822 * expr.c (expand_expr_real_2): Likewise.
2823 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
2824 fold_binary_loc): Likewise.
2825 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
2826 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
2827 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
2828 MINUS_EXPR transformations.
2829 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
2830 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
2831 * tree-inline.c (estimate_operator_cost): Likewise.
2832 * tree-pretty-print.c (dump_generic_node, op_code_prio,
2833 op_symbol_code): Likewise.
2834 * tree-vect-stmts.c (vectorizable_operation): Likewise.
2835 * vr-values.c (extract_range_from_binary_expr): Likewise.
2836 * varasm.c (initializer_constant_valid_p_1): Likewise.
2837 * tree.def: New tree code POINTER_DIFF_EXPR.
2839 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
2841 * config/i386/i386.md (*bswap<mode>2_movbe): Add
2842 integer suffix to movbe mnemonic.
2843 (*bswaphi2_movbe): Ditto.
2844 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
2846 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2849 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
2850 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
2851 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
2853 2017-11-21 Martin Liska <mliska@suse.cz>
2855 * tree-inline.c (expand_call_inline): Remove not needed
2858 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
2859 Jakub Jelinek <jakub@redhat.com>
2862 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
2863 Remove static keyword from f variable.
2865 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2867 PR tree-optimization/83086
2868 * gimple-ssa-store-merging.c
2869 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
2870 rather than n.base_addr.
2872 2017-11-21 Martin Liska <mliska@suse.cz>
2874 PR rtl-optimization/82044
2875 PR tree-optimization/82042
2876 * dse.c (check_mem_read_rtx): Check for overflow.
2878 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2880 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
2883 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2885 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
2886 and its corresponding call arg location note.
2888 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2889 Andrew Burgess <andrew.burgess@embecosm.com>
2891 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
2893 (arc_return_slot_offset): Likewise.
2894 (arc_eh_return_address_location): New declaration.
2895 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
2896 (MUST_SAVE_REGISTER): Add exception handler case.
2897 (MUST_SAVE_RETURN_ADDR): Likewise.
2898 (arc_frame_pointer_required): Likewise.
2899 (arc_frame_pointer_needed): New function.
2900 (arc_compute_frame_size): Changed.
2901 (arc_expand_prologue): Likewise.
2902 (arc_expand_epilogue): Likewise.
2903 (arc_initial_elimination_offset): Likewise.
2904 (arc_return_slot_offset): Delete.
2905 (arc_eh_return_address_location): New function.
2906 (arc_builtin_setjmp_frame_value): Likewise.
2907 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
2908 (EH_RETURN_STACKADJ_RTX): Define.
2909 (EH_RETURN_HANDLER_RTX): Likewise.
2910 * config/arc/arc.md (eh_return): Delete.
2912 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2914 * print-tree.h (debug_vec_tree): Remove prototype.
2915 * gdbinit.in (pvt): Remove macro.
2917 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2919 PR tree-optimization/83047
2920 * gimple-ssa-store-merging.c
2921 (imm_store_chain_info::output_merged_store): If the loads with the
2922 same vuse are in different basic blocks, for load_gsi pick a load
2923 location that is dominated by the other loads.
2926 * config/i386/i386.c (ix86_memmodel_check): Start
2927 -Winvalid-memory-model diagnostics with lowercase letter.
2930 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
2931 set in_cold_section_p to the partition containing loc_list->first.
2932 When seeing loc_list->last_before_switch node, update secname and
2933 perform range_across_switch second partition handling only after that.
2936 * run-rtl-passes.c: Include debug.h.
2937 (run_rtl_passes): Call debug_hooks->assembly_start.
2938 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
2942 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
2943 OPTAB_DIRECT in calls to expand_simple_binop.
2945 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2948 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
2950 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2953 * spellcheck.h (best_match::blithely_get_best_candidate): New
2956 2017-11-20 Jakub Jelinek <jakub@redhat.com>
2958 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
2959 with lower case letter.
2961 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
2963 * config/i386/i386.md (bswaphi2): New expander.
2964 (*bswaphi2_movbe): New insn pattern.
2965 (bswaphi -> rorhi pepehole2): New peephole pattern.
2967 2017-11-20 Jeff Law <law@redhat.com>
2969 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
2970 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
2971 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
2972 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
2973 gimple-ssa-evrp-analyze.h.
2975 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
2976 set BB_VISITED here.
2977 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
2979 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2981 * tree-predcom.c: Add general comment on Store-Store chains.
2982 (split_data_refs_to_components): Postpone clearing eliminate_store_p
2984 (get_chain_last_ref_at): Rename into...
2985 (get_chain_last_write_at): ...this.
2986 (get_chain_last_write_before_load): New function.
2987 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
2988 CT_STORE_STORE when write reference is added.
2989 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
2990 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
2991 (initialize_root_vars_store_elim_1): Ditto.
2992 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
2993 definition is created.
2994 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
2995 chain by replacing it with dominant stored value.
2997 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2999 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
3001 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
3003 * vr-values.c (extract_range_from_binary_expr): Use a full range
3006 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
3008 * config/arm/arm.md (R4_REGNUM): Define constant.
3009 (nonsecure_call_internal): Remove r4 clobber.
3010 (nonsecure_call_value_internal): Likewise.
3011 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
3012 clobber and resequence match_operands.
3013 (nonsecure_call_value_reg_thumb1_v5): Likewise.
3014 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
3015 (nonsecure_call_value_reg_thumb2): Likewise.
3017 2017-11-20 Jakub Jelinek <jakub@redhat.com>
3019 PR tree-optimization/78821
3020 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
3021 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
3022 to the address of the base rather than the base itself.
3023 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
3024 (find_bswap_or_nop_finalize): New function.
3025 (find_bswap_or_nop): Use it.
3026 (bswap_replace): Return a tree rather than bool, change first
3027 argument from gimple * to gimple_stmt_iterator, allow inserting
3028 into an empty sequence, allow ins_stmt to be NULL - then emit
3029 all stmts into gsi. Fix up MEM_REF address gimplification.
3030 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
3031 (struct store_immediate_info): Add N and INS_STMT non-static
3033 (store_immediate_info::store_immediate_info): Initialize them
3034 from newly added ctor args.
3035 (merged_store_group::apply_stores): Formatting fixes. Sort by
3037 (stmts_may_clobber_ref_p): For stores call also
3038 refs_anti_dependent_p.
3039 (gather_bswap_load_refs): New function.
3040 (imm_store_chain_info::try_coalesce_bswap): New method.
3041 (imm_store_chain_info::coalesce_immediate_stores): Use it.
3042 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
3043 (imm_store_chain_info::output_merged_store): Fail if number of
3044 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
3045 and NOP_EXPR rhs_code.
3046 (pass_store_merging::process_store): Compute n and ins_stmt, if
3047 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
3048 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
3050 (pass_store_merging::execute): Calculate dominators.
3052 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
3053 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
3054 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
3055 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3056 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
3057 class pass_optimize_bswap, bswap_replace,
3058 pass_optimize_bswap::execute): Moved to ...
3059 * gimple-ssa-store-merging.c: ... this file.
3060 Include optabs-tree.h.
3061 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
3062 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3063 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
3064 anonymous namespace, remove static keywords.
3065 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
3066 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
3068 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
3071 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
3073 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
3075 * vec.h (debug_helper): New function.
3076 (DEFINE_DEBUG_VEC): New macro.
3077 * hash-set.h (debug_helper): New function.
3078 (DEFINE_DEBUG_HASH_SET): New macro.
3079 * cfg.c (debug_slim (edge)): New function.
3080 Call DEFINE_DEBUG_VEC for edges.
3081 Call DEFINE_DEBUG_HASH_SET for edges.
3082 * cfghooks.c (debug_slim (basic_block)): New function.
3083 Call DEFINE_DEBUG_VEC for basic blocks.
3084 Call DEFINE_DEBUG_HASH_SET for basic blocks.
3085 * print-tree.c (debug_slim): New function to handle trees.
3086 Call DEFINE_DEBUG_VEC for trees.
3087 Call DEFINE_DEBUG_HASH_SET for trees.
3088 (debug (vec<tree, va_gc>) &): Remove.
3089 (debug (<vec<tree, va_gc>) *): Remove.
3090 * print-rtl.c (debug_slim): New function to handle const_rtx.
3091 Call DEFINE_DEBUG_VEC for rtx_def.
3092 Call DEFINE_DEBUG_VEC for rtx_insn.
3093 Call DEFINE_DEBUG_HASH_SET for rtx_def.
3094 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
3095 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
3096 (debug (vec<rtx_insn *> *ptr): Remove.
3097 (debug_insn_vector): Remove.
3098 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
3100 2017-11-20 Tom de Vries <tom@codesourcery.com>
3102 PR rtl-optimization/82020
3103 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
3104 IF_THEN_ELSE condition.
3106 2017-11-19 Jeff Law <law@redhat.com>
3108 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
3109 of degenerates resulting from ignoring an edge.
3111 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3114 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
3116 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3119 * profile-count.c (profile_count::to_sreal_scale): Fix return value
3120 for uninitialied counts.
3122 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3125 * tree-inline.c (estimate_num_insns): Set to 1 at least.
3127 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3130 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
3132 2017-11-19 Tom de Vries <tom@codesourcery.com>
3134 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
3137 2017-11-19 Tom de Vries <tom@codesourcery.com>
3139 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
3142 2017-11-19 Tom de Vries <tom@codesourcery.com>
3144 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
3147 2017-11-19 Tom de Vries <tom@codesourcery.com>
3149 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
3151 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
3153 2017-11-19 Tom de Vries <tom@codesourcery.com>
3155 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
3157 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
3158 semicolon after MCORE_EXPORT_NAME call.
3160 2017-11-19 Tom de Vries <tom@codesourcery.com>
3163 * vmsdbgout.c (vmsdbgout_early_finish): New function.
3164 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
3166 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
3168 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
3169 (cgraph_edge::clone): Cleanup updating of profile.
3170 * ipa-cp.c (update_profiling_info): Likewise.
3171 * ipa-inline-transform.c (inline_transform): Likewise.
3172 * ipa-inline.c (inline_small_functions): Add missing space to dump.
3173 * ipa-split.c (execute_split_functions): Do not split when function
3175 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
3176 * profile-count.c (profile_count::dump): Add global0.
3177 (profile_count::to_cgraph_frequency): Do not ICE when entry is
3179 (profile_count::to_sreal_scale): Likewise.
3180 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
3181 (profile_count::combine_with_ipa_count): New function.
3182 * profile-count.h (profile_guessed_global0adjusted): New.
3183 (profile_count::adjusted_zero): New.
3184 (profile_count::global0adjusted): New.
3185 (profile_count::combine_with_ipa_count): New.
3186 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
3187 correct profile of return block of split functions.
3188 (copy_cfg_body): Remove unused profile_count.
3189 (copy_body): Likewise.
3190 (expand_call_inline): Update.
3191 (tree_function_versioning): Update.
3193 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
3195 * hash-set.h (hash_set::empty): New.
3196 * tree-ssa-threadbackward.h: Delete.
3197 * tree-ssa-threadbackward.c (class thread_jumps): New.
3198 Move max_threaded_paths into class.
3199 (fsm_find_thread_path): Remove arguments that are now in class.
3200 (profitable_jump_thread_path): Rename to...
3201 (thread_jumps::profitable_jump_thread_path): ...this.
3202 (convert_and_register_jump_thread_path): Rename to...
3203 (thread_jumps::convert_and_register_current_path): ...this.
3204 (check_subpath_and_update_thread_path): Rename to...
3205 (thread_jumps::check_subpath_and_update_thread_path): ...this.
3206 (register_jump_thread_path_if_profitable): Rename to...
3207 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
3208 (handle_phi): Rename to...
3209 (thread_jumps::handle_phi): ...this.
3210 (handle_assignment): Rename to...
3211 (thread_jumps::handle_assignment): ...this.
3212 (fsm_find_control_statement_thread_paths): Rename to...
3213 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
3214 (find_jump_threads_backwards): Rename to...
3215 (thread_jumps::find_jump_threads_backwards): ...this.
3216 Initialize path local data.
3217 (pass_thread_jumps::execute): Call find_jump_threads_backwards
3218 from within thread_jumps class.
3219 (pass_early_thread_jumps::execute): Same.
3221 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3223 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
3225 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3227 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
3228 to canonical location.
3230 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3233 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
3235 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
3237 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
3239 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
3240 rcrt1.o%s/grcrt1.o%s for -static-pie.
3242 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3244 * i386.c (ix86_multiplication_cost, ix86_division_cost,
3245 ix86_shift_rotate_cost): Break out from ...
3246 (ix86_rtx_costs): ... here.
3247 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
3250 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3252 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
3253 when entry block was promoted unlikely.
3254 (estimate_bb_frequencies): Increase frequency scale.
3255 * profile-count.h (profile_count): Export precision info.
3257 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3259 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
3260 disturbin profile of entry block.
3262 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3264 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
3267 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3269 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
3270 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
3272 2017-11-17 Jeff Law <law@redhat.com>
3274 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
3275 from evrp_dom_walker class. Various methods moved into new class.
3276 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
3277 (evrp_range_analyzer::enter): New method.
3278 (evrp_range_analyzer::leave): New method.
3279 (evrp_dom_walker): Remove delegators no longer needed by this class.
3280 Replace vr_values data member with evrp_range_analyzer
3282 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
3283 method extracted from evrp_dom_walker::before_dom_children.
3284 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
3285 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
3287 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
3288 Add private copy constructor and move assignment operators.
3289 Privatize methods and class data where trivially possible.
3290 (evrp_dom_walker::cleanup): New function, extracted from
3291 execute_early_vrp. Simplify access to class data.
3293 * vr-values.h (get_output_for_vrp): Prototype.
3294 * vr-values.c (get_output_for_vrp): New function extracted from
3295 vrp_visit_assignment_or_call and extract_range_from_stmt.
3296 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
3298 2017-11-17 Luis Machado <luis.machado@linaro.org>
3300 * config/aarch64/aarch64.c
3301 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
3302 (qdf24xx_tunings) <autoprefetcher_model>: Set to
3303 tune_params::AUTOPREFETCHER_WEAK.
3305 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3308 * config/arm/arm.c (arm_valid_target_attribute_rec):
3309 Parse "arch=" and "+<ext>".
3310 (arm_valid_target_attribute_tree): Re-init global options.
3311 (arm_option_override): Make non-static.
3312 (arm_options_perform_arch_sanity_checks): Make errors fatal.
3313 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
3314 (__ARM_FEATURE_CRC32): Support undef.
3315 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
3316 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
3318 2017-11-17 David Malcolm <dmalcolm@redhat.com>
3320 * gdbinit.in (break-on-diagnostic): New command.
3322 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3324 * config/i386/i386.c (ix86_expand_epilogue): Change simple
3325 return to indirect jump for EH return if control-flow
3326 protection is enabled. Change explicit 'false' argument in
3327 pro_epilogue_adjust_stack with a value of
3329 * config/i386/i386.md (simple_return_indirect_internal):
3330 Remove SImode restriction to support 64-bit.
3332 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3334 * combine.c (added_notes_insn): New.
3335 (try_combine): Handle added_notes_insn like added_links_insn.
3336 Rewrite return value code.
3337 (distribute_notes): Set added_notes_insn to the earliest insn we added
3340 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3342 PR rtl-optimization/82621
3343 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
3344 dest of one of those SETs is unused.
3346 2017-11-17 Richard Biener <rguenther@suse.de>
3349 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
3350 * tree-pretty-print.c (dump_generic_node): Handle
3351 annot_expr_parallel_kind.
3352 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
3353 * gimplify.c (gimple_boolify): Likewise.
3355 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3357 * config.gcc (extra_headers): Add cet.h for x86 targets.
3358 * config/i386/cet.h: New file.
3359 * doc/install.texi: Add --enable-cet/--disable-cet.
3361 2017-11-17 Richard Biener <rguenther@suse.de>
3363 PR tree-optimization/83017
3364 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
3365 (gen_parallel_loop): Properly count iterations.
3366 (parallelize_loops): Handle loop->can_be_parallel independent
3367 of flag_loop_parallelize_all. Make static profitability test match
3369 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
3370 * invoke.texi (parloops-min-per-thread): Document.
3372 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
3374 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
3375 upstreaming review comments.
3377 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3379 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
3380 with fast unaligned access.
3381 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
3383 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
3385 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
3386 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
3388 2017-11-17 Richard Biener <rguenther@suse.de>
3390 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
3391 folding of references.
3393 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
3396 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
3397 of replacing call to strncmp with corresponding call to strcmp when
3400 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
3402 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
3403 option prefer-avx256 for skylake-avx512 configuration.
3404 * config/i386/i386.c (ix86_option_override_internal): Ditto.
3405 (get_builtin_code_for_version): Ditto.
3407 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3408 Monk Chiang <sh.chiang04@gmail.com>
3410 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
3411 (FIXED_REGISTERS): Reserve more register numbers.
3412 (CALL_USED_REGISTERS): Likewise.
3413 (REG_ALLOC_ORDER): Likewise.
3414 (REG_CLASS_CONTENTS): Likewise.
3415 (REGISTER_NAMES): Likewise.
3417 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3418 Kito Cheng <kito.cheng@gmail.com>
3420 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
3422 * config/nds32/iterators.md: Add vector mode iterators and attributes.
3424 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
3426 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
3427 parameter list for vec_splats.
3429 2017-11-16 Joseph Myers <joseph@codesourcery.com>
3431 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
3433 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
3435 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
3438 * doc/install.texi: Document incompatibility of Perl >=5.6.26
3439 with the required version of automake 1.11.6.
3441 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
3443 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
3445 (DU_C2_3_power9): Correct reservation combinations.
3446 (FP_DIV_power9, VEC_DIV_power9): New.
3447 (power9-alu): Split out rotate/shift...
3448 (power9-rot): ...to here, correct dispatch resource.
3449 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
3451 (power9-fp): Correct latency.
3452 (power9-sdiv): Add div/sqrt resource.
3453 (power9-ddiv): Correct latency, add div/sqrt resource.
3454 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
3455 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
3457 (power9-qpdiv, power9-qpmul): Adjust resource usage.
3459 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
3461 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
3462 switch statement mapping KF built-ins to TF built-ins if we don't
3463 have the proper ISA 3.0 assembler support.
3465 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3467 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
3468 (lower_emutls_function_body): Do not compute it.
3470 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3472 * ipa-split.c (split_bb_info): Turn time to sreal.
3473 (split_point): Likewise.
3474 (dump_split_point): Likewise.
3475 (fine_split_points): Likewise.
3476 (execute_split_functions): Only zero split_bbs; turn time to sreals.
3478 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3480 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
3483 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3485 * predict.c (combine_predictions_for_bb): Preserve zero predicted
3487 (expensive_function_p): Remove useless assert.
3488 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
3490 2017-11-16 Martin Sebor <msebor@redhat.com>
3492 PR tree-optimization/82588
3493 PR tree-optimization/82583
3494 * tree-vrp.c (check_array_ref): Handle flexible array members,
3495 string literals, and inner indices.
3496 (search_for_addr_array): Add detail to diagnostics.
3498 2017-11-16 Nathan Sidwell <nathan@acm.org>
3502 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
3503 (SET_DECL_ASSEMBLER_NAME): Forward to
3504 overwrite_decl_assembler_name.
3505 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
3506 (overwrite_decl_assembler_name): Declare.
3507 * tree.c (overwrite_decl_assembler_name): New.
3508 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
3509 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
3510 (LANG_HOOKS_INITIALIZER): Add it.
3511 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
3512 * langhooks.c (lhd_set_decl_assembler_name): Use
3513 SET_DECL_ASSEMBLER_NAME.
3514 (lhd_overwrite_decl_assembler_name): Default implementation.
3516 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
3517 Jackson Woodruff <jackson.woodruff@arm.com>
3519 PR tree-optimization/71026
3520 * match.pd: Canonicalize constant multiplies in division.
3522 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
3524 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
3525 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
3526 Remove OPT_fomit_frame_pointer entry.
3527 * common/config/arc/arc-common.c: Likewise.
3528 * common/config/arm/arm-common.c: Likewise.
3529 * common/config/avr/avr-common.c: Likewise.
3530 * common/config/c6x/c6x-common.c: Likewise.
3531 * common/config/cr16/cr16-common.c: Likewise.
3532 * common/config/cris/cris-common.c: Likewise.
3533 * common/config/epiphany/epiphany-common.c: Likewise.
3534 * common/config/fr30/fr30-common.c: Likewise.
3535 * common/config/frv/frv-common.c: Likewise.
3536 * common/config/ia64/ia64-common.c: Likewise.
3537 * common/config/iq2000/iq2000-common.c: Likewise.
3538 * common/config/lm32/lm32-common.c: Likewise.
3539 * common/config/m32r/m32r-common.c: Likewise.
3540 * common/config/mcore/mcore-common.c: Likewise.
3541 * common/config/microblaze/microblaze-common.c: Likewise.
3542 * common/config/mips/mips-common.c: Likewise.
3543 * common/config/mmix/mmix-common.c: Likewise.
3544 * common/config/mn10300/mn10300-common.c: Likewise.
3545 * common/config/nios2/nios2-common.c: Likewise.
3546 * common/config/pa/pa-common.c: Likewise.
3547 * common/config/pdp11/pdp11-common.c: Likewise.
3548 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
3549 * common/config/riscv/riscv-common.c: Likewise.
3550 * common/config/rs6000/rs6000-common.c: Likewise.
3551 * common/config/rx/rx-common.c: Likewise.
3552 * common/config/s390/s390-common.c: Likewise.
3553 * common/config/sh/sh-common.c: Likewise.
3554 * common/config/sparc/sparc-common.c: Likewise.
3555 * common/config/tilegx/tilegx-common.c: Likewise.
3556 * common/config/tilepro/tilepro-common.c: Likewise.
3557 * common/config/v850/v850-common.c: Likewise.
3558 * common/config/visium/visium-common.c: Likewise.
3559 * common/config/xstormy16/xstormy16-common.c: Likewise.
3560 * common/config/xtensa/xtensa-common.c: Likewise.
3561 * invoke.texi (-fomit-frame-pointer): Update documentation.
3563 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3565 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
3567 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3569 * cfg.c (scale_bbs_frequencies_int,
3570 cale_bbs_frequencies_gcov_type): Remove.
3571 * cfg.h (scale_bbs_frequencies_int,
3572 cale_bbs_frequencies_gcov_type): Remove.
3574 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3576 * tree-ssa-loop-manip.c
3577 (scale_dominated_blocks_in_loop): Update to profile counts.
3578 (tree_transform_and_unroll_loop): Likewise.
3580 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3582 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
3583 scale_bbs_frequencies_int.
3585 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3587 * final.c (compute_alignments): Use counts rather than frequencies.
3589 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3591 * cfgloopanal.c: Include sreal.h
3592 (average_num_loop_insns): Use counts and sreal for accounting.
3594 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3596 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
3599 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3600 Kito Cheng <kito.cheng@gmail.com>
3602 * config/nds32/constraints.md: Provide more constraints.
3603 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
3604 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
3605 support constraints usage.
3607 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3609 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
3611 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3612 Kito Cheng <kito.cheng@gmail.com>
3614 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
3615 * config/nds32/nds32.opt: Refine the layout.
3616 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
3617 TARGET_EXT_STRING): Support new options.
3618 * config/nds32/nds32.h: Likewise.
3619 * config/nds32/nds32.md: Likewise.
3620 * config/nds32/nds32-predicates.c: Likewise.
3621 * config/nds32/constraints.md: Likewise.
3622 * common/config/nds32/nds32-common.c: Likewise.
3624 2017-11-16 Julia Koval <julia.koval@intel.com>
3627 * config/i386/gfniintrin.h: Add sse check.
3628 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
3630 2017-11-16 Julia Koval <julia.koval@intel.com>
3632 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
3633 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
3634 (ix86_handle_option): Handle -mavx512vbmi2.
3635 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
3636 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3637 * config/i386/i386-c.c (__AVX512VBMI2__): New.
3638 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
3639 (ix86_valid_target_attribute_inner_p): Ditto.
3640 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
3641 * config/i386/i386.opt (mavx512vbmi2): New option.
3642 * doc/invoke.texi: Add new option.
3644 2017-11-16 Julia Koval <julia.koval@intel.com>
3646 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
3647 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
3648 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
3649 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
3650 _mm512_gf2p8mul_epi8): New intrinsics.
3651 * config/i386/i386-builtin-types.def
3652 (V64QI_FTYPE_V64QI_V64QI): New type.
3653 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
3654 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
3655 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
3656 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
3657 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
3658 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3660 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
3662 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
3663 explicitly as a stream of bytes.
3665 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3667 * config/rs6000/altivec.h (vec_xst_be): New #define.
3668 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
3669 and externalize from *altivec_vperm_<mode>_internal.
3670 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
3672 (XL_BE_V8HI): Likewise.
3673 (XL_BE_V4SI): Likewise.
3674 (XL_BE_V4SI): Likewise.
3675 (XL_BE_V2DI): Likewise.
3676 (XL_BE_V4SF): Likewise.
3677 (XL_BE_V2DF): Likewise.
3678 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
3679 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
3680 all array entries with these keys: VSX_BUILTIN_VEC_XL,
3681 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
3682 VSX_BUILTIN_VEC_XST_BE.
3683 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
3684 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
3686 (altivec_init_builtins): Replace conditional calls to def_builtin
3687 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
3688 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
3689 with unconditional calls. Remove calls to def_builtin for
3690 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
3691 __builtin_vec_xst_be.
3692 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
3693 to define_expand, and add alternate RTL generation for P8.
3694 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
3695 vsx_ld_elemrev_v8hi.
3696 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
3697 add alternate RTL generation for P8.
3698 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
3699 vsx_ld_elemrev_v16qi.
3700 (vsx_st_elemrev_v8hi): Convert define_insn
3701 to define_expand, and add alternate RTL generation for P8.
3702 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
3703 vsx_st_elemrev_v8hi.
3704 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
3705 add alternate RTL generation for P8.
3706 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
3707 vsx_st_elemrev_v16qi.
3709 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
3712 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
3713 TARGET_AVX512ER check.
3714 (ix86_option_override_internal): Set MASK_VZEROUPPER if
3715 neither -mzeroupper nor -mno-zeroupper is used and
3716 TARGET_EMIT_VZEROUPPER is set.
3717 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
3718 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
3720 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
3722 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3723 folding of vector compares.
3724 (fold_build_vec_cmp): New helper function.
3725 (fold_compare_helper): New helper function.
3726 (builtin_function_type): Add compare builtins to the list of functions
3727 having unsigned arguments. Cosmetic updates to comment indentation.
3728 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
3729 the not+eq combination.
3731 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3733 PR tree-optimization/82726
3734 PR tree-optimization/70754
3735 * tree-predcom.c (order_drefs_by_pos): New function.
3736 (combine_chains): Move code setting has_max_use_after to...
3737 (try_combine_chains): ...here. New parameter. Sort combined chains
3738 according to position information.
3739 (tree_predictive_commoning_loop): Update call to above function.
3740 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
3742 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3744 PR tree-optimization/82726
3746 2017-01-23 Bin Cheng <bin.cheng@arm.com>
3748 PR tree-optimization/70754
3749 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
3750 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
3751 combined stmt before it if not NULL.
3752 (combine_chains): Process refs reversely and compute dominance point
3756 2017-02-23 Bin Cheng <bin.cheng@arm.com>
3758 PR tree-optimization/79663
3759 * tree-predcom.c (combine_chains): Process refs in reverse order
3760 only for ZERO length chains, and add explaining comment.
3762 2017-11-15 Tamar Christina <tamar.christina@arm.com>
3764 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
3765 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
3766 * doc/invoke.texi (ARM Options): Add armv8.3-a.
3768 2017-11-15 Tamar Christina <tamar.christina@arm.com>
3770 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
3772 2017-11-15 Martin Liska <mliska@suse.cz>
3774 * tree-cfg.c (pass_warn_function_return::execute):
3775 Compare warn_return_type for greater than zero.
3777 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
3781 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
3782 to return true on Xeon and not on Xeon Phi.
3783 (ix86_check_avx256_register): Changed to ...
3784 (ix86_check_avx_upper_register): ... this. Add extra check for
3785 VALID_AVX512F_REG_OR_XI_MODE.
3786 (ix86_avx_u128_mode_needed): Changed
3787 ix86_check_avx256_register to ix86_check_avx_upper_register.
3788 (ix86_check_avx256_stores): Changed to ...
3789 (ix86_check_avx_upper_stores): ... this. Changed
3790 ix86_check_avx256_register to ix86_check_avx_upper_register.
3791 (ix86_avx_u128_mode_after): Changed
3792 avx_reg256_found to avx_upper_reg_found. Changed
3793 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
3794 (ix86_avx_u128_mode_entry): Changed
3795 ix86_check_avx256_register to ix86_check_avx_upper_register.
3796 (ix86_avx_u128_mode_exit): Ditto.
3797 * config/i386/i386.h: (host_detect_local_cpu): New define.
3799 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3801 * config/arm/xgene1.md (xgene1): Split into automatons
3802 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
3803 (xgene1_f_load): Adjust reservations and/or types.
3804 (xgene1_f_store): Likewise.
3805 (xgene1_load_pair): Likewise.
3806 (xgene1_store_pair): Likewise.
3807 (xgene1_fp_load1): Likewise.
3808 (xgene1_load1): Likewise.
3809 (xgene1_store1): Likewise.
3810 (xgene1_move): Likewise.
3811 (xgene1_alu): Likewise.
3812 (xgene1_simd): Likewise.
3813 (xgene1_bfm): Likewise.
3814 (xgene1_neon_load1): Likewise.
3815 (xgene1_neon_store1): Likewise.
3816 (xgene1_neon_logic): Likewise.
3817 (xgene1_neon_st1): Likewise.
3818 (xgene1_neon_ld1r): Likewise.
3819 (xgene1_alu_cond): Added.
3820 (xgene1_shift_reg): Likwise.
3821 (xgene1_bfx): Likewise.
3822 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
3824 2017-11-15 Jakub Jelinek <jakub@redhat.com>
3827 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
3829 (can_widen_mult_without_libcall): New function.
3830 (expand_mul_overflow): If only checking unsigned mul overflow,
3831 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
3832 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
3833 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
3834 (expand_DIVMOD): Formatting fix.
3835 * expmed.h (expand_mult): Add NO_LIBCALL argument.
3836 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
3837 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
3839 PR tree-optimization/82977
3840 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
3841 constructed temporary to strlen_to_stridx.put.
3843 2017-11-15 Martin Liska <mliska@suse.cz>
3845 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
3846 * configure: Regenerate.
3848 2017-11-15 Martin Liska <mliska@suse.cz>
3851 * config/sh/sh-mem.cc: Use proper probability for
3854 2017-11-14 Jeff Law <law@redhat.com>
3856 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
3857 the red zone for stack_clash_protection_final_dynamic_probe targets
3858 when the total dynamic stack size is zero bytes.
3860 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
3861 blocks is post order.
3863 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
3865 * dumpfile.h (TDF_COMPARE_DEBUG): New.
3866 * final.c (rest_of_clean_state): Set it for the
3867 -fcompare-debug dump.
3868 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
3869 class when TDF_COMPARE_DEBUG is set.
3871 * dwarf2out.c (gen_producer_string): Discard
3874 2017-11-15 Joseph Myers <joseph@codesourcery.com>
3877 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
3878 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
3879 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
3880 (__TGMATH_REAL_2_3): Remove macros.
3881 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
3882 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
3884 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
3886 (remquo): Define using __TGMATH_REAL_3.
3888 2017-11-14 Jeff Law <law@redhat.com>
3890 * vr-values.c: New file with contents extracted from tree-vrp.c.
3891 * Makefile.in (OBJS): Add vr-values.o
3892 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
3893 (set_value_range, set_and_canonicalize_value_range): Likewise.
3894 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
3895 (value_range_constant_singleton, symbolic_range_p): Likewise.
3896 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
3897 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
3898 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
3899 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
3900 (find_case_label_range, find_case_label_index): Likewise.
3901 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
3902 (range_int_cst_singleton_p, value_inside_range): Likewise.
3903 (get_single_symbol): Likewise.
3904 (switch_update): Move structure definition here.
3905 (to_remove_edges, to_update_switch_stmts): Provide externs.
3906 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
3907 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
3908 (vrp_val_is_min, set_value_range): Likewise.
3909 (set_and_canonicalize_value_range, copy_value_range): Likewise.
3910 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
3911 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
3912 (range_is_nonnull, range_int_cst_p): Likewwise.
3913 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
3914 (get_single_symbol, operand_less_p): Likewise
3915 (compare_values_warnv, compare_values): Likewise.
3916 (value_inside_range, value_range_constant_singleton): Likewise.
3917 (zero_nonzero_bitgs_from_vr): Likewise.
3918 (extract_range_from_binary_expr_1): Likewise.
3919 (overflow_comparison_p): Likewise.
3920 (to_remove_edges, to_update_switch_stmts): Likewise.
3921 (find_case_label-index, find_case_label_range): Likewise.
3922 (switch_update, set_value_range_to_nonnegative): Remove.
3923 (set_value_range_to_truthvalue): Likewise.
3924 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
3925 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
3926 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
3927 (find_case_label_ranges, test_for_singularity): Likewise.
3928 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
3929 (x_vr_values): Move to its remaining use site.
3931 2017-11-10 Jeff Law <law@redhat.com>
3933 * vr-values.h (VR_INITIALIZER): Move #define here.
3934 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
3935 * Makefile.in (OBJS): Add tree-evrp.o
3936 * tree-vrp.h (assert_info): Move structure definition here.
3937 (set_value_range_to_varying): Prototype.
3938 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3939 (infer_value_range, register_edge_assert_for): Likewise.
3940 (stmt_interesting_for_vrp): Likewise.
3941 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
3942 (set_value_range_to_varying): No longer static.
3943 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3944 (infer_value_range, register_edge_assert_for): Likewise.
3946 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
3948 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
3949 generate the XXBRD instruction.
3951 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
3952 (rs6000_builtin_type_compatible): Treat _Float128 and long double
3953 as being compatible if -mabi=ieeelongdouble.
3954 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
3955 to setup float128 built-ins with hardware support.
3956 (BU_FLOAT128_HW_2): Likewise.
3957 (BU_FLOAT128_HW_3): Likewise.
3958 (BU_FLOAT128_HW_VSX_1): Likewise.
3959 (BU_FLOAT128_HW_VSX_2): Likewise.
3960 (scalar_extract_expq): Change float128 built-in functions to
3961 accommodate having both KFmode and TFmode functions. Use the
3962 KFmode variant as the default.
3963 (scalar_extract_sigq): Likewise.
3964 (scalar_test_neg_qp): Likewise.
3965 (scalar_insert_exp_q): Likewise.
3966 (scalar_insert_exp_qp): Likewise.
3967 (scalar_test_data_class_qp): Likewise.
3968 (sqrtf128_round_to_odd): Delete processing the round to odd
3969 built-in functions as special built-in functions, and define them
3970 as float128 built-ins. Use the KFmode variant as the default.
3971 (truncf128_round_to_odd): Likewise.
3972 (addf128_round_to_odd): Likewise.
3973 (subf128_round_to_odd): Likewise.
3974 (mulf128_round_to_odd): Likewise.
3975 (divf128_round_to_odd): Likewise.
3976 (fmaf128_round_to_odd): Likewise.
3977 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
3978 support for KFmode and TFmode xststdcqp calls.
3979 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
3980 point, switch the built-in handlers for the get/set float128
3981 exponent, get float128 mantissa, float128 test built-ins, and the
3982 float128 round to odd built-in functions. Eliminate creating the
3983 float128 round to odd built-in functions as special built-ins.
3984 (rs6000_init_builtins): Eliminate special creation of the float128
3985 round to odd built-in functions.
3986 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
3987 function insns to support both TFmode and KFmode variants.
3988 (xsxsigqp_<mode>): Likewise.
3989 (xsiexpqpf_<mode>): Likewise.
3990 (xsiexpqp_<mode>): Likewise.
3991 (xststdcqp_<mode>): Likewise.
3992 (xststdcnegqp_<mode>): Likewise.
3993 (xststdcqp_<mode>): Likewise.
3995 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3997 * tree-ssa-threadupdate.c (compute_path_counts): Remove
3998 unused path_in_freq_ptr parameter.
3999 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
4001 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4003 * ipa-inline.c (edge_badness): Dump sreal frequency.
4004 (compute_inlined_call_time): Match natural implementaiton ...
4005 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
4006 forgotten division by CGRAPH_FREQ_BASE.
4008 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4010 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
4011 Solaris 11. Update comment.
4012 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
4014 * config/sol2.h (STARTFILE_SPEC): Likewise.
4015 * configure: Regenerate.
4017 2017-11-14 Carl Love <cel@us.ibm.com>
4019 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
4020 le_ and be_ prefixes to swap* variables. Remove
4021 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
4024 2017-11-14 Jason Merrill <jason@redhat.com>
4026 Support GTY((cache)) on hash_map.
4027 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
4028 (ggc_cache_remove): Override it instead of ggc_mx.
4029 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
4030 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
4031 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
4032 (simple_cache_map_traits): Override maybe_mx.
4033 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
4034 (hash_map): Friend gt_cleare_cache.
4035 (gt_cleare_cache): New.
4036 * tree.h (tree_cache_traits): New hash_map traits class.
4037 (tree_cache_map): New typedef.
4039 2017-11-14 Richard Biener <rguenther@suse.de>
4041 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
4042 paramter and handling.
4043 (cleanup_control_flow_bb): Likewise.
4044 (cleanup_control_flow_pre): New helper performing a DFS walk
4045 to call cleanup_control_flow_bb in PRE order.
4046 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
4047 via cleanup_control_flow_pre.
4049 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
4051 * config/aarch64/aarch64-simd.md
4052 (aarch64_simd_bsl<mode>_internal): Remove DImode.
4053 (*aarch64_simd_bsl<mode>_alt): Likewise.
4054 (aarch64_simd_bsldi_internal): New.
4055 (aarch64_simd_bsldi_alt): Likewise.
4057 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4059 * tracer.c (better_p): Do not compare frequencies.
4060 * reg-stack.c (better_edge): Likewise.
4061 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
4064 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4066 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
4067 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
4068 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
4069 * lto-streamer-in.c (input_function): Use update_max_bb_count.
4070 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
4071 * predict.c (maybe_hot_frequency_p): Inline to ...
4072 (maybe_hot_count_p): ... here; rewrite to counts.
4073 (counts_to_freqs): Rename to ...
4074 (update_max_bb_count): ... this one.
4075 (expensive_function_p): Use counts.
4076 (estimate_bb_frequencies): Update.
4077 (rebuild_frequencies): Update.
4078 * predict.h (counts_to_freqs): Rename to ...
4079 (update_max_bb_count): ... this one.
4080 * profile.c (compute_branch_probabilities): Add debug info
4081 * tree-inline.c (expand_call_inline): Update debug info.
4082 (optimize_inline_calls): Use update_max_bb_count..
4083 (tree_function_versioning): Use update_max_bb_count..
4084 * value-prof.c (gimple_value_profile_transformations):
4085 Do not use update_max_bb_count.
4087 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4089 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
4090 always use frequencies.
4092 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4094 * bb-reorder.c: Remove frequencies from comments.
4095 (better_edge_p): Use profile counts.
4096 (find_traces): Dump profile counts.
4097 (rotate_loop): Use profile counts.
4098 (find_traces_1_round): Likewise.
4099 (connect_better_edge_p): Use counts instead of probabilities for
4101 (copy_bb_p): Drop early check for non-0 frequency.
4102 (sanitize_hot_paths): Update comments.
4104 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4106 * ipa-split.c (struct split_point): Add count.
4107 (consider_split): Do not compute incoming frequency; compute incoming
4108 count and store it to split_point.
4109 (split_function): Set count of the call to split part correctly.
4111 2017-11-13 Carl Love <cel@us.ibm.com>
4113 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
4115 2017-11-13 Tom Tromey <tom@tromey.com>
4117 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
4119 2017-11-13 Carl Love <cel@us.ibm.com>
4121 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
4122 Add support for builtins:
4123 unsigned int vec_first_{,miss}_match_{,or_eos}index,
4124 vector {un,}signed {char,int,short},
4125 vector {un,}signed {char,int,short}) arguments.
4126 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
4127 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
4128 Add BU_P9V_AV_2 expansions for the builtins.
4129 * config/rs6000/altivec.h (vec_first_match_index,
4130 vec_first_mismatch_index, vec_first_match_or_eos_index,
4131 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
4132 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
4133 new extern declaration.
4134 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
4135 * config/rs6000/vsx.md (first_match_index_<mode>,
4136 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
4137 first_mismatch_or_eos_index_<mode>): Add define expand.
4138 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
4139 * doc/extend.texi: Update the built-in documenation file for the new
4142 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
4144 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
4145 into the min/max operations for _Float<N> and _Float<N>X types.
4147 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4150 * dwarf2out.c (do_eh_frame): New static variable.
4151 (dwarf2out_begin_prologue): Set it.
4152 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
4154 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4156 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
4158 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
4160 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4163 * debug.h (dwarf2out_do_eh_frame): Declare.
4164 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
4165 (dwarf2out_do_frame): Use it.
4166 (dwarf2out_do_cfi_asm): Likewise.
4167 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
4168 (dwarf2out_assembly_start): Likewise.
4169 (dwarf2out_begin_prologue): Fix comment.
4170 * toplev.c (compile_file): Always call dwarf2out_frame_finish
4171 if the target needs either debug or unwind DWARF2 info.
4172 * lto-opts.c (lto_write_options): Do not save -fexceptions,
4173 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
4174 -ftrapping-math, -ftrapv and -fwrapv.
4176 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4178 * cgraph.c (cgraph_edge::sreal_frequency): New function.
4179 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
4180 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
4181 (estimate_edge_size_and_time): Likewise.
4182 (ipa_merge_fn_summary_after_inlining): Likewise.
4183 * ipa-inline.c (cgraph_freq_base_rec): Remove.
4184 (compute_uninlined_call_time): Use sreal_frequency.
4185 (compute_inlined_call_time): Likewise.
4186 (ipa_inline): Do not initialize cgraph_freq_base_rec.
4187 * profile-count.c: Include sreal.h.
4188 (profile_count::to_sreal_scale): New.
4189 * profile-count.h: Forward declare sreal.
4190 (profile_count::to_sreal_scale): Declare.
4192 2017-11-13 Nathan Sidwell <nathan@acm.org>
4194 * diagnostic.c (maybe_line_and_column): New.
4195 (diagnostic_get_location_text): Use it.
4196 (diagnostic_report_current_module): Likewise.
4197 (test_diagnostic_get_location_text): Add tests.
4199 2017-11-13 Luis Machado <luis.machado@linaro.org>
4201 * doc/md.texi (Specifying processor pipeline description): Fix
4202 incorrect latency for the div instruction example.
4204 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4206 PR tree-optimization/78821
4207 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
4208 that bit_not_p is the same.
4209 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
4210 (split_group): Count precisely bit_not_p bits in each statement.
4211 (invert_op): New function.
4212 (imm_store_chain_info::output_merged_store): Use invert_op to
4213 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
4214 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
4216 2017-11-13 Martin Liska <mliska@suse.cz>
4218 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
4219 (struct source_info): Likewise.
4220 (add_branch_counts): Likewise.
4221 (add_line_counts): Likewise.
4222 (function_summary): Likewise.
4223 (output_intermediate_line): Likewise.
4224 (generate_results): Likewise.
4226 2017-11-13 Martin Liska <mliska@suse.cz>
4228 * gcov.c (struct block_info): Remove typedef for block_t.
4229 (struct line_info): Likewise.
4230 (line_info::has_block): Likewise.
4231 (EXIT_BLOCK): Likewise.
4232 (unblock): Likewise.
4233 (circuit): Likewise.
4234 (get_cycles_count): Likewise.
4235 (process_file): Likewise.
4236 (read_graph_file): Likewise.
4237 (solve_flow_graph): Likewise.
4238 (find_exception_blocks): Likewise.
4239 (add_line_counts): Likewise.
4240 (accumulate_line_info): Likewise.
4241 (output_line_details): Likewise.
4243 2017-11-13 Martin Liska <mliska@suse.cz>
4245 * gcov.c (struct arc_info): Remove typedef for arc_t.
4246 (struct line_info): Likewise.
4247 (add_branch_counts): Likewise.
4248 (output_branch_count): Likewise.
4249 (function_info::~function_info): Likewise.
4250 (circuit): Likewise.
4251 (output_intermediate_line): Likewise.
4252 (read_graph_file): Likewise.
4253 (solve_flow_graph): Likewise.
4254 (find_exception_blocks): Likewise.
4255 (add_line_counts): Likewise.
4256 (accumulate_line_info): Likewise.
4257 (output_line_details): Likewise.
4258 (output_function_details): Likewise.
4260 2017-11-13 Martin Liska <mliska@suse.cz>
4262 * gcov.c (struct function_info): Remove typedef for function_t.
4263 (struct source_info): Likewise.
4264 (source_info::get_functions_at_location): Likewise.
4265 (solve_flow_graph): Likewise.
4266 (find_exception_blocks): Likewise.
4267 (add_line_counts): Likewise.
4268 (output_intermediate_file): Likewise.
4269 (process_file): Likewise.
4270 (generate_results): Likewise.
4271 (release_structures): Likewise.
4272 (read_graph_file): Likewise.
4273 (read_count_file): Likewise.
4274 (accumulate_line_counts): Likewise.
4275 (output_lines): Likewise.
4277 2017-11-13 Martin Liska <mliska@suse.cz>
4279 * gcov.c (function_info::function_info): Remove num_counts
4280 and add vector<gcov_type>.
4281 (function_info::~function_info): Use the vector.
4282 (process_file): Likewise.
4283 (read_graph_file): Likewise.
4284 (read_count_file): Likewise.
4285 (solve_flow_graph): Likewise.
4287 2017-11-13 Martin Liska <mliska@suse.cz>
4289 * gcov.c (function_info::is_artificial): New function.
4290 (process_file): Erase all artificial early.
4291 (generate_results): Skip as all artificial are already
4294 2017-11-13 Martin Liska <mliska@suse.cz>
4296 * gcov.c (read_graph_file): Store to global vector of functions.
4297 (read_count_file): Iterate the vector.
4298 (process_file): Likewise.
4299 (generate_results): Likewise.
4300 (release_structures): Likewise.
4302 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4304 PR tree-optimization/82954
4305 * gimple-ssa-store-merging.c
4306 (imm_store_chain_info::coalesce_immediate_stores): If
4307 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
4309 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
4311 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
4312 Upddate call to ENDIAN_LANE_N.
4313 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
4314 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
4315 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
4316 and use aarch64_endian_lane_rtx.
4317 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
4319 2017-11-12 Tom de Vries <tom@codesourcery.com>
4321 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
4323 2017-11-12 Tom de Vries <tom@codesourcery.com>
4325 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
4327 2017-11-12 Tom de Vries <tom@codesourcery.com>
4329 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
4331 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4332 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4333 * defaults.h (ASM_OUTPUT_LABELREF): Same.
4335 2017-11-11 Martin Sebor <msebor@redhat.com>
4338 * doc/extend.texi (attribute nonstring): Remove spurious argument.
4341 * prefic.c (translate_name): Replace strncpy with memcpy to
4342 avoid -Wstringop-truncation.
4344 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4346 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
4348 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4350 * predict.c (maybe_hot_frequency_p): Do not use cfun.
4352 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4354 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
4357 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
4359 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
4360 enable generating XXBRH if the value is in a vector register.
4361 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
4362 value is in a vector register.
4363 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
4364 register to register bswap64's instead of doing the GPR sequence
4365 used on previous machines.
4366 (bswapdi2_xxbrd): New insn.
4367 (bswapdi2_reg): Disallow on ISA 3.0.
4368 (register to register bswap64 splitter): Do not split the insn on
4369 ISA 3.0 systems that use XXBRD.
4371 2017-11-10 Martin Sebor <msebor@redhat.com>
4374 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
4376 (find_subframework_file): Same.
4378 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4380 * auto-profile.c (afdo_indirect_call): Drop frequency.
4381 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
4382 (cgraph_node::create_edge): Drop frequency argument.
4383 (cgraph_node::create_indirect_edge): Drop frequency argument.
4384 (cgraph_edge::make_speculative): Drop frequency arguments.
4385 (cgraph_edge::resolve_speculation): Do not update frequencies
4386 (cgraph_edge::dump_edge_flags): Do not dump frequency.
4387 (cgraph_node::dump): Check consistency in IPA mode.
4388 (cgraph_edge::maybe_hot_p): Use IPA counter.
4389 (cgraph_edge::verify_count_and_frequency): Rename to ...
4390 (cgraph_edge::verify_count): ... this one; drop frequency checking.
4391 (cgraph_node::verify_node): Update.
4392 * cgraph.h (struct cgraph_edge): Drop frequency.
4393 (cgraph_edge::frequency): New function.
4394 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
4396 (cgraph_edge::rebuild_edges): Likewise.
4397 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
4398 (duplicate_thunk_for_node): Do not pass frequency.
4399 (cgraph_node::create_clone): Scale only counts.
4400 (cgraph_node::create_virtual_clone): Do not pass frequency.
4401 (cgraph_node::create_edge_including_clones): Do not pass frequency.
4402 (cgraph_node::create_version_clone): Do not pass frequency.
4403 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
4404 (cgraph_node::expand_thunk): Do not pass frequency.
4405 (cgraph_node::create_wrapper): Do not pass frequency.
4406 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
4408 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
4409 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
4410 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
4411 (gather_caller_stats): Use frequency function.
4412 (ipcp_cloning_candidate_p): Use frequency function.
4413 (ipcp_propagate_stage): Use frequency function.
4414 (get_info_about_necessary_edges): Use frequency function.
4415 (update_profiling_info): Update only IPA profile.
4416 (update_specialized_profile): Use frequency functoin.
4417 (perhaps_add_new_callers): Update only IPA profile.
4418 * ipa-devirt.c (ipa_devirt): Use IPA profile.
4419 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
4420 (dump_ipa_call_summary): Use frequency function.
4421 (estimate_edge_size_and_time): Use frequency function.
4422 (ipa_merge_fn_summary_after_inlining): Use frequency function.
4423 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
4424 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
4425 (update_noncloned_counts): ... ths one; scale counts only.
4426 (clone_inlined_nodes): Do not scale frequency.
4427 (inline_call): Do not pass frequency.
4428 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
4429 (compute_inlined_call_time): Use IPA profile.
4430 (want_inline_small_function_p): Use IPA profile.
4431 (want_inline_self_recursive_call_p): Use IPA profile.
4432 (edge_badness): Use IPA profile.
4433 (lookup_recursive_calls): Use IPA profile.
4434 (recursive_inlining): Do not pass frequency.
4435 (resolve_noninline_speculation): Do not update frequency.
4436 (inline_small_functions): Collect max of IPA profile.
4437 (dump_overall_stats): Dump IPA porfile.
4438 (dump_inline_stats): Dump IPA porfile.
4439 (ipa_inline): Collect IPA stats.
4440 * ipa-inline.h (clone_inlined_nodes): Update prototype.
4441 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
4442 (ipa_propagate_frequency): Use frequency function.
4443 (ipa_profile): Cleanup.
4444 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
4445 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
4446 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
4447 (input_node): Do not stream frequency.
4448 (input_edge): Do not stream frequency.
4449 (merge_profile_summaries): Scale only IPA profiles.
4450 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
4451 * predict.c (drop_profile): Do not recompute frequency.
4452 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
4453 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
4454 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
4455 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
4456 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
4457 * tree-inline.c (copy_bb): Do not scale frequency.
4458 (expand_call_inline): Do not scale frequency.
4459 (tree_function_versioning): Do not scale frequency.
4460 * ubsan.c (ubsan_create_edge): Do not pass frequency.
4462 2017-11-10 Julia Koval <julia.koval@intel.com>
4464 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
4465 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
4466 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
4467 (_mm256_maskz_gf2p8affine_epi64_epi8)
4468 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
4469 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
4470 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
4471 (__builtin_ia32_vgf2p8affineqb_v32qi)
4472 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
4473 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
4475 2017-11-10 Tamar Christina <tamar.christina@arm.com>
4479 (arm_option_override): Refactor.
4480 (arm_option_reconfigure_globals): New.
4481 (arm_options_perform_arch_sanity_checks): New.
4482 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
4484 (arm_options_perform_arch_sanity_checks): Likewise
4486 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
4488 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
4489 (power9-qpmul): New.
4490 * rs6000/rs6000.md ("type" attr): Add qmul.
4491 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
4492 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
4493 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
4495 2017-11-10 Martin Sebor <msebor@redhat.com>
4498 * builtins.c (compute_objsize): Handle arrays that
4499 compute_builtin_object_size likes to fail for. Make extern.
4500 * builtins.h (compute_objsize): Declare.
4501 (check_strncpy_sizes): New function.
4502 (expand_builtin_strncpy): Call check_strncpy_sizes.
4503 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
4504 -Wstringop-truncation.
4505 (gimple_fold_builtin_strncat): Same.
4506 * gimple.c (gimple_build_call_from_tree): Set call location.
4507 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
4508 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
4509 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
4510 (handle_builtin_strlen): Use strlen_to_stridx.
4511 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
4513 Use strlen_to_stridx.
4514 (pass_strlen::execute): Release strlen_to_stridx.
4515 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
4516 (-Wstringop-truncation): Document new option.
4518 2017-11-10 Martin Liska <mliska@suse.cz>
4520 PR gcov-profile/82702
4521 * gcov.c (main): Handle intermediate files in a different
4523 (get_gcov_intermediate_filename): New function.
4524 (output_gcov_file): Remove support of intermediate files.
4525 (generate_results): Allocate intermediate file.
4526 (release_structures): Clean-up properly fn_end.
4527 (output_intermediate_file): Start iterating with line 1.
4529 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4531 PR tree-optimization/82929
4532 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
4533 ops_swapped_p non-static data member.
4534 (store_immediate_info::store_immediate_info): Clear it.
4535 (imm_store_chain_info::coalesce_immediate_stores): If swapping
4536 ops set ops_swapped_p.
4537 (count_multiple_uses): Handle ops_swapped_p.
4539 2017-11-10 Martin Liska <mliska@suse.cz>
4541 * coverage.c (coverage_init): Stream information about
4542 support of has_unexecuted_blocks.
4543 * doc/gcov.texi: Document that.
4544 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
4545 * gcov.c (read_graph_file): Likewise.
4546 (output_line_beginning): Fix a small issue with
4549 2017-11-10 Bin Cheng <bin.cheng@arm.com>
4551 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
4552 reference of trivial component.
4554 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4557 * gimple-ssa-store-merging.c
4558 (pass_store_merging::terminate_all_aliasing_chains): For
4559 gimple_store_p stmts also call refs_output_dependent_p.
4561 PR rtl-optimization/82913
4562 * compare-elim.c (try_merge_compare): Punt if def_insn is not
4565 2017-11-09 Jeff Law <law@redhat.com>
4567 * vr-values.h: New file with vr_values class.
4568 * tree-vrp.c: Include vr-values.h
4569 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
4570 data objects into the vr_values class.
4571 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
4572 (get_value_range): Make it a member function within vr_values class.
4573 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
4574 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
4575 (op_with_constant_singleton_value_range): Likewise.
4576 (extract_range_for_var_from_comparison_expr): Likewise.
4577 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
4578 (extract_range_from_binary_expr): Likewise.
4579 (extract_range_from_unary_expr): Likewise.
4580 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
4581 (check_for_binary_op_overflow, extract_range_basic): Likewise.
4582 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
4583 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
4584 (compare_name_with_value, compare_names): Likewise.
4585 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
4586 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
4587 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
4588 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
4589 (extract_range_from_phi_node): Likewise.
4590 (simplify_truth_ops_using_ranges): Likewise.
4591 (simplify_div_or_mod_using_ranges): Likewise.
4592 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
4593 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
4594 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
4595 (simplify_float_conversion_using_ranges): Likewise.
4596 (simplify_internal_call_using_ranges): Likewise.
4597 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
4598 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
4599 poitner via x_vr_values for calls into gimple folder.
4600 (vrp_initialize_lattice): Make this the vr_values ctor.
4601 (vrp_free_lattice): Make this the vr_values dtor.
4602 (set_vr_value): New function.
4603 (class vrp_prop): Add vr_values data member. Add various member
4604 functions as well as member functions that delegate to vr_values.
4605 (check_array_ref): Make a member function within vrp_prop class.
4606 (search_for_addr_array, vrp_initialize): Likewise.
4607 (vrp_finalize): Likewise. Revamp to avoid direct access to
4608 vr_value, values_propagated, etc.
4609 (check_array_bounds): Extract vrp_prop class instance pointer from
4610 walk info structure. Use it to call member functions.
4611 (check_all_array_refs): Make a member function within vrp_prop class.
4612 Smuggle class instance pointer via walk info structure.
4613 (x_vr_values): New local static.
4614 (vrp_valueize): Use x_vr_values to get class instance.
4615 (vr_valueize_1): Likewise.
4616 (class vrp_folder): Add vr_values data member. Add various member
4617 functions as well as member functions that delegate to vr_values.
4618 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
4619 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
4620 class instance from vr_values. Use it to call member functions.
4621 (vrp_dom_walker): Add vr_values data member.
4622 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
4623 instance via x_vr_values.
4624 (identify_jump_threads): Accept vr_values as argument. Store
4625 it into the walker structure.
4626 (evrp_dom_walker): Add vr_values class data member. Add various
4628 (evrp_dom_walker::try_find_new_range): Use vr_values data
4629 member to access the memory allocator.
4630 (evrp_dom_walker::before_dom_children): Store vr_values class
4631 instance into the vrp_folder class.
4632 (evrp_dom_walker::push_value_range): Rework to avoid direct
4633 access to num_vr_values and vr_value.
4634 (evrp_dom_walker::pop_value_range): Likewise.
4635 (execute_early_vrp): Remove call to vrp_initialize_lattice.
4636 Use vr_values to get to dump_all_value_ranges member function.
4637 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
4638 and simplify_cond_using_ranges_2 via vrp_prop class instance.
4639 Pass vr_values class instance down to identify_jump_threads.
4640 Remove call to vrp_free_lattice.
4641 (debug_all_value_ranges): Remove.
4643 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
4644 (vrp_folder): Likewise.
4646 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
4647 Get it from the existing bitmap instead.
4648 (vrp_intersect_ranges_1): Likewise.
4650 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4652 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
4654 (store_immediate_info::store_immediate_info): Add bitnotp argument,
4655 set bit_not_p to it.
4656 (imm_store_chain_info::coalesce_immediate_stores): Break group
4657 if bit_not_p is different.
4658 (count_multiple_uses, split_group,
4659 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
4660 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
4661 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
4662 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
4663 to store_immediate_info ctor.
4665 2017-11-09 Jim Wilson <jimw@sifive.com>
4667 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
4668 (scan_prog_file): Likewise.
4670 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4672 * bb-reorder.c (max_entry_frequency): Remove.
4673 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
4674 connect_traces, push_to_next_round_p): Remove prototypes.
4675 (find_traces_1_round): Use counts only.
4676 (push_to_next_round_p): Likewise.
4677 (find_traces): Likewise.
4678 (rotate_loop): Likewise.
4679 (find_traces_1_round): Likewise.
4680 (connect_traces): Likewise.
4681 (edge_order): Likewise.
4683 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
4685 * config/arm/arm.c (output_return_instruction): Add comments to
4686 indicate requirement for cmse_nonsecure_entry return to account
4687 for the size of clearing instruction output here.
4688 (thumb_exit): Likewise.
4689 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
4690 return in hardfloat mode.
4692 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
4694 * config/rs6000/rs6000.c (machine_function): Add a bool,
4695 "toc_is_wrapped_separately".
4696 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
4697 if it wasn't explicitly set or unset, we are optimizing for speed, and
4698 doing separate shrink-wrapping.
4699 (rs6000_get_separate_components): Enable the TOC component if
4700 saving the TOC register in the prologue.
4701 (rs6000_components_for_bb): Handle the TOC component.
4702 (rs6000_emit_prologue_components): Store the TOC register where needed.
4703 (rs6000_set_handled_components): Mark TOC as handled, if handled.
4704 (rs6000_emit_prologue): Don't save the TOC if that is already done.
4706 2017-11-09 Martin Jambor <mjambor@suse.cz>
4708 * ipa-param-manipulation.c: New file.
4709 * ipa-param-manipulation.h: Likewise.
4710 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
4711 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
4712 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
4713 (ipa_parm_adjustment): Likewise.
4714 (ipa_parm_adjustment_vec): Likewise.
4715 (ipa_get_vector_of_formal_parms): Moved declaration to
4716 ipa-param-manipulation.h.
4717 (ipa_get_vector_of_formal_parm_types): Likewise.
4718 (ipa_modify_formal_parameters): Likewise.
4719 (ipa_modify_call_arguments): Likewise.
4720 (ipa_combine_adjustments): Likewise.
4721 (ipa_dump_param_adjustments): Likewise.
4722 (ipa_modify_expr): Likewise.
4723 (ipa_get_adjustment_candidate): Likewise.
4724 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
4725 ipa-param-manipulation.c.
4726 (ipa_get_vector_of_formal_parm_types): Likewise.
4727 (ipa_modify_formal_parameters): Likewise.
4728 (ipa_modify_call_arguments): Likewise.
4729 (ipa_modify_expr): Likewise.
4730 (get_ssa_base_param): Likewise.
4731 (ipa_get_adjustment_candidate): Likewise.
4732 (index_in_adjustments_multiple_times_p): Likewise.
4733 (ipa_combine_adjustments): Likewise.
4734 (ipa_dump_param_adjustments): Likewise.
4735 * tree-sra.c: Also include ipa-param-manipulation.h
4736 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
4739 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4740 Alan Hayward <alan.hayward@arm.com>
4741 David Sherwood <david.sherwood@arm.com>
4743 * doc/sourcebuild.texi (vect_masked_store): Document.
4745 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4746 Alan Hayward <alan.hayward@arm.com>
4747 David Sherwood <david.sherwood@arm.com>
4749 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
4751 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4752 Alan Hayward <alan.hayward@arm.com>
4753 David Sherwood <david.sherwood@arm.com>
4755 * doc/sourcebuild.texi (vect_variable_length): Document.
4757 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4758 Alan Hayward <alan.hayward@arm.com>
4759 David Sherwood <david.sherwood@arm.com>
4761 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
4763 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4764 Alan Hayward <alan.hayward@arm.com>
4765 David Sherwood <david.sherwood@arm.com>
4767 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
4769 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4770 Alan Hayward <alan.hayward@arm.com>
4771 David Sherwood <david.sherwood@arm.com>
4773 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
4774 previously undocumented selectors.
4775 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
4777 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4779 * doc/rtl.texi (const_vector): Say that elements can be
4780 const_wide_ints too.
4781 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
4782 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
4783 (gen_vec_duplicate): Use it instead of CONSTANT_P.
4784 * optabs.c (expand_vector_broadcast): Likewise.
4786 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4787 Alan Hayward <alan.hayward@arm.com>
4788 David Sherwood <david.sherwood@arm.com>
4790 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
4791 scaled index even if the unscaled address was invalid.
4792 Don't increase the complexity of using a scale in that case.
4794 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4795 Alan Hayward <alan.hayward@arm.com>
4796 David Sherwood <david.sherwood@arm.com>
4798 * doc/rtl.texi: Rewrite the subreg rules so that they partition
4799 the inner register into REGMODE_NATURAL_SIZE bytes rather than
4800 UNITS_PER_WORD bytes.
4801 * emit-rtl.c (validate_subreg): Divide subregs into blocks
4802 based on REGMODE_NATURAL_SIZE of the inner mode.
4803 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
4804 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
4805 * expmed.c (lowpart_bit_field_p): Divide the value up into
4806 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
4807 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
4808 whether something is likely to occupy more than one register.
4810 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4813 * ipa-inline-transform.c (update_noncloned_frequencies): Use
4814 profile_count::adjust_for_ipa_scaling.
4815 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
4816 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
4818 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
4820 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4822 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
4823 (split_group): Add total_orig and total_new arguments, estimate the
4824 number of statements related to the store group without store merging
4825 and with store merging.
4826 (imm_store_chain_info::output_merged_store): Adjust split_group
4827 callers, punt if estimated number of statements with store merging
4828 is not smaller than estimated number of statements without it.
4830 (handled_load): Remove has_single_use checks.
4831 (pass_store_merging::process_store): Likewise.
4833 2017-11-09 Richard Biener <rguenther@suse.de>
4835 PR tree-optimization/82902
4836 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
4838 2017-11-09 Martin Liska <mliska@suse.cz>
4841 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
4844 2017-11-09 Martin Liska <mliska@suse.cz>
4846 PR tree-optimization/82669
4847 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
4849 2017-11-09 Martin Liska <mliska@suse.cz>
4851 PR gcov-profile/48463
4852 * coverage.c (coverage_begin_function): Output also end locus
4853 of a function and information whether the function is
4855 * gcov-dump.c (tag_function): Parse and print the information.
4856 * gcov.c (INCLUDE_MAP): Add include.
4857 (INCLUDE_SET): Likewise.
4858 (struct line_info): Move earlier in the source file because
4859 of vector<line_info> in function_info structure.
4860 (line_info::line_info): Likewise.
4861 (line_info::has_block): Likewise.
4862 (struct source_info): Add new member index.
4863 (source_info::get_functions_at_location): New function.
4864 (function_info::group_line_p): New function.
4865 (output_intermediate_line): New function.
4866 (output_intermediate_file): Use the mentioned function.
4867 (struct function_start): New.
4868 (struct function_start_pair_hash): Likewise.
4869 (process_file): Add code that identifies group functions.
4870 Assign lines either to global or function scope.
4871 (generate_results): Skip artificial functions.
4872 (find_source): Assign index for each source file.
4873 (read_graph_file): Read new flag artificial and end_line.
4874 (add_line_counts): Assign it either to global of function scope.
4875 (accumulate_line_counts): Isolate core of the function to
4876 accumulate_line_info and call it for both function and global
4878 (accumulate_line_info): New function.
4879 (output_line_beginning): Fix GNU coding style.
4880 (print_source_line): New function.
4881 (output_line_details): Likewise.
4882 (output_function_details): Likewise.
4883 (output_lines): Iterate both source (global) scope and function
4885 (struct function_line_start_cmp): New class.
4886 * doc/gcov.texi: Reflect changes in documentation.
4888 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4891 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
4892 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
4893 and similarly for not instead of neg.
4895 2017-11-08 Andi Kleen <ak@linux.intel.com>
4897 * config/i386/i386.opt: Add -mforce-indirect-call.
4898 * config/i386/predicates.md: Check for flag_force_indirect_call.
4899 * doc/invoke.texi: Document -mforce-indirect-call
4901 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
4903 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
4905 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
4906 config/riscv/riscv.c (predict.h): New include.
4907 (riscv_slow_unaligned_access_p): No longer static.
4908 (riscv_block_move_straight): Add require.
4909 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
4911 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4914 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
4915 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
4916 nonimmediate_operand predicate for operand 1 instead of
4919 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4921 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
4923 * config/aarch64/constraints.md (Uml): New constraint.
4924 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
4927 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4929 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
4930 of two vec_duplicates into a vec_concat.
4932 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4934 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4935 Simplify vec_merge of vec_duplicate and vec_concat.
4936 * config/aarch64/constraints.md (Utq): New constraint.
4937 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
4940 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4942 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4943 Simplify vec_merge of vec_duplicate and const_vector.
4944 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
4946 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
4947 mode iterator. Update predicate on operand 1 to
4948 handle non-const_vec constants. Delete constraints.
4949 (*aarch64_combinez_be<mode>): Likewise for operand 2.
4951 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4953 PR tree-optimization/78821
4954 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
4956 (store_operand_info::store_operand_info): Initialize it to false.
4957 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
4958 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
4959 store in the group, and if chain_info is non-NULL, to ignore altogether
4961 (compatible_load_p): Fail if bit_not_p does not match.
4962 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
4963 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
4964 (pass_store_merging::process_store): Adjust
4965 terminate_all_aliasing_chains calls to pass NULL in all current spots,
4966 call terminate_all_aliasing_chains newly when adding a store into
4967 a chain with non-NULL chain_info.
4969 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
4971 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
4973 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
4975 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4977 (aarch64_layout_frame): Initialise emit_frame_chain.
4978 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
4979 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
4981 2017-11-08 Martin Liska <mliska@suse.cz>
4983 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
4984 of gimple_call_internal_p.
4986 2017-11-07 Tom de Vries <tom@codesourcery.com>
4988 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
4990 2017-11-07 Tom de Vries <tom@codesourcery.com>
4992 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
4995 2017-11-08 Martin Liska <mliska@suse.cz>
4998 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
5000 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
5002 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
5004 2017-11-07 Jakub Jelinek <jakub@redhat.com>
5007 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
5008 (*cmp<mode>_ccz_1): New insn with $k alternative.
5011 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
5012 RTX_COMM_COMPARE as commutative as well.
5013 (ix86_binary_operator_ok): Formatting fix.
5014 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
5015 *<code><mode>3<mask_name><round_saeonly_name>,
5016 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
5017 *<s>mul<mode>3_highpart<mask_name>,
5018 *vec_widen_umult_even_v16si<mask_name>,
5019 *vec_widen_umult_even_v8si<mask_name>,
5020 *vec_widen_umult_even_v4si<mask_name>,
5021 *vec_widen_smult_even_v16si<mask_name>,
5022 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
5023 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
5024 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
5025 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
5026 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
5027 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
5028 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
5029 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
5030 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
5031 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
5032 ix86_binary_operator_ok. Formatting fixes.
5033 (*<plusminus_insn><mode>3<mask_name><round_name>,
5034 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
5037 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
5039 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
5040 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
5041 eq and ne if TARGET_ISEL.
5042 (cmp): New code_iterator.
5043 (UNS, UNSU_, UNSIK): New code_attrs.
5044 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
5045 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
5047 ("eq<mode>3"): ... this.
5048 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
5050 ("ne<mode>3"): ... this.
5052 2017-11-07 Julia Koval <julia.koval@intel.com>
5055 * common/config/i386/i386-common.c
5056 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
5057 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
5058 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
5059 * config/i386/i386.opt: Ditto.
5060 * config/i386/i386.c (ix86_target_string): Ditto.
5061 (ix86_option_override_internal): Ditto.
5062 (ix86_init_mpx_builtins): Move MPX to args2.
5063 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
5064 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
5065 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
5066 __builtin_ia32_vgf2p8affineinvqb_v32qi,
5067 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
5068 __builtin_ia32_vgf2p8affineinvqb_v16qi,
5069 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
5071 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
5074 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
5075 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
5076 (zero-extendsidi peephole2): Remove peephole.
5078 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
5081 * stor-layout.c: Include attribs.h.
5082 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
5083 explicit lookup of "aligned" attribute.
5085 2017-11-07 Andrew Waterman <andrew@sifive.com>
5087 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
5088 (riscv_expand_block_move): Likewise.
5089 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
5091 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
5092 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
5093 * config/riscv/riscv.c (riscv_block_move_straight): New function.
5094 (riscv_adjust_block_mem): Likewise.
5095 (riscv_block_move_loop): Likewise.
5096 (riscv_expand_block_move): Likewise.
5097 * config/riscv/riscv.md (movmemsi): New pattern.
5099 2017-11-07 Michael Clark <michaeljclark@mac.com>
5101 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
5102 (MUSL_DYNAMIC_LINKER): Likewise.
5104 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
5106 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
5107 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
5110 2017-11-07 Richard Biener <rguenther@suse.de>
5112 * match.pd: Fix build.
5114 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
5115 Jackson Woodruff <jackson.woodruff@arm.com>
5117 PR tree-optimization/71026
5118 * match.pd: Canonicalize negate in division.
5120 2017-11-07 Sudakshina Das <sudi.das@arm.com>
5123 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
5125 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5127 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
5128 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
5130 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5132 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
5133 non-scalar integral types.
5134 * match.pd (negate_expr_p): Handle MINUS_EXPR.
5135 (-(A-B), -(~A)): New transformations.
5137 2017-11-07 Tom de Vries <tom@codesourcery.com>
5139 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
5140 semicolon after "do {} while (0)".
5141 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5142 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5143 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5144 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5145 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5147 2017-11-07 Tom de Vries <tom@codesourcery.com>
5149 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
5150 after "do {} while (0)".
5151 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5152 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5153 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5154 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5155 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5157 2017-11-07 Tom de Vries <tom@codesourcery.com>
5160 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
5162 (arm_rtx_costs_internal): Add missing semicolon after
5163 HANDLE_NARROW_SHIFT_ARITH call.
5165 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5167 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
5168 disable isel if it was not set explicitly.
5170 2017-11-06 James Bowman <james.bowman@ftdichip.com>
5172 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
5173 (add_type_attribute) likewise.
5175 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
5177 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
5178 of struct ix86_frame.
5179 (ix86_initial_elimination_offset): Likewise.
5180 (ix86_expand_split_stack_prologue): Likewise.
5182 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
5184 * tree-vrp.h (enum value_range_type): Update stale comment.
5186 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5187 Alan Hayward <alan.hayward@arm.com>
5188 David Sherwood <david.sherwood@arm.com>
5190 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
5191 (aarch64_expand_vec_perm_const): Take the number of units too.
5192 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
5193 (aarch64_expand_vec_perm_const): Likewise.
5194 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
5195 (vec_perm<mode>): Update accordingly.
5197 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5198 Alan Hayward <alan.hayward@arm.com>
5199 David Sherwood <david.sherwood@arm.com>
5201 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
5202 Take the number of units too.
5203 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
5204 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
5205 but check for a vector mode before rather than after the call.
5206 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
5207 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
5208 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
5209 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
5210 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
5211 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
5212 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
5213 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
5214 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
5215 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
5216 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
5217 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
5218 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
5219 (aarch64_sqdmull2_n<mode>): Update accordingly.
5221 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5222 Alan Hayward <alan.hayward@arm.com>
5223 David Sherwood <david.sherwood@arm.com>
5225 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
5226 the number of units too.
5227 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
5228 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
5229 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
5230 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
5231 (vec_store_lanesxi<mode>): Update accordingly.
5233 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5234 Alan Hayward <alan.hayward@arm.com>
5235 David Sherwood <david.sherwood@arm.com>
5237 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
5238 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
5239 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
5240 of units rather than the mode.
5241 * config/aarch64/iterators.md (nunits): New mode attribute.
5242 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
5243 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
5244 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
5245 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
5246 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
5247 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
5248 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
5249 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
5250 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
5251 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
5252 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5253 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
5254 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
5255 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
5256 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
5257 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
5258 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
5259 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
5260 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
5261 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
5262 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
5263 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
5264 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
5265 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
5266 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
5267 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
5268 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
5269 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
5270 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
5271 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
5272 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
5273 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
5274 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
5275 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
5276 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
5277 (aarch64_simd_vec_setv2di): Likewise.
5279 2017-11-06 Carl Love <cel@us.ibm.com>
5281 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
5283 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
5284 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
5286 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
5288 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
5289 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
5290 Add power 8 macro expansions.
5291 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
5292 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
5293 power 8 instructions. (VSX_XXBR): Add iterator.
5295 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
5297 * config/arm/arm.md (predicable_short_it): Change default to "no",
5298 improve documentation, remove uses that are identical to the default.
5299 (enabled_for_depr_it): Rename to enabled_for_short_it.
5300 * gcc/config/arm/arm-fixed.md (predicable_short_it):
5301 Remove default uses.
5302 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
5303 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
5304 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
5305 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
5307 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
5310 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
5311 float128 helper macros, which are no longer used after deleting
5312 the old 'q' built-in functions, and moving the round to odd
5313 built-in functions to being special built-in functions.
5314 (BU_FLOAT128_2): Likewise.
5315 (BU_FLOAT128_1_HW): Likewise.
5316 (BU_FLOAT128_2_HW): Likewise.
5317 (BU_FLOAT128_3_HW): Likewise.
5318 (FABSQ): Delete old 'q' built-in functions.
5319 (COPYSIGNQ): Likewise.
5320 (SQRTF128_ODD): Move round to odd built-in functions to be
5321 special built-in functions, so that we can handle
5322 -mabi=ieeelongdouble.
5323 (TRUNCF128_ODD): Likewise.
5324 (ADDF128_ODD): Likewise.
5325 (SUBF128_ODD): Likewise.
5326 (MULF128_ODD): Likewise.
5327 (DIVF128_ODD): Likewise.
5328 (FMAF128_ODD): Likewise.
5329 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
5330 built-in names to 'f128'.
5331 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
5332 old 'q' built-in functions, as the machine independent code for
5333 'f128' built-in functions handles this.
5334 (rs6000_expand_builtin): Add expansion for float128 round to odd
5335 functions, keying off on -mabi=ieeelongdouble of whether to use
5336 the KFmode or TFmode variant.
5337 (rs6000_init_builtins): Initialize the _Float128 round to odd
5339 * doc/extend.texi (PowerPC Built-in Functions): Document the old
5340 _Float128 'q' built-in functions are now mapped into the new
5341 'f128' built-in functions.
5343 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
5345 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
5346 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
5348 2017-11-06 David Malcolm <dmalcolm@redhat.com>
5351 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
5352 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
5353 * toplev.c: Include "ipa-fnsummary.h".
5354 (toplev::finalize): Call ipa_fnsummary_c_finalize.
5356 2017-11-06 Jakub Jelinek <jakub@redhat.com>
5358 PR tree-optimization/82838
5359 * gimple-ssa-store-merging.c
5360 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
5361 on a separate gimple_seq which is then appended to seq.
5363 2017-11-06 Jeff Law <law@redhat.com>
5366 * config/i386/i386.c (PROBE_INTERVAL): Remove.
5367 (get_probe_interval): New functions.
5368 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
5369 (ix86_adjust_stack_and_probe): Likewise.
5370 (output_adjust_stack_and_probe): Likewise.
5371 (ix86_emit_probe_stack_range): Likewise.
5372 (ix86_expand_prologue): Likewise.
5374 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5376 PR tree-optimization/82816
5377 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
5378 if the modes of the two types are the same.
5379 (convert_plusminus_to_widen): Likewise.
5381 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5383 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
5385 (usadv16qi): New define_expand.
5386 (usadv8hi): New define_expand.
5388 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
5391 * ipa-inline-transform.c (update_noncloned_frequencies): Always
5393 (inline_transform): Likewise.
5394 * predict.c (counts_to_freqs): Remove useless conditional.
5395 * profile-count.h (profile_count::apply_scale): Move sanity check.
5396 * tree-inline.c (copy_bb): Always scale.
5397 (copy_cfg_body): Likewise.
5399 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
5402 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
5404 (*cmp_ite0): Add enabled_for_depr_it attribute.
5405 (*cmp_ite1): Likewise.
5407 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5409 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
5412 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5414 * tree-vrp.c (vrp_int_const_binop): Return true on success and
5415 return the value by pointer.
5416 (extract_range_from_multiplicative_op_1): Update accordingly.
5417 Return as soon as an operation fails.
5419 2017-11-05 Tom de Vries <tom@codesourcery.com>
5422 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
5423 (DEF_SANITIZER_BUILTIN): ... here.
5424 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
5425 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
5427 2017-11-05 Tom de Vries <tom@codesourcery.com>
5430 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
5432 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
5433 ASM_OUTPUT_BEFORE_CASE_LABEL call.
5434 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
5436 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5437 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5438 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5440 2017-11-05 Tom de Vries <tom@codesourcery.com>
5443 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
5446 2017-11-04 Michael Clark <michaeljclark@mac.com>
5448 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
5449 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
5451 (*addsi3_extended): Likewise.
5452 (*addsi3_extended2): Likewise.
5453 (<optab>si3): Likewise.
5454 (<optab>di3): Likewise.
5455 (<optab><mode>3): Likewise.
5456 (<*optabe>si3_internal): Likewise.
5457 (zero_extendqi<SUPERQI:mode>2): Likewise.
5458 (*add<mode>hi3): Likewise.
5459 (*xor<mode>hi3): Likewise.
5460 (<optab>di3): Likewise.
5461 (*<optab>si3_extend): Likewise.
5462 (*sge<u>_<X:mode><GPR:mode>): Likewise.
5463 (*slt<u>_<X:mode><GPR:mode>): Likewise.
5464 (*sle<u>_<X:mode><GPR:mode>): Likewise.
5466 2017-11-04 Andrew Waterman <andrew@sifive.com>
5468 * config/riscv/riscv.c (riscv_option_override): Conditionally set
5469 TARGET_STRICT_ALIGN based upon -mtune argument.
5471 2017-11-04 Andrew Waterman <andrew@sifive.com>
5473 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
5475 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
5477 * config/i386/i386.c (choose_basereg): Use optional scratch
5478 register and add assertion.
5479 (x86_emit_outlined_ms2sysv_save): Use scratch register when
5480 needed, and don't allocate stack.
5481 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
5482 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
5483 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
5485 2017-11-03 Jeff Law <law@redhat.com>
5487 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
5488 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
5489 to probe at the start of a noreturn function.
5491 2017-11-03 Jakub Jelinek <jakub@redhat.com>
5493 PR tree-optimization/78821
5494 * gimple-ssa-store-merging.c: Update the file comment.
5495 (MAX_STORE_ALIAS_CHECKS): Define.
5496 (struct store_operand_info): New type.
5497 (store_operand_info::store_operand_info): New constructor.
5498 (struct store_immediate_info): Add rhs_code and ops data members.
5499 (store_immediate_info::store_immediate_info): Add rhscode, op0r
5500 and op1r arguments to the ctor, initialize corresponding data members.
5501 (struct merged_store_group): Add load_align_base and load_align
5503 (merged_store_group::merged_store_group): Initialize them.
5504 (merged_store_group::do_merge): Update them.
5505 (merged_store_group::apply_stores): Pick the constant for
5506 encode_tree_to_bitpos from one of the two operands, or skip
5507 encode_tree_to_bitpos if neither operand is a constant.
5508 (class pass_store_merging): Add process_store method decl. Remove
5509 bool argument from terminate_all_aliasing_chains method decl.
5510 (pass_store_merging::terminate_all_aliasing_chains): Remove
5511 var_offset_p argument and corresponding handling.
5512 (stmts_may_clobber_ref_p): New function.
5513 (compatible_load_p): New function.
5514 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
5515 if there is overlap and rhs_code is not INTEGER_CST. For
5516 non-overlapping stores terminate group if rhs is not mergeable.
5517 (get_alias_type_for_stmts): Change first argument from
5518 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
5519 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
5520 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
5522 (get_location_for_stmts): Change first argument from
5523 auto_vec<gimple *> & to vec<gimple *> &.
5524 (struct split_store): Remove orig_stmts data member, add orig_stores.
5525 (split_store::split_store): Create orig_stores rather than orig_stmts.
5526 (find_constituent_stmts): Renamed to ...
5527 (find_constituent_stores): ... this. Change second argument from
5528 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
5529 to info structures rather than the statements.
5530 (split_group): Rename ALLOW_UNALIGNED argument to
5531 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
5532 it. Adjust find_constituent_stores caller.
5533 (imm_store_chain_info::output_merged_store): Handle rhs_code other
5534 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
5535 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
5536 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
5537 (mem_valid_for_store_merging): New function.
5538 (handled_load): New function.
5539 (pass_store_merging::process_store): New method.
5540 (pass_store_merging::execute): Use process_store method. Adjust
5541 terminate_all_aliasing_chains caller.
5543 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5545 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
5546 Return true for more constants, symbols and label references.
5547 (aarch64_valid_floating_const): Remove unused function.
5549 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5552 * config/aarch64/aarch64.c (aarch64_layout_frame):
5553 Undo forcing of LR at bottom of frame.
5555 2017-11-03 Jeff Law <law@redhat.com>
5558 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
5559 for int_registers_saved.
5561 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
5562 extracted from tree-ssa-dom.c.
5563 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
5564 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
5565 (record_equivalences_from_incoming_edge): Add additional argument
5566 to single_pred_edge_ignoring_loop_edges call.
5567 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
5568 (uncprop_dom_walker::before_dom_children): Add additional argument
5569 to single_pred_edge_ignoring_loop_edges call.
5570 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
5571 single_pred_edge_ignoring_loop_edges rather than open coding.
5572 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
5574 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
5576 * match.pd (-(-A)): Rewrite.
5578 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5580 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
5581 (rs6000_emit_int_cmove): New declaration.
5582 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
5583 (rs6000_emit_sISEL): Delete.
5584 (rs6000_emit_int_cmove): Make non-static.
5585 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
5586 instead of rs6000_emit_sISEL.
5588 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
5590 * asan.c (create_cond_insert_point): Maintain profile.
5591 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
5593 * basic-block.h (struct basic_block_def): Remove frequency.
5594 (EDGE_FREQUENCY): Use to_frequency
5595 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
5597 (find_traces): Update to use to_frequency.
5598 (find_traces_1_round): Likewise; use only IPA counts.
5599 (bb_to_key): Likewise.
5600 (connect_traces): Use IPA counts only.
5601 (copy_bb_p): Update to use to_frequency.
5602 (fix_up_crossing_landing_pad): Likewise.
5603 (sanitize_hot_paths): Likewise.
5604 * bt-load.c (basic_block_freq): Likewise.
5605 * cfg.c (init_flow): Set count_max to uninitialized.
5606 (check_bb_profile): Remove frequencies; check counts.
5607 (dump_bb_info): Do not dump frequencies.
5608 (update_bb_profile_for_threading): Update counts only.
5609 (scale_bbs_frequencies_int): Likewise.
5610 (MAX_SAFE_MULTIPLIER): Remove.
5611 (scale_bbs_frequencies_gcov_type): Update counts only.
5612 (scale_bbs_frequencies_profile_count): Update counts only.
5613 (scale_bbs_frequencies): Update counts only.
5614 * cfg.h (struct control_flow_graph): Add count-max.
5615 (update_bb_profile_for_threading): Update prototype.
5616 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
5617 (find_many_sub_basic_blocks): Likewise.
5618 * cfgcleanup.c (try_forward_edges): Likewise.
5619 (try_crossjump_to_edge): Likewise.
5620 * cfgexpand.c (expand_gimple_cond): Likewise.
5621 (expand_gimple_tailcall): Likewise.
5622 (construct_init_block): Likewise.
5623 (construct_exit_block): Likewise.
5624 * cfghooks.c (verify_flow_info): Check consistency of counts.
5625 (dump_bb_for_graph): Do not dump frequencies.
5626 (split_block_1): Do not update frequencies.
5627 (split_edge): Do not update frequencies.
5628 (make_forwarder_block): Do not update frequencies.
5629 (duplicate_block): Do not update frequencies.
5630 (account_profile_record): Do not update frequencies.
5631 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
5632 for global heuristics.
5633 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
5634 (expected_loop_iterations_unbounded): Use counts only.
5635 * cfgloopmanip.c (scale_loop_profile): Simplify.
5636 (create_empty_loop_on_edge): Simplify
5638 (duplicate_loop_to_header_edge): Simplify
5639 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
5640 (update_br_prob_note): Take care of removing note when profile
5642 (relink_block_chain): Do not dump frequency.
5643 (rtl_account_profile_record): Use to_frequency.
5644 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
5645 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
5646 (cgraph_update_edges_for_call_stmt_node): Likewise.
5647 (cgraph_edge::verify_count_and_frequency): Update.
5648 (cgraph_node::verify_node): Temporarily disable frequency verification.
5649 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
5650 to_cgraph_frequency.
5651 (cgraph_edge::rebuild_edges): Convert to ipa counts.
5652 * cgraphunit.c (init_lowered_empty_function): Do not initialize
5654 (cgraph_node::expand_thunk): Update profile.
5655 * except.c (dw2_build_landing_pads): Do not update frequency.
5656 * final.c (compute_alignments): Use to_frequency.
5657 (dump_basic_block_info): Do not dump frequency.
5658 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
5659 (dump_gimple_bb_header): Do not dump frequency.
5660 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
5662 * gimple-streamer-in.c (input_bb): Do not stream frequency.
5663 * gimple-streamer-out.c (output_bb): Do not stream frequency.
5664 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
5665 (init_before_recovery): Do not update frequency.
5666 (sched_create_recovery_edges): Do not update frequency.
5667 * hsa-gen.c (convert_switch_statements): Do not update frequency.
5668 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
5669 (ipa_cp_c_finalize): Set max_count to uninitialized.
5670 * ipa-fnsummary.c (get_minimal_bb): Use counts.
5671 (param_change_prob): Use counts.
5672 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
5674 * ipa-split.c (consider_split): Use to_frequency.
5675 (split_function): Use to_frequency.
5676 * ira-build.c (loop_compare_func): Likewise.
5677 (mark_loops_for_removal): Likewise.
5678 (mark_all_loops_for_removal): Likewise.
5679 * loop-doloop.c (doloop_modify): Do not update frequency.
5680 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
5682 * lto-streamer-in.c (input_function): Update count_max.
5683 * omp-expand.c (expand_omp_taskreg): Update count_max.
5684 * omp-simd-clone.c (simd_clone_adjust): Update profile.
5685 * predict.c (maybe_hot_frequency_p): Use to_frequency.
5686 (maybe_hot_count_p): Use ipa counts only.
5687 (maybe_hot_bb_p): Simplify.
5688 (maybe_hot_edge_p): Simplify.
5689 (probably_never_executed): Do not take frequency argument.
5690 (probably_never_executed_bb_p): Do not pass frequency.
5691 (probably_never_executed_edge_p): Likewise.
5692 (combine_predictions_for_bb): Check that profile is nonzero.
5693 (propagate_freq): Do not set frequency.
5694 (drop_profile): Simplify.
5695 (counts_to_freqs): Simplify.
5696 (expensive_function_p): Use to_frequency.
5697 (propagate_unlikely_bbs_forward): Simplify.
5698 (determine_unlikely_bbs): Simplify.
5699 (estimate_bb_frequencies): Add hack to silence graphite issues.
5700 (compute_function_frequency): Use ipa counts.
5701 (pass_profile::execute): Update.
5702 (rebuild_frequencies): Use counts only.
5703 (force_edge_cold): Use counts only.
5704 * profile-count.c (profile_count::dump): Dump new count types.
5705 (profile_count::differs_from_p): Check compatiblity.
5706 (profile_count::to_frequency): New function.
5707 (profile_count::to_cgraph_frequency): New function.
5708 * profile-count.h (struct function): Declare.
5709 (enum profile_quality): Add profile_guessed_local and
5710 profile_guessed_global0.
5711 (class profile_proability): Decrease number of bits to 29;
5712 update from_reg_br_prob_note and to_reg_br_prob_note.
5713 (class profile_count: Update comment; decrease number of bits
5714 to 61. Check compatibility.
5715 (profile_count::compatible_p): New private member function.
5716 (profile_count::ipa_p): New member function.
5717 (profile_count::operator<): Handle global zero correctly.
5718 (profile_count::operator>): Handle global zero correctly.
5719 (profile_count::operator<=): Handle global zero correctly.
5720 (profile_count::operator>=): Handle global zero correctly.
5721 (profile_count::nonzero_p): New member function.
5722 (profile_count::force_nonzero): New member function.
5723 (profile_count::max): New member function.
5724 (profile_count::apply_scale): Handle IPA scalling.
5725 (profile_count::guessed_local): New member function.
5726 (profile_count::global0): New member function.
5727 (profile_count::ipa): New member function.
5728 (profile_count::to_frequency): Declare.
5729 (profile_count::to_cgraph_frequency): Declare.
5730 * profile.c (OVERLAP_BASE): Delete.
5731 (compute_frequency_overlap): Delete.
5732 (compute_branch_probabilities): Do not use compute_frequency_overlap.
5733 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
5734 * sched-ebb.c (rank): Use counts only.
5735 * shrink-wrap.c (handle_simple_exit): Use counts only.
5736 (try_shrink_wrapping): Use counts only.
5737 (place_prologue_for_one_component): Use counts only.
5738 * tracer.c (find_best_predecessor): Use to_frequency.
5739 (find_trace): Use to_frequency.
5740 (tail_duplicate): Use to_frequency.
5741 * trans-mem.c (expand_transaction): Do not update frequency.
5742 * tree-call-cdce.c: Do not update frequency.
5743 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
5744 (gimple_merge_blocks): Likewise.
5745 (gimple_split_edge): Likewise.
5746 (gimple_duplicate_sese_region): Likewise.
5747 (gimple_duplicate_sese_tail): Likewise.
5748 (move_sese_region_to_fn): Likewise.
5749 (gimple_account_profile_record): Likewise.
5750 (insert_cond_bb): Likewise.
5751 * tree-complex.c (expand_complex_div_wide): Likewise.
5752 * tree-eh.c (lower_resx): Update profile.
5753 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
5755 (initialize_cfun): Do not initialize frequencies
5756 (freqs_to_counts): Delete.
5757 (copy_cfg_body): Ignore count parameter.
5758 (copy_body): Update.
5759 (expand_call_inline): Update count_max.
5760 (optimize_inline_calls): Update count_max.
5761 (tree_function_versioning): Update count_max.
5762 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
5763 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
5765 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
5766 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
5767 (try_peel_loop): Likewise.
5768 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
5770 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
5771 (tree_transform_and_unroll_loop): Do not use frequencies
5772 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
5773 Use reliable prediction only.
5774 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
5775 * tree-ssa-sink.c (select_best_block): Use to_frequency.
5776 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
5777 probability scaling.
5778 * tree-ssa-threadupdate.c (create_block_for_threading): Do
5779 not update frequency
5780 (any_remaining_duplicated_blocks): Likewise.
5781 (update_profile): Likewise.
5782 (estimated_freqs_path): Delete.
5783 (freqs_to_counts_path): Delete.
5784 (clear_counts_path): Delete.
5785 (ssa_fix_duplicate_block_edges): Likewise.
5786 (duplicate_thread_path): Likewise.
5787 * tree-switch-conversion.c (gen_inbound_check): Use counts.
5788 * tree-tailcall.c (decrease_profile): Do not update frequency.
5789 (eliminate_tail_call): Likewise.
5790 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
5791 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
5792 (optimize_mask_stores): Likewise.
5793 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
5794 * ubsan.c (ubsan_expand_null_ifn): Update profile.
5795 (ubsan_expand_ptr_ifn): Update profile.
5796 * value-prof.c (gimple_ic): Simplify.
5797 * value-prof.h (gimple_ic): Update prototype.
5798 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
5799 * ipa-inline.c (compute_uninlined_call_time): Be sure that
5801 (want_inline_self_recursive_call_p): Likewise.
5802 (resolve_noninline_speculation): Only cummulate defined counts.
5803 (inline_small_functions): Use nonzero_p.
5804 (ipa_inline): Do not access freed node.
5806 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5808 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5809 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
5811 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
5813 * config/riscv/riscv.c (riscv_legitimize_move): Handle
5814 non-legitimate address.
5816 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5818 * config/rs6000/rs6000.md (*lt0_disi): Delete.
5819 (*lt0_<mode>di, *lt0_<mode>si): New.
5821 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5823 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
5824 TARGET_PAIRED_FLOAT.
5826 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
5827 Jim Wilson <jim.wilson@linaro.org>
5829 * config/aarch64/aarch64-cores.def (saphira): New CPU.
5830 * config/aarch64/aarch64-tune.md: Regenerated.
5831 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
5832 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
5834 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
5836 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
5837 (arc_expand_prologue): Restore blink for millicode.
5838 * config/arc/linux.h (LINK_EH_SPEC): Defined.
5840 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
5843 * config/i386/i386.c (ix86_vector_duplicate_value): Use
5844 gen_vec_duplicate after forcing the scalar into a register.
5846 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
5848 * combine (try_combine): Print the insns input to try_combine to the
5851 2017-11-02 Steve Ellcey <sellcey@cavium.com>
5854 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
5855 Remove second argument from aarch64_process_target_attr call.
5856 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
5858 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
5860 (aarch64_handle_attr_arch): Remove second argument.
5861 (aarch64_handle_attr_cpu): Ditto.
5862 (aarch64_handle_attr_tune): Ditto.
5863 (aarch64_handle_attr_isa_flags): Ditto.
5864 (aarch64_process_one_target_attr): Ditto.
5865 (aarch64_process_target_attr): Ditto.
5866 (aarch64_option_valid_attribute_p): Remove second argument.
5867 on aarch64_process_target_attr call.
5869 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5871 * diagnostic.c: Include "selftest-diagnostic.h".
5872 (selftest::assert_location_text): New function.
5873 (selftest::test_diagnostic_get_location_text): New function.
5874 (selftest::diagnostic_c_tests): Call it.
5876 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5878 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
5879 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
5880 (class selftest::test_diagnostic_context): Move to...
5881 * selftest-diagnostic.c: New file.
5882 * selftest-diagnostic.h: New file.
5884 2017-11-02 James Bowman <james.bowman@ftdichip.com>
5886 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
5887 offset range for FT32B.
5888 * config/ft32/ft32.h: option "mcompress" enables relaxation.
5889 * config/ft32/ft32.md: Add TARGET_NOPM.
5890 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
5891 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
5893 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
5895 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
5897 2017-11-02 Jeff Law <law@redhat.com>
5899 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
5900 virtual keyword on FINAL OVERRIDE members.
5902 * tree-ssa-propagate.h (ssa_propagation_engine): Group
5903 virtuals together. Add virtual destructor.
5904 (substitute_and_fold_engine): Similarly.
5906 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
5908 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
5910 2017-11-02 Richard Biener <rguenther@suse.de>
5912 PR tree-optimization/82795
5913 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
5915 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5917 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
5919 (gcc_GAS_CHECK_FEATURE): Remove.
5920 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
5921 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
5922 * configure: Regenerate.
5924 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
5926 * config/arc/arc.c (hwloop_optimize): Account for empty
5929 2017-11-02 Richard Biener <rguenther@suse.de>
5932 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
5933 Truncate ARRAY_REF index and element size.
5935 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
5937 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
5939 2017-11-01 Jeff Law <law@redhat.com>
5941 * tree-ssa-ccp.c (ccp_folder): New class derived from
5942 substitute_and_fold_engine.
5943 (ccp_folder::get_value): New member function.
5944 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
5945 (ccp_fold_stmt): Remove prototype.
5946 (ccp_finalize): Call substitute_and_fold from the ccp_class.
5947 * tree-ssa-copy.c (copy_folder): New class derived from
5948 substitute_and_fold_engine.
5949 (copy_folder::get_value): Renamed from get_value.
5950 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
5951 * tree-vrp.c (vrp_folder): New class derived from
5952 substitute_and_fold_engine.
5953 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
5954 (vrp_folder::get_value): New member function.
5955 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
5956 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
5957 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
5958 provide a class interface to folder/substitute routines.
5959 (ssa_prop_fold_stmt_fn): Remove typedef.
5960 (ssa_prop_get_value_fn): Likewise.
5961 (subsitute_and_fold): Remove prototype.
5962 (replace_uses_in): Likewise.
5963 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
5964 Renamed from replace_uses_in. Call the virtual member function
5965 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
5966 (substitute_and_fold_dom_walker): Remove initialization of
5967 data member entries for calbacks. Add substitute_and_fold_engine
5968 member and initialize it.
5969 (substitute_and_fold_dom_walker::before_dom_children0: Use the
5970 member functions for get_value, replace_phi_args_in c
5971 replace_uses_in, and fold_stmt calls.
5972 (substitute_and_fold_engine::substitute_and_fold): Renamed from
5973 substitute_and_fold. Remove assert. Update ctor call.
5975 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
5976 (ssa_prop_visit_phi_fn): Likewise.
5977 (class ssa_propagation_engine): New class to provide an interface
5979 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
5981 (ssa_prop_visit_phi): Likewise.
5982 (ssa_propagation_engine::simulate_stmt): Moved into class.
5983 Call visit_phi/visit_stmt from the class rather than via
5984 file scoped static variables.
5985 (ssa_propagation_engine::simulate_block): Moved into class.
5986 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
5987 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
5988 set file scoped statics for the visit_stmt/visit_phi callbacks.
5989 * tree-complex.c (complex_propagate): New class derived from
5990 ssa_propagation_engine.
5991 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
5992 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
5993 (tree_lower_complex): Call ssa_propagate via the complex_propagate
5995 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
5996 ssa_propagation_engine.
5997 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
5998 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
5999 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
6000 * tree-ssa-copy.c (copy_prop): New class derived from
6001 ssa_propagation_engine.
6002 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
6003 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
6004 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
6005 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
6006 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
6007 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
6008 (execute_vrp): Call ssa_propagate from the vrp_prop class.
6010 2017-11-01 Jakub Jelinek <jakub@redhat.com>
6012 PR rtl-optimization/82778
6013 PR rtl-optimization/82597
6014 * compare-elim.c (struct comparison): Add in_a_setter field.
6015 (find_comparison_dom_walker::before_dom_children): Remove killed
6016 bitmap and df_simulate_find_defs call, instead walk the defs.
6017 Compute last_setter and initialize in_a_setter. Merge definitions
6018 with first initialization for a few variables.
6019 (try_validate_parallel): Use insn_invalid_p instead of
6020 recog_memoized. Return insn rather than just the pattern.
6021 (try_merge_compare): Fix up comment. Don't uselessly test if
6022 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
6024 (execute_compare_elim_after_reload): Remove df_chain_add_problem
6027 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6028 Alan Hayward <alan.hayward@arm.com>
6029 David Sherwood <david.sherwood@arm.com>
6031 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
6032 aarch64_hard_regno_nregs to get the number of registers
6035 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6036 Alan Hayward <alan.hayward@arm.com>
6037 David Sherwood <david.sherwood@arm.com>
6039 * config/aarch64/constraints.md (Upl): Rename to...
6041 * config/aarch64/aarch64.md
6042 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
6045 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6046 Alan Hayward <alan.hayward@arm.com>
6047 David Sherwood <david.sherwood@arm.com>
6049 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
6050 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
6053 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6054 Alan Hayward <alan.hayward@arm.com>
6055 David Sherwood <david.sherwood@arm.com>
6057 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
6058 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
6059 (aarch64_evpc_dup): Generate rtl direcly, rather than using
6061 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
6062 of a single element.
6063 * config/aarch64/iterators.md: Add a comment above the permute
6064 unspecs to say that they are generated directly by
6065 aarch64_expand_vec_perm_const.
6066 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
6068 2017-11-01 Nathan Sidwell <nathan@acm.org>
6070 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
6072 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
6074 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
6075 and medany code models, and describe what they do.
6077 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6079 Revert accidental duplicate:
6081 * combine.c (can_change_dest_mode): Reject changes in
6082 REGMODE_NATURAL_SIZE.
6084 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
6086 PR rtl-optimization/64682
6087 PR rtl-optimization/69567
6088 PR rtl-optimization/69737
6089 PR rtl-optimization/82683
6090 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
6091 register mentioned in the note, drop the note, unless it came from I3,
6092 in which case it should go to I3 again.
6094 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6096 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
6097 and return false if not.
6098 (clear_bytes_written_by, live_bytes_read): Update accordingly.
6100 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6102 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
6103 range is known to be empty.
6105 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6106 Alan Hayward <alan.hayward@arm.com>
6107 David Sherwood <david.sherwood@arm.com>
6109 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
6110 and CONST_VECTOR_NUNITS instead of computing the number of units from
6111 the byte sizes of the vector and element.
6112 (simplify_binary_operation_1): Likewise.
6113 (simplify_const_binary_operation): Likewise.
6114 (simplify_ternary_operation): Likewise.
6116 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6117 Alan Hayward <alan.hayward@arm.com>
6118 David Sherwood <david.sherwood@arm.com>
6120 * var-tracking.c (INT_MEM_OFFSET): Replace with...
6121 (int_mem_offset): ...this new function.
6122 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
6123 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
6124 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
6127 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6128 Alan Hayward <alan.hayward@arm.com>
6129 David Sherwood <david.sherwood@arm.com>
6131 * lower-subreg.c (interesting_mode_p): New function.
6132 (compute_costs, find_decomposable_subregs, decompose_register)
6133 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
6134 (resolve_clobber, dump_choices): Use it.
6136 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6137 Alan Hayward <alan.hayward@arm.com>
6138 David Sherwood <david.sherwood@arm.com>
6140 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
6142 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6143 Alan Hayward <alan.hayward@arm.com>
6144 David Sherwood <david.sherwood@arm.com>
6146 * alias.c (find_base_value, find_base_term): Only process integer
6147 truncations. Check the precision rather than the size.
6149 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6150 Alan Hayward <alan.hayward@arm.com>
6151 David Sherwood <david.sherwood@arm.com>
6153 * machmode.h (is_narrower_int_mode): New function
6154 * optabs.c (expand_float, expand_fix): Use it.
6155 * dwarf2out.c (rotate_loc_descriptor): Likewise.
6157 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6158 Alan Hayward <alan.hayward@arm.com>
6159 David Sherwood <david.sherwood@arm.com>
6161 * rtl.h (narrower_subreg_mode): New function.
6162 * ira-color.c (update_costs_from_allocno): Use it.
6164 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6165 Alan Hayward <alan.hayward@arm.com>
6166 David Sherwood <david.sherwood@arm.com>
6168 * optabs-query.h (convert_optab_p): New function, split out from...
6169 (convert_optab_handler): ...here.
6170 (widening_optab_handler): Delete.
6171 (find_widening_optab_handler): Remove permit_non_widening parameter.
6172 (find_widening_optab_handler_and_mode): Likewise. Provide an
6173 override that operates on mode class wrappers.
6174 * optabs-query.c (widening_optab_handler): Delete.
6175 (find_widening_optab_handler_and_mode): Remove permit_non_widening
6176 parameter. Assert that the two modes are the same class and that
6177 the "from" mode is narrower than the "to" mode. Use
6178 convert_optab_handler instead of widening_optab_handler.
6179 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
6180 instead of widening_optab_handler.
6181 * expr.c (expand_expr_real_2): Update calls to
6182 find_widening_optab_handler.
6183 * optabs.c (expand_widen_pattern_expr): Likewise.
6184 (expand_binop_directly): Take the insn_code as a parameter.
6185 (expand_binop): Only call find_widening_optab_handler for
6186 conversion optabs; use optab_handler otherwise. Update calls
6187 to find_widening_optab_handler and expand_binop_directly.
6188 Use convert_optab_handler instead of widening_optab_handler.
6189 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
6190 find_widening_optab_handler and use scalar_mode rather than
6192 (convert_plusminus_to_widen): Likewise.
6194 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6195 Alan Hayward <alan.hayward@arm.com>
6196 David Sherwood <david.sherwood@arm.com>
6198 * machmode.h (fixed_size_mode): New class.
6199 * rtl.h (get_pool_mode): Return fixed_size_mode.
6200 * gengtype.c (main): Add fixed_size_mode.
6201 * target.def (get_raw_result_mode): Return a fixed_size_mode.
6202 (get_raw_arg_mode): Likewise.
6203 * doc/tm.texi: Regenerate.
6204 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
6205 * targhooks.c (default_get_reg_raw_mode): Likewise.
6206 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
6207 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
6208 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
6209 (msp430_get_raw_result_mode): Likewise.
6210 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
6211 * dbxout.c (dbxout_parms): Require fixed-size modes.
6212 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
6213 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
6214 * omp-low.c (lower_oacc_reductions): Likewise.
6215 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
6216 (simplify_subreg): Update accordingly.
6217 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
6218 (force_const_mem): Update accordingly. Return NULL_RTX for modes
6219 that aren't fixed-size.
6220 (get_pool_mode): Return a fixed_size_mode.
6221 (output_constant_pool_2): Take a fixed_size_mode.
6223 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6224 Alan Hayward <alan.hayward@arm.com>
6225 David Sherwood <david.sherwood@arm.com>
6227 * doc/rtl.texi (vec_series): Document.
6228 (const): Say that the operand can be a vec_series.
6229 * rtl.def (VEC_SERIES): New rtx code.
6230 * rtl.h (const_vec_series_p_1): Declare.
6231 (const_vec_series_p): New function.
6232 * emit-rtl.h (gen_const_vec_series): Declare.
6233 (gen_vec_series): Likewise.
6234 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
6235 (gen_vec_series): Likewise.
6236 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
6237 * simplify-rtx.c (simplify_unary_operation): Handle negations
6239 (simplify_binary_operation_series): New function.
6240 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
6241 (test_vector_ops_series): New function.
6242 (test_vector_ops): Call it.
6243 * config/powerpcspe/altivec.md (altivec_lvsl): Use
6244 gen_const_vec_series.
6245 (altivec_lvsr): Likewise.
6246 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
6248 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6249 Alan Hayward <alan.hayward@arm.com>
6250 David Sherwood <david.sherwood@arm.com>
6252 * doc/rtl.texi (const): Update description of address constants.
6253 Say that vector constants are allowed too.
6254 * common.md (E, F): Use CONSTANT_P instead of checking for
6256 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
6257 checking for CONST_VECTOR.
6258 * expmed.c (make_tree): Use build_vector_from_val for a CONST
6260 * expr.c (expand_expr_real_2): Check for vector modes instead
6261 of checking for CONST_VECTOR.
6262 * rtl.h (const_vec_p): New function.
6263 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
6264 (unwrap_const_vec_duplicate): Handle them here too.
6266 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6267 David Malcolm <dmalcolm@redhat.com>
6268 Alan Hayward <alan.hayward@arm.com>
6269 David Sherwood <david.sherwood@arm.com>
6271 * rtl.h (vec_duplicate_p): New function.
6272 * selftest-rtl.c (assert_rtx_eq_at): New function.
6273 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
6274 (assert_rtx_eq_at): Declare.
6275 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
6276 * selftest-run-tests.c (selftest::run_tests): Call it.
6277 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
6278 (simplify_unary_operation_1): Recursively handle vector duplicates.
6279 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
6281 (simplify_subreg): Handle subregs of vector duplicates.
6282 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
6283 (selftest::simplify_rtx_c_tests): New functions.
6285 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6286 Alan Hayward <alan.hayward@arm.com>
6287 David Sherwood <david.sherwood@arm.com>
6289 * emit-rtl.h (gen_const_vec_duplicate): Declare.
6290 (gen_vec_duplicate): Likewise.
6291 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
6293 (gen_const_vector): ...here.
6294 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
6295 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
6296 whose elements are all equal.
6297 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
6298 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
6299 (simplify_relational_operation): Likewise.
6300 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
6302 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
6303 (aarch64_expand_vector_init): Likewise.
6304 * config/arm/arm.c (neon_vdup_constant): Likewise.
6305 (neon_expand_vector_init): Likewise.
6306 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
6307 (arm_block_set_unaligned_vect): Likewise.
6308 (arm_block_set_aligned_vect): Likewise.
6309 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
6310 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
6311 (expand_vec_perm_even_odd_pack): Likewise.
6312 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
6313 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
6314 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
6315 gen_const_vec_duplicate.
6316 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
6317 * config/mips/mips.c (mips_gen_const_int_vector): Use
6318 gen_const_vec_duplicate.
6319 (mips_expand_vector_init): Use CONST0_RTX.
6320 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
6321 (define_split): Use gen_const_vec_duplicate.
6322 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
6323 (define_split): Use gen_const_vec_duplicate.
6324 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
6325 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
6326 * config/spu/spu.c (spu_const): Likewise.
6328 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6329 Alan Hayward <alan.hayward@arm.com>
6330 David Sherwood <david.sherwood@arm.com>
6332 * combine.c (can_change_dest_mode): Reject changes in
6333 REGMODE_NATURAL_SIZE.
6335 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
6337 * configure.ac (--enable-libssp): New.
6338 (gcc_cv_libc_provides_ssp): Check for explicit setting before
6339 trying to determine target-specific default. Adjust indentation.
6340 * configure: Regenerated.
6341 * doc/install.texi (Configuration): Expand --disable-libssp
6344 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
6346 config/i386/i386.c (ix86_expand_epilogue): Correct stack
6349 2017-10-31 Martin Jambor <mjambor@suse.cz>
6352 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
6354 2017-10-31 David Malcolm <dmalcolm@redhat.com>
6356 * auto-profile.c (autofdo_source_profile::read): Use
6357 UNKNOWN_LOCATION rather than 0.
6358 * diagnostic-core.h (warning_at_rich_loc): Rename to...
6359 (warning_at): ...this overload.
6360 (warning_at_rich_loc_n): Rename to...
6361 (warning_n): ...this overload.
6362 (error_at_rich_loc): Rename to...
6363 (error_at): ...this overload.
6364 (pedwarn_at_rich_loc): Rename to...
6365 (pedwarn): ...this overload.
6366 (permerror_at_rich_loc): Rename to...
6367 (permerror): ...this overload.
6368 (inform_at_rich_loc): Rename to...
6369 (inform): ...this overload.
6370 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
6371 (diagnostic_n_impl_richloc): Rename to...
6372 (diagnostic_n_impl): ...this rich_location *-based decl.
6373 (inform_at_rich_loc): Rename to...
6374 (inform): ...this, and add an assertion.
6375 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
6376 (warning_at_rich_loc): Rename to...
6377 (warning_at): ...this, and add an assertion.
6378 (warning_at_rich_loc_n): Rename to...
6379 (warning_n): ...this, and add an assertion.
6380 (warning_n): Update location_t-based implementation for removal of
6381 location_t-based diagnostic_n_impl.
6382 (pedwarn_at_rich_loc): Rename to...
6383 (pedwarn): ...this, and add an assertion.
6384 (permerror_at_rich_loc): Rename to...
6385 (permerror): ...this, and add an assertion.
6386 (error_n): Update for removal of location_t-based diagnostic_n_impl.
6387 (error_at_rich_loc): Rename to...
6388 (error_at): ...this, and add an assertion.
6389 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
6390 (driver::do_spec_on_infiles): Likewise.
6391 * substring-locations.c (format_warning_va): Update for renaming
6392 of inform_at_rich_loc.
6394 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
6396 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
6397 _Float<N>X built-in functions so that the variant without the
6398 "__builtin_" prefix is only enabled for the GNU C and Objective C
6399 languages when they are in non-strict ANSI/ISO mode.
6400 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
6401 * target.def (floatn_builtin_p): Add a target hook to control
6402 whether _Float<N> and _Float<N>X built-in functions without the
6403 "__builtin_" prefix are enabled, and return true for C and
6404 Objective C in the default hook. Include langhooks.h in
6406 * targhooks.h (default_floatn_builtin_p): Likewise.
6407 * targhooks.c (default_floatn_builtin_p): Likewise.
6408 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
6409 floatn_builtin_p target hook.
6410 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
6412 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
6413 Eric Botcazou <ebotcazou@adacore.com>
6415 PR rtl-optimization/81803
6416 * lra-constraints.c (curr_insn_transform): Also reload the whole
6417 register for a strict subreg no wider than a word if this is for
6418 a WORD_REGISTER_OPERATIONS target.
6420 2017-10-31 Jason Merrill <jason@redhat.com>
6422 * gdbinit.in: Skip over inlines from timevar.h.
6424 2017-10-31 Martin Liska <mliska@suse.cz>
6426 * doc/gcov.texi: Document new option.
6427 * gcov.c (print_usage): Likewise print it.
6428 (process_args): Support the argument.
6429 (format_count): New function.
6430 (format_gcov): Use the function.
6432 2017-10-31 Martin Liska <mliska@suse.cz>
6434 * gcov.c (struct name_map): do not use typedef.
6435 Define operator== and operator<.
6436 (name_search): Remove.
6437 (name_sort): Remove.
6438 (main): Do not allocate names.
6439 (process_file): Add vertical space.
6440 (generate_results): Use std::find.
6441 (release_structures): Do not release memory.
6442 (find_source): Use std::find.
6444 2017-10-31 Martin Liska <mliska@suse.cz>
6446 * gcov.c (struct line_info): Remove it's typedef.
6447 (line_info::line_info): Add proper ctor.
6448 (line_info::has_block): Do not use a typedef.
6449 (struct source_info): Do not use typedef.
6450 (circuit): Likewise.
6451 (get_cycles_count): Likewise.
6452 (output_intermediate_file): Iterate via vector iterator.
6453 (add_line_counts): Use std::vector methods.
6454 (accumulate_line_counts): Likewise.
6455 (output_lines): Likewise.
6457 2017-10-31 Martin Liska <mliska@suse.cz>
6459 * gcov.c (struct source_info): Remove typedef.
6460 (source_info::source_info): Add proper ctor.
6461 (accumulate_line_counts): Use struct, not it's typedef.
6462 (output_gcov_file): Likewise.
6463 (output_lines): Likewise.
6464 (main): Do not allocate an array.
6465 (output_intermediate_file): Use size of vector container.
6466 (process_file): Resize the vector.
6467 (generate_results): Do not preallocate, use newly added vector
6469 (release_structures): Do not release sources.
6470 (find_source): Use vector methods.
6471 (add_line_counts): Do not use typedef.
6473 2017-10-31 Martin Liska <mliska@suse.cz>
6475 * doc/gcov.texi: Document that.
6476 * gcov.c (add_line_counts): Mark lines with a non-executed
6478 (output_line_beginning): Handle such lines.
6479 (output_lines): Pass new argument.
6480 (output_intermediate_file): Print it in intermediate format.
6482 2017-10-31 Martin Liska <mliska@suse.cz>
6484 * color-macros.h: New file.
6485 * diagnostic-color.c: Factor out color related to macros to
6487 * doc/gcov.texi: Document -k option.
6488 * gcov.c (INCLUDE_STRING): Include string.h.
6489 (print_usage): Add -k option.
6490 (process_args): Parse it.
6491 (pad_count_string): New function.
6492 (output_line_beginning): Likewise.
6493 (DEFAULT_LINE_START): New macro.
6494 (output_lines): Support color output.
6496 2017-10-31 Martin Liska <mliska@suse.cz>
6498 PR gcov-profile/82633
6499 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
6500 their interaction with GCOV infrastructure.
6501 * configure.ac: Add -fkeep-{inline,static}-functions to
6503 * configure: Regenerate.
6505 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
6508 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
6510 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
6513 * config/rs6000/rs6000.md (allocate_stack): Force update interval
6514 into a register if it does not fit into an immediate offset field.
6516 2017-10-31 Olivier Hainque <hainque@adacore.com>
6518 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
6520 2017-10-31 Julia Koval <julia.koval@intel.com>
6522 * config.gcc: Add gfniintrin.h.
6523 * config/i386/gfniintrin.h: New.
6524 * config/i386/i386-builtin-types.def
6525 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
6526 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
6527 __builtin_ia32_vgf2p8affineinvqb_v32qi,
6528 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
6529 __builtin_ia32_vgf2p8affineinvqb_v16qi,
6530 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
6531 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
6532 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
6533 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
6534 V64QI_FTYPE_V64QI_V64QI_INT): New types.
6535 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
6536 * config/i386/immintrin.h: Include gfniintrin.h.
6537 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
6539 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
6541 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
6543 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
6545 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
6546 (arm_ashldi3_1bit): Remove pattern.
6547 (ashrdi3): Remove shift by 1 expansion.
6548 (arm_ashrdi3_1bit): Remove pattern.
6549 (lshrdi3): Remove shift by 1 expansion.
6550 (arm_lshrdi3_1bit): Remove pattern.
6551 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
6552 cost of ashldi3 by 1.
6553 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
6554 (<shift>di3_neon): Likewise.
6556 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6558 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
6559 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
6560 and (*aarch64_simd_mov<VQ:mode>).
6561 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
6562 pattern alternative.
6563 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
6564 attributes to match pattern alternative.
6566 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
6568 * config.gcc (powerpc*-*-*): Add emmintrin.h.
6569 * config/rs6000/emmintrin.h: New file.
6570 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
6572 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
6574 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
6575 * (movdi_vfp_cortexa8): Remove pattern.
6577 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6579 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
6581 (Specific, alpha*-dec-osf5.1): Remove.
6582 (Specific, mips-sgi-irix5): Remove.
6583 (Specific, mips-sgi-irix6): Remove.
6585 2017-10-30 Jakub Jelinek <jakub@redhat.com>
6588 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
6589 arguments to clear_bit_region_be.
6591 2017-10-30 Jim Wilson <wilson@tuliptree.org>
6593 * gimplify.c: Include memmodel.h.
6595 2017-10-30 Martin Jambor <mjambor@suse.cz>
6597 * omp-grid.c (grid_attempt_target_gridification): Also insert a
6598 condition whether loop should be executed at all.
6600 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
6602 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6603 gimple folding of vec_madd() intrinsics.
6604 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
6605 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
6606 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
6608 2017-10-30 Richard Biener <rguenther@suse.de>
6610 PR tree-optimization/82762
6612 2017-10-23 Richard Biener <rguenther@suse.de>
6614 PR tree-optimization/82129
6616 2017-08-01 Richard Biener <rguenther@suse.de>
6618 PR tree-optimization/81181
6619 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6620 (compute_antic): ... end of iteration here.
6622 2017-10-30 Joseph Myers <joseph@codesourcery.com>
6624 * doc/invoke.texi (C Dialect Options): Document -std=c17,
6625 -std=iso9899:2017 and -std=gnu17.
6626 * doc/standards.texi (C Language): Document C17 support.
6627 * doc/cpp.texi (Overview): Mention -std=c17.
6628 (Standard Predefined Macros): Document C11 and C17 values of
6629 __STDC_VERSION__. Do not refer to C99 support as incomplete.
6630 * doc/extend.texi (Inline): Do not list individual options for
6631 standards newer than C99.
6632 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
6634 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
6637 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
6639 * asan.c (asan_finish_file): Align asan globals array by shadow
6642 2017-10-30 Jakub Jelinek <jakub@redhat.com>
6645 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
6646 (struct store_immediate_info): Add bitregion_start and bitregion_end
6648 (store_immediate_info::store_immediate_info): Add brs and bre
6649 arguments and initialize bitregion_{start,end} from those.
6650 (struct merged_store_group): Add bitregion_start, bitregion_end,
6651 align_base and mask fields. Drop unnecessary struct keyword from
6652 struct store_immediate_info. Add do_merge method.
6653 (clear_bit_region_be): Use memset instead of loop storing zeros.
6654 (merged_store_group::do_merge): New method.
6655 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
6656 stores as long as the surrounding bitregions have no gaps.
6657 (merged_store_group::merge_overlapping): Use do_merge.
6658 (merged_store_group::apply_stores): Test that bitregion_{start,end}
6659 is byte aligned, rather than requiring that start and width are
6660 byte aligned. Drop unnecessary struct keyword from
6661 struct store_immediate_info. Allocate and populate also mask array.
6662 Make start of the arrays relative to bitregion_start rather than
6663 start and size them according to bitregion_{end,start} difference.
6664 (struct imm_store_chain_info): Drop unnecessary struct keyword from
6665 struct store_immediate_info.
6666 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
6667 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
6668 struct keyword from struct store_immediate_info.
6669 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
6670 between stores as long as the surrounding bitregions have no gaps.
6672 (struct split_store): Add orig non-static data member.
6673 (split_store::split_store): Initialize orig to false.
6674 (find_constituent_stmts): Return store_immediate_info *, non-NULL
6675 if there is exactly a single original stmt. Change stmts argument
6676 to pointer from reference, if NULL, don't push anything to it. Add
6677 first argument, use it to optimize skipping over orig stmts that
6678 are known to be before bitpos already. Simplify.
6679 (split_group): Return unsigned int count how many stores are or
6680 would be needed rather than a bool. Add allow_unaligned argument.
6681 Change split_stores argument from reference to pointer, if NULL,
6682 only do a dry run computing how many stores would be produced.
6683 Rewritten algorithm to use both alignment and misalign if
6684 !allow_unaligned and handle bitfield stores with gaps.
6685 (imm_store_chain_info::output_merged_store): Set start_byte_pos
6686 from bitregion_start instead of start. Compute allow_unaligned
6687 here, if true, do 2 split_group dry runs to compute which one
6688 produces fewer stores and prefer aligned if equal. Punt if
6689 new count is bigger or equal than original before emitting any
6690 statements, rather than during that. Remove no longer needed
6691 new_ssa_names tracking. Replace num_stmts with
6692 split_stores.length (). Use 32-bit stack allocated entries
6693 in split_stores auto_vec. Try to reuse original store lhs/rhs1
6694 if possible. Handle bitfields with gaps.
6695 (pass_store_merging::execute): Ignore bitsize == 0 stores.
6696 Compute bitregion_{start,end} for the stores and construct
6697 store_immediate_info with that. Formatting fixes.
6699 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
6702 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
6703 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
6705 2017-10-29 Jim Wilson <wilson@tuliptree.org>
6707 * gimplify.c: Include tm_p.h.
6709 * common.opt (gcoff): Re-add as ignored option.
6710 (gcoff1, gcoff2, gcoff3): Likewise.
6712 * Makefile.in (OBJS): Delete sdbout.o.
6713 (GTFILES): Delete $(srcdir)/sdbout.c.
6714 * debug.h: Delete sdb_debug_hooks.
6715 * final.c: Delete sdbout.h include.
6716 (final_scan_insn): Delete SDB_DEBUG check.
6717 (rest_of_clean_state): Likewise.
6718 * output.h: Delete sdb_begin_function_line.
6721 * toplev.c: Delete sdbout.h include.
6722 (process_options): Delete SDB_DEBUG check.
6723 * tree-core.h (tree_type_common): Delete pointer field of
6725 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
6726 TYPE_SYMTAB_POINTER.
6727 * tree.h (TYPE_SYMTAB_POINTER): Delete.
6728 (TYPE_SYMTAB_IS_POINTER): Delete.
6729 (TYPE_SYMTAB_IS_DIE): Renumber.
6730 * xcoffout.c: Refer to former sdbout.c file.
6731 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
6733 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
6734 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
6735 (Debugging Options): Delete -gcoff.
6736 (-gstabs): Delete SDB reference.
6738 (-gcoff@var{level}): Delete.
6739 * doc/passes.texi (Debugging information output): Delete SDB and
6740 sdbout.c references.
6741 * doc/tm.texi: Regenerate.
6742 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
6743 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
6745 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
6746 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
6747 (SDB_DEBUGGING_INFO): Delete.
6748 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
6749 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
6750 * target.def (output_source_filename): Delete COFF reference.
6752 * common.opt (gcoff): Delete.
6753 (gxcoff+): Update Negative chain.
6754 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
6756 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
6757 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
6758 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
6760 (expand_function_start): Change sdb reference to past tense.
6761 (expand_function_end): Change sdb reference to past tense.
6762 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
6763 * opts.c (debug_type_names): Delete coff entry.
6764 (common_handle_option): Delete OPT_gcoff case.
6765 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
6767 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
6768 * config/cris/cris.h: Delete SDB reference in comment.
6769 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
6770 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
6771 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
6772 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
6774 (ix86_expand_prologue): Likewise.
6775 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
6776 * config/ia64/ia64.h: Likewise.
6777 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
6778 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
6780 * config/mmix/mmix.h: Likewise.
6781 * config/nds32/nds32.c: Likewise.
6782 * config/stormy/storym16.h: Likewise.
6783 * config/visium/visium.h: Likewise.
6784 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
6786 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
6788 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
6789 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
6790 FRAME_POINTER_REGNUM point at high end of local var area.
6792 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
6794 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
6795 Move comment around. Do not reset best_edge for a copiable
6796 destination if the copy would cause a partition change.
6797 (better_edge_p): Remove redundant check.
6799 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6801 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
6803 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
6805 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
6806 for math functions that have _Float<N> and _Float<N>X variants.
6807 (mathfn_built_in_2): Add support for math functions that have
6808 _Float<N> and _Float<N>X variants.
6809 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
6810 (expand_builtin_mathfn_ternary): Add support for fma with
6811 _Float<N> and _Float<N>X variants.
6812 (expand_builtin): Likewise.
6813 (fold_builtin_3): Likewise.
6814 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
6815 create math function _Float<N> and _Float<N>X variants as external
6817 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
6818 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
6819 the __builtin_ prefix and if not strict ansi, without the prefix.
6820 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
6821 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
6822 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
6823 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
6824 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
6825 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
6826 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
6827 function signatures for fma _Float<N> and _Float<N>X variants.
6828 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
6829 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
6830 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
6831 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
6832 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
6833 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
6834 * gencfn-macros.c (print_case_cfn): Add support for math functions
6835 that have _Float<N> and _Float<N>X variants.
6836 (print_define_operator_list): Likewise.
6837 (fltfn_suffixes): Likewise.
6839 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
6840 for math functions that have _Float<N> and _Float<N>X variants.
6841 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
6842 and _Float<N>X variants.
6843 (COPYSIGN): Likewise.
6846 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
6847 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
6849 (integer_valued_read_call_p): Likewise.
6850 * fold-const-call.c (fold_const_call_ss): Likewise.
6851 (fold_const_call_sss): Add support for copysign, fmin, and fmax
6852 _Float<N> and _Float<N>X variants.
6853 (fold_const_call_ssss): Add support for fma _Float<N> and
6854 _Float<N>X variants.
6855 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
6856 support for copysign and fma _Float<N> and _Float<N>X variants.
6857 (backprop::process_builtin_call_use): Likewise.
6858 * tree-call-cdce.c (can_test_argument_range); Add support for
6859 sqrt _Float<N> and _Float<N>X variants.
6860 (edom_only_function): Likewise.
6861 (get_no_error_domain): Likewise.
6862 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
6863 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
6864 copysign _Float<N> and _Float<N>X variants.
6865 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
6866 handled by machine independent code.
6867 (FMAF128): Likewise.
6868 * doc/cpp.texi (Common Predefined Macros): Document defining
6869 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
6870 fma _Float<N> and _Float<N>X variants.
6872 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6875 * config/i386/i386-modes.def (CCFPU): Remove definition.
6876 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
6877 (ix86_cc_modes_compatible): Ditto.
6878 (ix86_expand_carry_flag_compare): Ditto.
6879 (ix86_expand_int_movcc): Ditto.
6880 (ix86_expand_int_addcc): Ditto.
6881 (ix86_reverse_condition): Ditto.
6882 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
6883 Return true/false for unordered/ordered fp comparisons.
6884 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
6885 (ix86_prepare_fp_compare_args): Update for rename.
6886 (ix86_expand_fp_compare): Update for rename. Generate unordered
6887 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
6888 (ix86_expand_sse_compare_and_jump): Ditto.
6889 * config/i386/predicates.md (fcmov_comparison_operator):
6890 Remove CCFPU mode handling.
6891 (ix86_comparison_operator): Ditto.
6892 (ix86_carry_flag_operator): Ditto.
6893 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
6894 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
6895 (*cmpu<mode>_cc_i387): Ditto.
6896 (FPCMP): Remove mode iterator.
6897 (unord): Remove mode attribute.
6898 (unord_subst): New define_subst transformation
6899 (unord): New define_subst attribute.
6901 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
6902 (*cmpi<unord>xf_i387): Ditto.
6903 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
6904 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
6905 using unord_subst transformation.
6906 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
6907 (round_saeonly): Also handle CCFP mode.
6908 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
6909 Remove UNSPEC_SAHF unspec handling.
6911 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6913 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
6915 2017-10-27 Jeff Law <law@redhat.com>
6917 * gimple-ssa-sprintf.c: Include domwalk.h.
6918 (class sprintf_dom_walker): New class, derived from dom_walker.
6919 (sprintf_dom_walker::before_dom_children): New function.
6920 (struct call_info): Moved into sprintf_dom_walker class
6921 (compute_formath_length, handle_gimple_call): Likewise.
6922 (sprintf_length::execute): Call the dominator walker rather
6923 than walking the statements.
6925 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
6926 gimple statement locations.
6927 (check_array_bounds): Corresponding changes. Get the statement's
6928 location directly from wi->stmt.
6930 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
6933 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
6935 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6937 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
6938 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
6940 2017-10-27 Jakub Jelinek <jakub@redhat.com>
6943 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
6944 * config/i386/i386.c (maybe_get_pool_constant): Removed.
6945 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
6946 maybe_get_pool_constant.
6947 * config/i386/predicates.md (zero_extended_scalar_load_operand):
6950 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6952 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
6953 2.26 caveat. Update gas and gld versions.
6954 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
6957 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6959 * cgraph.h (set_malloc_flag): Declare.
6960 * cgraph.c (set_malloc_flag_1): New function.
6961 (set_malloc_flag): Likewise.
6962 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
6963 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
6965 (read_ipa_call_summary): Add support for reading is_return_callee.
6966 (write_ipa_call_summary): Stream is_return_callee.
6967 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
6968 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
6969 ipa-prop.h, ipa-fnsummary.h.
6970 (pure_const_names): Change to static.
6971 (malloc_state_e): Define.
6972 (malloc_state_names): Define.
6973 (funct_state_d): Add field malloc_state.
6974 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
6975 (check_retval_uses): New function.
6976 (malloc_candidate_p): Likewise.
6977 (analyze_function): Add support for malloc attribute.
6978 (pure_const_write_summary): Stream malloc_state.
6979 (pure_const_read_summary): Add support for reading malloc_state.
6980 (dump_malloc_lattice): New function.
6981 (propagate_malloc): New function.
6982 (warn_function_malloc): New function.
6983 (ipa_pure_const::execute): Call propagate_malloc and
6984 ipa_free_fn_summary.
6985 (pass_local_pure_const::execute): Add support for malloc attribute.
6986 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
6987 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
6989 2017-10-27 Martin Liska <mliska@suse.cz>
6991 PR gcov-profile/82457
6992 * doc/invoke.texi: Document that one needs a non-strict ISO mode
6993 for fork-like functions to be properly instrumented.
6995 2017-10-27 Richard Biener <rguenther@suse.de>
6998 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
6999 info when we redirected EH.
7001 2017-10-26 Michael Collison <michael.collison@arm.com>
7003 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
7005 (<optab>_trunchf<GPI:mode>2: New pattern.
7006 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
7007 * config/aarch64/iterators.md (wv): New mode attribute.
7008 (vf, VF): New mode attributes.
7009 (vgp, VGP): New mode attributes.
7010 (s): Update attribute with SImode and DImode prefixes.
7012 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
7014 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
7015 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
7016 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
7017 (nios2_option_overide): Initialize it. Don't allow R0-relative
7018 addressing with PIC.
7019 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
7020 (nios2_symbolic_constant_p): Likewise.
7021 (nios2_legitimate_address_p): Likewise.
7022 (nios2_r0rel_section_name_p): New.
7023 (nios2_symbol_ref_in_r0rel_data_p): New.
7024 (nios2_emit_move_sequence): Handle r0rel_constant_p.
7025 (r0rel_constant_p): New.
7026 (nios2_print_operand_address): Handle r0rel_constant_p.
7027 (nios2_cdx_narrow_form_p): Likewise.
7028 * config/nios2/nios2.opt (mr0rel-sec=): New option.
7029 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
7030 (Nios II Options): Document -mr0rel-sec.
7032 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
7034 * config/nios2/nios2.c: Include xregex.h.
7035 (nios2_gprel_sec_regex): New.
7036 (nios2_option_overide): Initialize it. Don't allow GP-relative
7037 addressing with PIC.
7038 (nios2_small_section_name_p): Check for regex match.
7039 * config/nios2/nios2.opt (mgprel-sec=): New option.
7040 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
7041 (Nios II Options): Document -mgprel-sec.
7043 2017-10-26 Jim Wilson <wilson@tuliptree.org>
7045 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
7047 2017-10-26 Tom de Vries <tom@codesourcery.com>
7049 PR tree-optimization/82707
7050 * gimple.c (gimple_copy): Fix unsharing of
7051 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
7053 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
7055 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
7056 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
7057 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
7058 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
7059 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
7060 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
7061 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
7062 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
7063 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
7064 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
7065 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
7066 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
7067 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
7068 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
7069 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
7070 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
7071 _mm512_mask_cmpunord_ps_mask): New intrinsics.
7073 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7075 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
7077 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
7078 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
7079 warning to rs6000.c. Remove the Undocumented flag, since it has
7081 (-mabi=ibmlongdouble): Likewise.
7082 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
7083 already set, set the default format for long double.
7084 (rs6000_debug_reg_global): Print whether long double is IBM or
7086 (rs6000_option_override_internal): Rework setting long double
7087 format. Only warn if the user is changing the long double default
7088 and they did not use -Wno-psabi.
7089 * doc/invoke.texi (PowerPC options): Update the documentation for
7090 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
7092 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7093 Alan Hayward <alan.hayward@arm.com>
7094 David Sherwood <david.sherwood@arm.com>
7096 * rtl.h (wider_subreg_mode): New function.
7097 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
7098 rather than an unsigned int *.
7099 * ira-color.c (regno_max_ref_width): Replace with...
7100 (regno_max_ref_mode): ...this new variable.
7101 (coalesced_pseudo_reg_slot_compare): Update accordingly.
7102 Use wider_subreg_mode.
7103 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
7104 rather than an unsigned int *.
7105 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
7106 (process_alt_operands): Likewise.
7107 (invariant_p): Likewise.
7108 * lra-spills.c (assign_mem_slot): Likewise.
7109 (add_pseudo_to_slot): Likewise.
7110 * lra.c (collect_non_operand_hard_regs): Likewise.
7111 (add_regs_to_insn_regno_info): Likewise.
7112 * reload1.c (regno_max_ref_width): Replace with...
7113 (regno_max_ref_mode): ...this new variable.
7114 (reload): Update accordingly. Update call to
7115 ira_sort_regnos_for_alter_reg.
7116 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
7117 (init_eliminable_invariants): Update to use regno_max_ref_mode.
7118 (scan_paradoxical_subregs): Likewise.
7120 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7122 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
7123 (aarch64_frame): Add emit_frame_chain boolean.
7124 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7125 Move eh_return case to aarch64_layout_frame.
7126 (aarch64_layout_frame): Initialize emit_frame_chain.
7127 (aarch64_expand_prologue): Use emit_frame_chain.
7129 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7131 * config/aarch64/aarch64.c (aarch64_layout_frame):
7132 Ensure LR is always stored at the bottom of the callee-saves.
7133 Remove rarely used frame layout which saves callee-saves at top of
7134 frame, so the store of LR can be used as a valid probe in all cases.
7136 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7138 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
7139 Improve unaligned TImode/TFmode base/offset split.
7141 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7142 Alan Hayward <alan.hayward@arm.com>
7143 David Sherwood <david.sherwood@arm.com>
7145 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
7146 * combine.c (find_single_use_1): Likewise.
7147 (expand_field_assignment): Likewise.
7148 (move_deaths): Likewise.
7149 * lra-constraints.c (simplify_operand_subreg): Likewise.
7150 (curr_insn_transform): Likewise.
7151 * lra.c (collect_non_operand_hard_regs): Likewise.
7152 (add_regs_to_insn_regno_info): Likewise.
7153 * rtlanal.c (reg_referenced_p): Likewise.
7154 (covers_regno_no_parallel_p): Likewise.
7156 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7158 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
7159 Don't print any bits outside the precision of the value.
7160 * wide-int.cc (test_printing): Add some new tests.
7162 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7164 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
7165 supports -xbrace_comment option.
7166 * configure: Regenerate.
7167 * config.in: Regenerate.
7168 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
7169 (ASM_CPU_SPEC): Use it.
7171 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7173 * target.def (static_rtx_alignment): New hook.
7174 * targhooks.h (default_static_rtx_alignment): Declare.
7175 * targhooks.c (default_static_rtx_alignment): New function.
7176 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
7177 * doc/tm.texi: Regenerate.
7178 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
7179 instead of targetm.constant_alignment. Remove call to
7181 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7182 (cris_preferred_mininum_alignment): New function, split out from...
7183 (cris_constant_alignment): ...here.
7184 (cris_static_rtx_alignment): New function.
7185 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
7187 (ix86_constant_alignment): ...here.
7188 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7189 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7190 (mmix_static_rtx_alignment): New function.
7191 * config/spu/spu.c (spu_static_rtx_alignment): New function.
7192 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7194 2017-10-26 Tamar Christina <tamar.christina@arm.com>
7197 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
7198 Add flag_trapping_math and flag_fp_int_builtin_inexact.
7200 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
7202 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
7203 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
7204 mark as a sign-extending load.
7205 (local_pic_load_u): Define.
7207 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
7210 * fold-const.c (operand_equal_for_comparison_p): Also return true
7211 if ARG0 is a simple variant of ARG1 with narrower precision.
7212 (fold_ternary_loc): Always pass unstripped operands to the predicate.
7214 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
7216 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
7218 * i386.h (processor_costs): Add gather_static, gather_per_elt,
7219 scatter_static, scatter_per_elt.
7220 * x86-tune-costs.h: Add new cost entries.
7222 2017-10-25 Richard Biener <rguenther@suse.de>
7224 * tree-ssa-sccvn.h (vn_eliminate): Declare.
7225 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
7226 class pass_fre): Move to ...
7227 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
7228 class pass_fre): ... here and adjust for statistics.
7230 2017-10-25 Jakub Jelinek <jakub@redhat.com>
7233 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
7234 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
7235 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
7236 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
7239 2017-10-25 Richard Biener <rguenther@suse.de>
7241 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
7242 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
7243 eliminate_push_avail, eliminate_insert): Move inside...
7244 (class eliminate_dom_walker): ... this class in preparation
7246 (fini_eliminate): Remove by merging with ...
7247 (eliminate): ... this function. Adjust for class changes.
7248 (pass_pre::execute): Remove fini_eliminate call.
7249 (pass_fre::execute): Likewise.
7251 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7254 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
7255 (VPERMI2, VPERMI2I): New mode iterators.
7256 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
7257 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
7259 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
7260 mode iterator. Remove 3 old define_insn patterns.
7261 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
7262 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
7263 VPERMI2 mode iterator, remove the other two expanders.
7264 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
7265 to use VPERMI2 mode iterator, add another alternative for vpermi2*
7266 instructions, remove the other two patterns.
7267 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
7268 mode iterator, remove the other two patterns.
7269 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
7270 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
7271 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
7272 and adjust argument order accordingly.
7273 (ix86_expand_vec_perm): Adjust caller.
7274 (expand_vec_perm_1): Likewise.
7275 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
7276 (expand_vec_perm_vpermt2_vpshub2): ... this.
7277 (ix86_expand_vec_perm_const_1): Adjust caller.
7278 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
7281 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
7282 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
7283 (vec_shl_<mode>): Remove unused expander.
7284 (avx512bw_<shift_insn><mode>3): New define_insn.
7285 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
7286 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
7288 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
7291 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
7294 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7296 PR rtl-optimization/82396
7297 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
7298 (autopref_multipass_init): Simplify initialization.
7299 (autopref_rank_data): Simplify sort order.
7300 * gcc/sched-int.h (autopref_multipass_data_): Remove
7301 multi_mem_insn_p, min_offset and max_offset.
7303 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7306 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7307 Check special value of flag_omit_frame_pointer.
7308 (aarch64_can_eliminate): Likewise.
7309 (aarch64_override_options_after_change_1): Simplify handling of
7310 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
7312 2017-10-24 Richard Biener <rguenther@suse.de>
7314 PR tree-optimization/82697
7315 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
7316 zero for conditional load and unconditional store.
7318 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7320 * doc/install.texi: Document bootstrap-cet.
7322 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7325 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
7326 ENDBR instruction at function entrance if function is only
7329 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7332 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
7333 patterns to better describe from which operation the CF is computed.
7334 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
7335 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
7336 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
7337 is 0, use _0 suffixed expanders instead of emitting a comparison
7340 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
7342 * config/i386/i386.md(*movsf_internal, *movdf_internal):
7343 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
7345 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
7348 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
7349 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
7350 * loop-iv.c (iv_get_reaching_def): Likewise.
7351 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
7352 variable is promoted and the partition contains undefined values.
7354 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7356 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
7358 (nios2_address_cost): Define.
7359 (nios2_legitimize_address): Recognize (exp + constant) directly.
7360 (TARGET_ADDRESS_COST): Define.
7362 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7364 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
7365 (nios2_symbolic_memory_operand_p): Declare.
7366 (nios2_split_large_constant): Declare.
7367 (nios2_split_symbolic_memory_operand): Declare.
7368 * config/nios2/nios2.c: Adjust includes.
7369 (nios2_symbolic_constant_allowed): New.
7370 (nios2_symbolic_constant_p): New.
7371 (nios2_plus_symbolic_constant_p): New.
7372 (nios2_valid_addr_expr_p): Recognize addresses involving
7374 (nios2_legitimate_address_p): Likewise, also LO_SUM.
7375 (nios2_symbolic_memory_operand_p): New.
7376 (nios2_large_constant_p): New.
7377 (nios2_split_large_constant): New.
7378 (nios2_split_plus_large_constant): New.
7379 (nios2_split_symbolic_memory_operand): New.
7380 (nios2_legitimize_address): Code refactoring. Handle addresses
7381 involving symbolic constants.
7382 (nios2_emit_move_sequence): Likewise.
7383 (nios2_print_operand): Improve error output.
7384 (nios2_print_operand_address): Handle LO_SUM.
7385 (nios2_cdx_narrow_form_p): Likewise.
7386 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
7387 operands involving symbolic constants.
7388 (movhi_internal, movsi_internal): Likewise.
7389 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
7390 (extendhisi2, extendqi<mode>2): Likewise.
7392 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7394 * tree-pass.h (PROP_rtl_split_insns): Define.
7395 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
7397 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7399 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
7401 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7404 * target.def (const_not_ok_for_debug_p): Default to
7405 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
7406 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
7407 * targhooks.c (default_const_not_ok_for_debug_p): New function.
7408 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
7409 which targetm.const_not_ok_for_debug_p returned true.
7410 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
7412 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
7414 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
7415 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
7416 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
7418 (ix86_const_not_ok_for_debug_p): New function.
7419 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
7420 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
7422 2017-10-23 David Malcolm <dmalcolm@redhat.com>
7425 * system.h: Conditionally include "unique-ptr.h" if
7426 INCLUDE_UNIQUE_PTR is defined.
7427 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
7428 of defining INCLUDE_UNIQUE_PTR before including "system.h".
7430 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
7432 * config/rl78/rl78.md: New define_expand "subdi3".
7434 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
7437 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
7438 DF_REF_INSN if DF_REF_INSN_INFO is false.
7440 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
7442 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
7443 xmm_move instead of sse_move.
7444 (sse_store_index): New function.
7445 (ix86_register_move_cost): Be more sensible about mismatch stall;
7446 model AVX moves correctly; make difference between sse->integer and
7448 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
7449 moves; make difference between SSE and AVX.
7450 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
7451 and zmm_move. Increase size of sse load and store tables;
7452 add unaligned load and store tables; add ssemmx_to_integer.
7453 * x86-tune-costs.h: Update all entries according to real
7454 move latencies from Agner Fog's manual and chip documentation.
7456 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7459 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
7460 * config/i386/constraints.md (Wf): New constraint.
7461 * config/i386/i386.md (UNSPEC_SBB): New unspec.
7462 (cmp<dwi>_doubleword): Removed.
7463 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
7464 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
7465 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
7466 expand with cmp<dwi>_doubleword. For LTU and GEU use
7467 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
7469 * common.opt (gcolumn-info): Enable by default.
7470 * doc/invoke.texi (gcolumn-info): Document new default.
7472 2017-10-23 Richard Biener <rguenther@suse.de>
7474 PR tree-optimization/82672
7475 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
7476 Fold the stmt if we propagated into it.
7478 2017-10-23 Richard Biener <rguenther@suse.de>
7480 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
7481 (bitmap_remove_expr_from_set): ... this. All callers call this
7482 for non-constant values.
7483 (bitmap_set_subtract): Rename to...
7484 (bitmap_set_subtract_expressions): ... this. Adjust and
7486 (bitmap_set_contains_value): Remove superfluous check.
7487 (bitmap_set_replace_value): Inline into single caller ...
7488 (bitmap_value_replace_in_set): ... here and simplify.
7489 (dependent_clean): Merge into ...
7490 (clean): ... this using an overload. Adjust.
7491 (prune_clobbered_mems): Adjust.
7492 (compute_antic_aux): Likewise.
7493 (compute_partial_antic_aux): Likewise.
7495 2017-10-23 Richard Biener <rguenther@suse.de>
7497 PR tree-optimization/82129
7499 2017-08-01 Richard Biener <rguenther@suse.de>
7501 PR tree-optimization/81181
7502 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7503 (compute_antic): ... end of iteration here.
7505 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7507 * target.def (starting_frame_offset): New hook.
7508 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
7509 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
7510 * doc/tm.texi.in: Regenerate.
7511 * hooks.h (hook_hwi_void_0): Declare.
7512 * hooks.c (hook_hwi_void_0): New function.
7513 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
7514 STARTING_FRAME_OFFSET.
7515 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
7516 * reload1.c (reload): Likewise.
7517 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
7518 instead of STARTING_FRAME_OFFSET.
7519 * function.c (try_fit_stack_local): Likewise.
7520 (assign_stack_local_1): Likewise
7521 (instantiate_virtual_regs): Likewise.
7522 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
7523 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
7524 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
7525 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
7526 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
7527 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
7528 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
7529 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
7530 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
7531 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
7532 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
7533 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
7534 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
7535 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
7536 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
7537 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
7538 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
7539 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
7540 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
7541 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
7542 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
7543 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
7544 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
7545 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
7546 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
7547 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
7548 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
7549 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
7550 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
7551 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
7552 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
7553 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
7554 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
7555 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
7556 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
7557 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
7558 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
7559 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
7560 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
7561 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
7562 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
7563 * config/avr/avr.c (avr_starting_frame_offset): Make static and
7564 return a HOST_WIDE_INT.
7565 (avr_builtin_setjmp_frame_value): Use it instead of
7566 STARTING_FRAME_OFFSET.
7567 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7568 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
7569 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
7571 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7572 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
7573 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
7574 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7575 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
7576 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
7577 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7578 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
7579 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
7580 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7581 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
7582 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
7584 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7585 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
7586 * config/mips/mips.c (mips_compute_frame_info): Refer to
7587 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
7588 (mips_starting_frame_offset): New function.
7589 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7590 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
7591 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
7592 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
7593 and return a HOST_WIDE_INT.
7594 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7595 (mmix_initial_elimination_offset): Refer to
7596 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
7597 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
7598 * config/pa/pa.c (pa_starting_frame_offset): New function.
7599 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
7600 (pa_expand_prologue): Likewise.
7601 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7602 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
7603 !FRAME_GROWS_DOWNWARD handling to...
7604 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7605 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
7606 !FRAME_GROWS_DOWNWARD handling to...
7607 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7608 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
7609 !FRAME_GROWS_DOWNWARD handling to...
7610 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7611 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
7613 (rs6000_starting_frame_offset): New function.
7614 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
7615 !FRAME_GROWS_DOWNWARD handling to...
7616 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7617 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
7618 !FRAME_GROWS_DOWNWARD handling to...
7619 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7620 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
7621 !FRAME_GROWS_DOWNWARD handling to...
7622 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7623 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
7624 (rs6000_starting_frame_offset): New function.
7625 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
7626 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
7627 * config/vax/vax.c (vax_starting_frame_offset): New function.
7628 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
7629 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7630 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
7631 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
7632 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7633 * system.h (STARTING_FRAME_OFFSET): Poison.
7635 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7637 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
7638 SCALAR_TYPE_MODE instead of TYPE_MODE.
7640 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7641 Alan Hayward <alan.hayward@arm.com>
7642 David Sherwood <david.sherwood@arm.com>
7644 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
7646 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7647 Alan Hayward <alan.hayward@arm.com>
7648 David Sherwood <david.sherwood@arm.com>
7650 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
7652 2017-10-23 Richard Biener <rguenther@suse.de>
7654 PR tree-optimization/82129
7655 * tree-ssa-pre.c (bitmap_set_and): Remove.
7656 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
7657 canonicalizing expressions in the set to those with lowest
7658 ID rather than taking that from the first edge.
7660 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7662 * combine.c (rtx_equal_for_field_assignment_p): Use
7663 byte_lowpart_offset.
7665 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7666 Alan Hayward <alan.hayward@arm.com>
7667 David Sherwood <david.sherwood@arm.com>
7669 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
7670 to a SUBREG_PROMOTED_VAR.
7672 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7673 Alan Hayward <alan.hayward@arm.com>
7674 David Sherwood <david.sherwood@arm.com>
7676 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
7677 (expand_debug_source_expr): Likewise.
7678 * combine.c (combine_simplify_rtx): Likewise.
7679 * cse.c (fold_rtx): Likewise.
7680 * optabs.c (expand_float): Likewise.
7681 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7682 (simplify_binary_operation_1): Likewise.
7684 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7685 Alan Hayward <alan.hayward@arm.com>
7686 David Sherwood <david.sherwood@arm.com>
7688 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
7689 (record_promoted_value): Likewise.
7690 * expr.c (expand_expr_real_2): Likewise.
7691 * ree.c (update_reg_equal_equiv_notes): Likewise.
7692 (combine_set_extension): Likewise.
7693 * rtlanal.c (low_bitmask_len): Likewise.
7694 * simplify-rtx.c (neg_const_int): Likewise.
7695 (simplify_binary_operation_1): Likewise.
7697 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7698 Alan Hayward <alan.hayward@arm.com>
7699 David Sherwood <david.sherwood@arm.com>
7701 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
7702 * regcprop.c (maybe_mode_change): Likewise.
7703 * reload1.c (alter_reg): Likewise.
7705 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7707 * inchash.h (inchash::hash::add_wide_int): New function.
7708 * lto-streamer-out.c (hash_tree): Use it.
7710 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7712 * inchash.h (inchash::hash::add_wide_int): Rename to...
7713 (inchash::hash::add_hwi): ...this.
7714 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
7715 (polymorphic_call_target_hasher::hash): Likewise.
7716 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
7717 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
7718 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
7719 * lto-streamer-out.c (hash_tree): Likewise.
7720 * optc-save-gen.awk: Likewise.
7721 * tree.c (add_expr): Likewise.
7723 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
7726 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
7727 for ordered inequality comparisons even with TARGET_IEEE_FP.
7729 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
7732 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
7733 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
7734 Expand with cmp<dwi>_doubleword.
7736 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7738 * extend.texi: Add x86 specific to 'nocf_check' attribute.
7739 List CET intrinsics.
7740 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
7741 specific to -fcf-protection option.
7743 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7745 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
7746 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
7747 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
7748 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
7749 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
7750 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
7751 (extra_objs): Add cet.o for Linux/x86 targets.
7752 (tmake_file): Add i386/t-cet for Linux/x86 targets.
7753 * config/i386/cet.c: New file.
7754 * config/i386/cetintrin.h: Likewise.
7755 * config/i386/t-cet: Likewise.
7756 * config/i386/cpuid.h (bit_SHSTK): New.
7757 (bit_IBT): Likewise.
7758 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
7759 pass IBT and SHSTK bits.
7760 * config/i386/i386-builtin-types.def
7761 (VOID_FTYPE_UNSIGNED_PVOID): New.
7762 (VOID_FTYPE_UINT64_PVOID): Likewise.
7763 * config/i386/i386-builtin.def: Add CET intrinsics.
7764 * config/i386/i386-c.c (ix86_target_macros_internal): Add
7765 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
7766 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
7767 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
7769 * config/i386/i386.c (rest_of_insert_endbranch): New.
7770 (pass_data_insert_endbranch): Likewise.
7771 (pass_insert_endbranch): Likewise.
7772 (make_pass_insert_endbranch): Likewise.
7773 (ix86_notrack_prefixed_insn_p): Likewise.
7774 (ix86_target_string): Add -mibt, -mshstk flags.
7775 (ix86_option_override_internal): Add flag_cf_protection
7777 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
7778 (ix86_print_operand): Add 'notrack' prefix output.
7779 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
7780 (ix86_expand_builtin): Expand CET intrinsics.
7781 (x86_output_mi_thunk): Add 'endbranch' instruction.
7782 * config/i386/i386.h (TARGET_IBT): New.
7783 (TARGET_IBT_P): Likewise.
7784 (TARGET_SHSTK): Likewise.
7785 (TARGET_SHSTK_P): Likewise.
7786 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
7787 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
7788 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
7789 (builtin_setjmp_setup): New pattern.
7790 (builtin_longjmp): Likewise.
7791 (rdssp<mode>): Likewise.
7792 (incssp<mode>): Likewise.
7793 (saveprevssp): Likewise.
7794 (rstorssp): Likewise.
7795 (wrss<mode>): Likewise.
7796 (wruss<mode>): Likewise.
7797 (setssbsy): Likewise.
7798 (clrssbsy): Likewise.
7799 (nop_endbr): Likewise.
7800 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
7802 * config/i386/immintrin.h: Include <cetintrin.h>.
7803 * config/i386/linux-common.h
7804 (file_end_indicate_exec_stack_and_cet): New prototype.
7805 (TARGET_ASM_FILE_END): New.
7807 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7809 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
7810 latencies instead of having separate table; make difference between
7811 integer and float costs.
7812 * i386.h (processor_costs): Remove scalar_stmt_cost,
7813 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
7814 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
7816 * x86-tune-costs.h: Remove entries which has been removed in
7817 procesor_costs from all tables; make cond_taken_branch_cost
7818 and cond_not_taken_branch_cost COST_N_INSNS based.
7820 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7822 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
7824 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7826 * config/i386/i386.md (isa): Remove fma_avx512f.
7827 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
7828 <avx512>_fmadd_<mode>_mask3<round_name>,
7829 <avx512>_fmsub_<mode>_mask<round_name>,
7830 <avx512>_fmsub_<mode>_mask3<round_name>,
7831 <avx512>_fnmadd_<mode>_mask<round_name>,
7832 <avx512>_fnmadd_<mode>_mask3<round_name>,
7833 <avx512>_fnmsub_<mode>_mask<round_name>,
7834 <avx512>_fnmsub_<mode>_mask3<round_name>,
7835 <avx512>_fmaddsub_<mode>_mask<round_name>,
7836 <avx512>_fmaddsub_<mode>_mask3<round_name>,
7837 <avx512>_fmsubadd_<mode>_mask<round_name>,
7838 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
7839 (*vec_widen_umult_even_v16si<mask_name>,
7840 *vec_widen_smult_even_v16si<mask_name>): Likewise.
7841 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
7843 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7845 * extend.texi: Add 'nocf_check' documentation.
7846 * gimple.texi: Add second parameter to
7847 gimple_build_call_from_tree.
7848 * invoke.texi: Add -fcf-protection documentation.
7849 * rtl.texi: Add REG_CALL_NOTRACK documenation.
7851 2017-10-20 Richard Biener <rguenther@suse.de>
7853 PR tree-optimization/82473
7854 * tree-vect-loop.c (vectorizable_reduction): Properly get at
7855 the largest input type.
7857 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7859 * c-attribs.c (handle_nocf_check_attribute): New function.
7860 (c_common_attribute_table): Add 'nocf_check' handling.
7861 * gimple-parser.c: Add second argument NULL to
7862 gimple_build_call_from_tree.
7863 * attrib.c (comp_type_attributes): Check nocf_check attribute.
7864 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
7866 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
7867 * common.opt: Add fcf-protection flag.
7868 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
7869 * flag-types.h: Add enum cf_protection_level.
7870 * gimple.c (gimple_build_call_from_tree): Add second parameter.
7871 Add 'nocf_check' attribute propagation to gimple call.
7872 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
7873 (gimple_build_call_from_tree): Update prototype.
7874 (gimple_call_nocf_check_p): New function.
7875 (gimple_call_set_nocf_check): Likewise.
7876 * gimplify.c: Add second argument to gimple_build_call_from_tree.
7877 * ipa-icf.c: Add nocf_check attribute in statement hash.
7878 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
7879 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
7880 * toplev.c (process_options): Add flag_cf_protection handling.
7882 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7884 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
7886 2017-10-20 Richard Biener <rguenther@suse.de>
7888 PR tree-optimization/82603
7889 * tree-if-conv.c (predicate_mem_writes): Make sure to only
7890 remove false predicated stores.
7892 2017-10-20 Richard Biener <rguenther@suse.de>
7894 * graphite-isl-ast-to-gimple.c
7895 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
7896 Remove return value and simplify, dump copied stmt after lhs
7898 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
7899 Reduce dump verbosity.
7900 (gsi_insert_earliest): Likewise.
7901 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
7902 * graphite.c (print_global_statistics): Adjust dumping.
7903 (print_graphite_scop_statistics): Likewise.
7904 (print_graphite_statistics): Do not dump loops here.
7905 (graphite_transform_loops): But here.
7907 2017-10-20 Nicolas Roche <roche@adacore.com>
7909 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
7910 * configure: Regenerate.
7912 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7915 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
7916 functions when optimizing replace GIMPLE_RETURN stmts with
7917 calls to __builtin_unreachable ().
7920 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
7921 for -fsanitize=thread link of executables.
7922 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
7923 link of executables.
7926 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
7928 (<shift_insn><mode>3<mask_name>): Change the last of the 3
7929 define_insns for logical vector shifts to use VI248_AVX512BW
7930 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
7931 condition, useless isa and prefix attributes. Change the first
7932 2 of these define_insns to ...
7933 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
7934 define_insn for avx512vl.
7935 (<shift_insn><mode>3): ... and this, new define_insn without
7936 masking for non-avx512vl.
7939 * config/i386/sse.md (*andnot<mode>3,
7940 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
7941 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
7942 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
7943 not applied use empty suffix even for TARGET_AVX512VL.
7944 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
7945 is applied, supply evex,evex or evex,evex,evex instead of just
7948 2017-10-20 Julia Koval <julia.koval@intel.com>
7950 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
7951 (OPTION_MASK_ISA_GFNI_UNSET): New.
7952 (ix86_handle_option): Handle OPT_mgfni.
7953 * config/i386/cpuid.h (bit_GFNI): New.
7954 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
7955 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
7956 * config/i386/i386.c (ix86_target_string): Add -mgfni.
7957 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
7958 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
7959 * config/i386/i386.opt: Add mgfni.
7961 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
7963 * config/msp430/msp430.c (msp430_option_override): Disable
7964 -fdelete-null-pointer-checks.
7965 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
7967 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7969 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
7970 of x87 and SSE instructions.
7972 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7974 * asan.c (create_cond_insert_point): Do not update edge count.
7975 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
7976 (afdo_propagate_circuit): Likewise.
7977 (afdo_calculate_branch_prob): Likewise.
7978 (afdo_annotate_cfg): Likewise.
7979 * basic-block.h (struct edge_def): Remove count.
7980 (edge_def::count): New accessor.
7981 * bb-reorder.c (rotate_loop): Update.
7982 (find_traces_1_round): Update.
7983 (connect_traces): Update.
7984 (sanitize_hot_paths): Update.
7985 * cfg.c (unchecked_make_edge): Update.
7986 (make_single_succ_edge): Update.
7987 (check_bb_profile): Update.
7988 (dump_edge_info): Update.
7989 (update_bb_profile_for_threading): Update.
7990 (scale_bbs_frequencies_int): Update.
7991 (scale_bbs_frequencies_gcov_type): Update.
7992 (scale_bbs_frequencies_profile_count): Update.
7993 (scale_bbs_frequencies): Update.
7994 * cfganal.c (connect_infinite_loops_to_exit): Update.
7995 * cfgbuild.c (compute_outgoing_frequencies): Update.
7996 (find_many_sub_basic_blocks): Update.
7997 * cfgcleanup.c (try_forward_edges): Update.
7998 (try_crossjump_to_edge): Update
7999 * cfgexpand.c (expand_gimple_cond): Update
8000 (expand_gimple_tailcall): Update
8001 (construct_exit_block): Update
8002 * cfghooks.c (verify_flow_info): Update
8003 (redirect_edge_succ_nodup): Update
8004 (split_edge): Update
8005 (make_forwarder_block): Update
8006 (duplicate_block): Update
8007 (account_profile_record): Update
8008 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
8009 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
8010 * cfgloopmanip.c (scale_loop_profile): Update.
8012 (lv_adjust_loop_entry_edge): Update.
8013 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
8014 (force_nonfallthru_and_redirect): Update.
8015 (purge_dead_edges): Update.
8016 (rtl_flow_call_edges_add): Update.
8017 * cgraphunit.c (init_lowered_empty_function): Update.
8018 (cgraph_node::expand_thunk): Update.
8019 * gimple-pretty-print.c (dump_probability): Update.
8020 (dump_edge_probability): Update.
8021 * gimple-ssa-isolate-paths.c (isolate_path): Update.
8022 * haifa-sched.c (sched_create_recovery_edges): Update.
8023 * hsa-gen.c (convert_switch_statements): Update.
8024 * ifcvt.c (dead_or_predicable): Update.
8025 * ipa-inline-transform.c (inline_transform): Update.
8026 * ipa-split.c (split_function): Update.
8027 * ipa-utils.c (ipa_merge_profiles): Update.
8028 * loop-doloop.c (add_test): Update.
8029 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
8030 * lto-streamer-in.c (input_cfg): Update.
8031 (input_function): Update.
8032 * lto-streamer-out.c (output_cfg): Update.
8033 * modulo-sched.c (sms_schedule): Update.
8034 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
8035 * predict.c (maybe_hot_edge_p): Update.
8036 (unlikely_executed_edge_p): Update.
8037 (probably_never_executed_edge_p): Update.
8038 (dump_prediction): Update.
8039 (drop_profile): Update.
8040 (propagate_unlikely_bbs_forward): Update.
8041 (determine_unlikely_bbs): Update.
8042 (force_edge_cold): Update.
8043 * profile.c (compute_branch_probabilities): Update.
8044 * reg-stack.c (better_edge): Update.
8045 * shrink-wrap.c (handle_simple_exit): Update.
8046 * tracer.c (better_p): Update.
8047 * trans-mem.c (expand_transaction): Update.
8048 (split_bb_make_tm_edge): Update.
8049 * tree-call-cdce.c: Update.
8050 * tree-cfg.c (gimple_find_sub_bbs): Update.
8051 (gimple_split_edge): Update.
8052 (gimple_duplicate_sese_region): Update.
8053 (gimple_duplicate_sese_tail): Update.
8054 (gimple_flow_call_edges_add): Update.
8055 (insert_cond_bb): Update.
8056 (execute_fixup_cfg): Update.
8057 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
8058 * tree-complex.c (expand_complex_div_wide): Update.
8059 * tree-eh.c (lower_resx): Update.
8060 (unsplit_eh): Update.
8061 (cleanup_empty_eh_move_lp): Update.
8062 * tree-inline.c (copy_edges_for_bb): Update.
8063 (freqs_to_counts): Update.
8064 (copy_cfg_body): Update.
8065 * tree-ssa-dce.c (remove_dead_stmt): Update.
8066 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
8067 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
8068 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
8069 (unloop_loops): Update.
8070 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8071 * tree-ssa-loop-split.c (connect_loops): Update.
8072 (split_loop): Update.
8073 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
8074 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
8075 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
8076 * tree-ssa-reassoc.c (branch_fixup): Update.
8077 * tree-ssa-tail-merge.c (replace_block_by): Update.
8078 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
8079 (compute_path_counts): Update.
8080 (update_profile): Update.
8081 (recompute_probabilities): Update.
8082 (update_joiner_offpath_counts): Update.
8083 (estimated_freqs_path): Update.
8084 (freqs_to_counts_path): Update.
8085 (clear_counts_path): Update.
8086 (ssa_fix_duplicate_block_edges): Update.
8087 (duplicate_thread_path): Update.
8088 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
8089 (case_bit_test_cmp): Update.
8090 (collect_switch_conv_info): Update.
8091 (gen_inbound_check): Update.
8092 (do_jump_if_equal): Update.
8093 (emit_cmp_and_jump_insns): Update.
8094 * tree-tailcall.c (decrease_profile): Update.
8095 (eliminate_tail_call): Update.
8096 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
8097 (vect_do_peeling): Update.
8098 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8099 * ubsan.c (ubsan_expand_null_ifn): Update.
8100 (ubsan_expand_ptr_ifn): Update.
8101 * value-prof.c (gimple_divmod_fixed_value): Update.
8102 (gimple_mod_pow2): Update.
8103 (gimple_mod_subtract): Update.
8104 (gimple_ic): Update.
8105 (gimple_stringop_fixed_value): Update.
8107 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8110 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
8112 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
8114 PR rtl-optimization/82395
8115 * ira-color.c (allocno_priority_compare_func): Fix comparison step
8116 based on non_spilled_static_chain_regno_p.
8118 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8120 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
8121 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
8122 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
8124 2017-10-19 Martin Sebor <msebor@redhat.com>
8126 PR tree-optimization/82596
8127 * tree.c (array_at_struct_end_p): Handle STRING_CST.
8129 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8131 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
8132 (get_mem_refs_of_builtin_call): Likewise.
8133 * builtins.c (expand_builtin_apply): Adjust call to
8134 allocate_dynamic_stack_space.
8135 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
8136 the third argument to allocate_dynamic_stack_space, otherwise -1.
8137 (expand_builtin): Deal with all alloca variants.
8138 (is_inexpensive_builtin): Likewise.
8139 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
8140 * calls.c (special_function_p): Deal with all alloca variants.
8141 (initialize_argument_information): Adjust call to
8142 allocate_dynamic_stack_space.
8143 (expand_call): Likewise.
8144 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
8145 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
8146 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
8147 use it for the stack usage computation.
8148 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
8149 * function.c (gimplify_parameters): Call build_alloca_call_expr.
8150 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
8151 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
8152 (in_loop_p): Remove first argument and useless check.
8153 (pass_walloca::execute): Remove useless test and adjust call to above.
8154 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
8155 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
8156 (gimplify_call_expr): Deal with all alloca variants.
8157 * hsa-gen.c (gen_hsa_alloca): Likewise.
8158 (gen_hsa_insns_for_call): Likewise.
8159 * ipa-pure-const.c (special_builtin_state): Likewise.
8160 * tree-chkp.c (chkp_build_returned_bound): Likewise.
8161 * tree-object-size.c (alloc_object_size): Likewise.
8162 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
8163 (call_may_clobber_ref_p_1): Likewise.
8164 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
8165 (ccp_fold_stmt): Likewise.
8166 (optimize_stack_restore): Likewise.
8167 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
8168 (mark_all_reaching_defs_necessary_1): Likewise.
8169 (propagate_necessity): Likewise.
8170 (eliminate_unnecessary_stmts): Likewise.
8171 * tree.c (build_common_builtin_nodes): Build
8172 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
8173 (build_alloca_call_expr): New function.
8174 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
8175 (CASE_BUILT_IN_ALLOCA): Likewise.
8176 (build_alloca_call_expr): Declare.
8177 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
8179 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8182 * dwarf2out.c (new_die_raw): New static inline function.
8183 (new_die): Use it to create the DIE.
8184 (add_AT_external_die_ref): Likewise.
8185 (clone_die): Likewise.
8186 (clone_as_declaration): Likewise.
8187 (dwarf2out_vms_debug_main_pointer): Likewise.
8188 (base_type_die): Likewise. Remove early return for corner cases.
8189 Do not call add_pubtype on the DIE here.
8190 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
8191 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
8192 typedefs for base types with DW_AT_endianity. Make sure a DIE with
8193 native order exists for base types, attach the DIE manually and call
8194 add_pubtype on it. Do not equate a reverse order DIE to the type.
8196 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
8198 * config/arm/arm.c (align_ok_ldrd_strd): New function.
8199 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
8201 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
8203 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8205 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
8206 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
8207 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
8208 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
8209 * opts.c (sanitizer_opts): Add builtin.
8210 * ubsan.c (instrument_builtin): New function.
8211 (pass_ubsan::execute): Call it.
8212 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
8213 * doc/invoke.texi: Document -fsanitize=builtin.
8215 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
8216 builtins, store max (log2 (align), 0) into uchar field instead of
8217 align into uptr field.
8218 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
8219 store uchar 0 field instead of uptr 0 field.
8220 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
8221 instead of passing one address of struct with 2 locations pass
8222 two addresses of structs with 1 location each.
8223 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
8224 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
8225 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
8226 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
8227 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
8228 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
8229 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
8230 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
8232 2017-10-19 Martin Liska <mliska@suse.cz>
8235 * file-find.c (remove_prefix): Remove.
8236 * file-find.h (remove_prefix): Likewise.
8237 * gcc-ar.c: Remove smartness of lookup.
8239 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
8241 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
8242 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
8243 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
8245 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8248 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
8249 (setcc + and to xor + setcc): New peephole2.
8251 2017-10-19 Tom de Vries <tom@codesourcery.com>
8253 * doc/sourcebuild.texi (Test Directives, Variants of
8254 dg-require-support): Add dg-require-stack-size.
8256 2017-10-19 Martin Liska <mliska@suse.cz>
8259 * gimplify.c (gimplify_decl_expr): Do not instrument variables
8260 that have a large alignment.
8261 (gimplify_target_expr): Likewise.
8263 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
8265 PR rtl-optimization/82602
8266 * ira.c (rtx_moveable_p): Return false for volatile asm.
8268 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
8271 * config/i386/i386-modes.def (CCGZ): New CC mode.
8272 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
8273 * config/i386/predicates.md (ix86_comparison_operator):
8275 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
8276 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
8277 with double-word subtraction.
8278 (put_condition_code): Handle CCGZmode.
8280 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
8282 * wide-int.cc (debug (const wide_int &)): New.
8283 (debug (const wide_int *)): New.
8284 (debug (const widest_int &)): New.
8285 (debug (const widest_int *)): New.
8287 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
8290 * lra-constraints.c (curr_insn_transform): Use non-input operand
8291 instead of output one for matched reload.
8293 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8295 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
8296 (tree-ssa-loop-ivopts.h): New header file.
8297 (struct builtin_info): New fields.
8298 (classify_builtin_1): Compute and record base and offset parts for
8299 memset builtin partition by calling strip_offset.
8300 (offset_cmp, fuse_memset_builtins): New functions.
8301 (finalize_partitions): Fuse adjacent memset partitions by calling
8303 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
8304 Expose the interface.
8305 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
8307 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8309 PR tree-optimization/82574
8310 * tree-loop-distribution.c (find_single_drs): New parameter. Check
8311 that data reference must be executed exactly once per iteration
8312 against the outermost loop in nest.
8313 (classify_partition): Update call to above function.
8315 2017-10-18 Richard Biener <rguenther@suse.de>
8317 PR tree-optimization/82591
8318 * graphite.c (graphite_transform_loops): Move code gen message
8320 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8321 Here. Handle scop_to_isl_ast failing.
8322 (scop_to_isl_ast): Limit the number of ISL operations.
8324 2017-10-18 Richard Biener <rguenther@suse.de>
8326 * graphite-isl-ast-to-gimple.c
8327 (translate_isl_ast_to_gimple::set_rename): Simplify.
8328 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
8329 (graphite_copy_stmts_from_block): ... here.
8330 (copy_bb_and_scalar_dependences): Simplify.
8331 (add_parameters_to_ivs_params): Canonicalize.
8332 (generate_entry_out_of_ssa_copies): Simplify.
8333 * graphite-sese-to-poly.c (extract_affine_name): Simplify
8334 by passing in ISL dimension.
8335 (parameter_index_in_region_1): Rename to ...
8336 (parameter_index_in_region): ... this.
8337 (extract_affine): Adjust assert, pass down parameter index.
8338 (add_param_constraints): Use range-info when available.
8339 (build_scop_context): Adjust.
8340 * sese.c (new_sese_info): Adjust.
8341 (free_sese_info): Likewise.
8342 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
8343 Remove unused typedefs.
8344 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
8346 2017-10-18 Martin Liska <mliska@suse.cz>
8348 * combine.c (simplify_compare_const): Add gcc_fallthrough.
8350 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8352 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
8353 (s390_sched_init): Do not reset s390_sched_state if we entered the
8354 current basic block via a fallthru edge and all others are unlikely.
8356 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8358 * config/s390/s390.c (NUM_SIDES): New variable.
8359 (LONGRUNNING_THRESHOLD): New variable.
8360 (LATENCY_FACTOR): New variable.
8361 (s390_sched_score): Decrease score for long-running instructions on
8363 (s390_sched_variable_issue): Perform bookkeeping for long-running
8366 2017-10-18 Richard Biener <rguenther@suse.de>
8368 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
8369 Simplify with removal of the parameter rename map.
8370 (set_rename): Likewise.
8371 (should_copy_to_new_region): Likewise.
8372 (graphite_copy_stmts_from_block): Likewise.
8373 (copy_bb_and_scalar_dependences): Remove initialization of
8374 unused copied_bb_map.
8376 (copy_internal_parameters): Likewise.
8377 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
8378 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
8379 Use INTEGRAL_TYPE_P.
8380 (parameter_index_in_region_1): Rename to ...
8381 (assign_parameter_index_in_region): ... this. Assert we have
8382 a parameter we handle.
8383 (scan_tree_for_params): Adjust.
8384 * sese.h (parameter_rename_map_t): Remove.
8385 (struct sese_info_t): Remove unused parameter_rename_map and
8386 copied_bb_map members.
8387 * sese.c (new_sese_info): Adjust.
8388 (free_sese_info): Likewise.
8390 2017-10-18 Martin Liska <mliska@suse.cz>
8393 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
8394 on an abnormal edge.
8396 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
8398 * doc/invoke.texi (ffunction-sections and fdata-sections):
8401 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8403 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
8404 the use statement can throw internally.
8406 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8408 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
8409 any RTX present on the RHS of a SET.
8410 * compare-elim.c (try_eliminate_compare): Restore comment.
8412 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8414 * langhooks.h (struct lang_hooks): Document that tree_size langhook
8415 may be also called on tcc_type nodes.
8416 * langhooks.c (lhd_tree_size): Likewise.
8418 2017-10-17 David Malcolm <dmalcolm@redhat.com>
8420 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
8421 format_warning_at_substring.
8422 (maybe_warn): Convert source_range * param to a location_t. Pass
8423 UNKNOWN_LOCATION rather than NULL to fmtwarn.
8424 (format_directive): Remove code to extract source_ranges and
8425 source_range * in favor of just a location_t.
8426 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
8428 * substring-locations.c (format_warning_va): Convert
8429 source_range * param to a location_t.
8430 (format_warning_at_substring): Likewise.
8431 * substring-locations.h (format_warning_va): Likewise.
8432 (format_warning_at_substring): Likewise.
8434 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
8436 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
8438 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
8439 and scatter/gather ops.
8441 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
8442 vec_gather_load and vec_scatter_store.
8443 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
8444 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
8445 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
8446 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
8447 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
8448 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
8450 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
8452 * reg-stack.c (compare_for_stack_reg): Add bool argument.
8453 Detect FTST instruction and handle its register pops. Only pop
8454 second operand if can_pop_second_op is true.
8455 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
8456 set can_pop_second_op to false in the compare_for_stack_reg call.
8458 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
8459 output_fp_compare for stack register operands.
8460 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
8461 instructions here. Do not emit stack register pops here. Assert
8462 that FCOMPP pops next to top stack register. Rewrite function.
8464 2017-10-17 Nathan Sidwell <nathan@acm.org>
8467 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
8468 use DECL_ASSEMBLER_NAME_RAW.
8471 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
8474 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
8475 Wilco Dijkstra <wilco.dijkstra@arm.com>
8477 * builtins.c (expand_builtin_update_setjmp_buf): Add a
8478 converstion to Pmode from the buf_addr.
8480 2017-10-17 Richard Biener <rguenther@suse.de>
8482 * graphite-dependences.c (scop_get_reads_and_writes): Change
8483 output parameters to references.
8485 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
8487 PR 71026/tree-optimization
8488 * fold-const.c (distribute_real_division): Removed.
8489 (fold_binary_loc): Remove calls to distribute_real_divison.
8491 2017-10-17 Richard Biener <rguenther@suse.de>
8493 * graphite-scop-detection.c
8494 (scop_detection::stmt_has_simple_data_refs_p): Always use
8495 the full nest as region.
8496 (try_generate_gimple_bb): Likewise.
8497 * sese.c (scalar_evolution_in_region): Simplify now that
8498 SCEV can handle instantiation in regions.
8499 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
8500 in the non-loop part of a function if requested.
8502 2017-10-17 Richard Biener <rguenther@suse.de>
8504 PR tree-optimization/82563
8505 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
8507 (graphite_regenerate_ast_isl): Call it.
8508 * graphite-scop-detection.c (build_scops): Remove entry edge split.
8510 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8512 PR tree-optimization/82549
8513 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
8514 Formatting fixes. Instead of calling make_bit_field_ref with negative
8517 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
8519 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
8520 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
8521 _mm_maskz_reduce_ss): New.
8522 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
8523 __builtin_ia32_reducess_mask): Ditto..
8524 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
8525 * config/i386/sse.md (reduces<mode>): Renamed to ...
8526 (reduces<mode><mask_scalar_name>): ... this.
8527 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
8529 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
8530 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
8532 2017-10-16 David Malcolm <dmalcolm@redhat.com>
8534 * Makefile.in (OBJS): Add unique-ptr-tests.o.
8535 * selftest-run-tests.c (selftest::run_tests): Call
8536 selftest::unique_ptr_tests_cc_tests.
8537 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
8538 * unique-ptr-tests.cc: New file.
8540 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
8543 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8545 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8546 (make_hard_regno_born, make_hard_regno_dead): Update
8547 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
8549 2017-10-16 Jeff Law <law@redhat.com>
8551 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
8553 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
8555 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
8557 2017-10-16 Olivier Hainque <hainque@adacore.com>
8559 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
8560 with_cpu if we were configured for an e500v2 target cpu name.
8562 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
8564 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
8565 * doc/invoke.texi: Document +nodsp as a valid extension for
8568 2017-10-16 Martin Liska <mliska@suse.cz>
8570 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
8571 (test_set_range): Likewise.
8572 (test_range_functions): Rename to ...
8573 (test_bit_in_range): ... this.
8574 (sbitmap_c_tests): Add new test.
8576 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8578 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
8580 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
8581 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
8583 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8585 * config/aarch64/aarch64-builtins.c
8586 (aarch64_types_quadopu_lane_qualifiers): New.
8587 (TYPES_QUADOPU_LANE): New.
8588 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
8589 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
8590 (aarch64_<sur>dot_laneq<vsi2qi>): New.
8591 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
8592 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
8593 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
8594 (Vdottype, DOTPROD): New.
8595 (sur): Add SDOT and UDOT.
8597 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8599 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
8600 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
8601 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
8603 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
8604 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
8605 Enable TARGET_DOTPROD.
8606 (cortex-a75.cortex-a55): Likewise.
8607 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
8609 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8611 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
8612 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
8614 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
8616 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
8617 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
8618 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
8619 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
8620 * config/arm/types.md (neon_dot, neon_dot_q): New.
8621 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
8623 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8625 * config/arm/arm.h (TARGET_DOTPROD): New.
8626 * config/arm/arm.c (arm_arch_dotprod): New.
8627 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
8628 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
8629 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
8630 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
8631 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
8632 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
8633 * doc/invoke.texi (armv8.2-a): Document dotprod
8635 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
8637 * i386.c (ix86_vec_cost): New function.
8638 (ix86_rtx_costs): Handle vector operations better.
8639 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
8640 * x86-tune-costs.h: Add new costs to all tables.
8642 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
8644 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
8646 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
8647 divsd, sqrtss and sqrtsd
8648 * x86-tune-costs.h: Add new entries to all costs.
8649 (znver1_cost): Fix to match real instruction latencies.
8651 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8652 Michael Collison <michael.collison@arm.com>
8654 * compare-elim.c: Include emit-rtl.h.
8655 (can_merge_compare_into_arith): New function.
8656 (try_validate_parallel): Likewise.
8657 (try_merge_compare): Likewise.
8658 (try_eliminate_compare): Call the above when no previous clobber
8660 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
8663 2017-10-14 Jakub Jelinek <jakub@redhat.com>
8667 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
8668 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
8672 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
8673 to be any operand_equal_p operands. For & (B - 1) require
8674 B to be power of 2. Recognize
8675 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
8677 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8680 * optabs.c (expand_memory_blockage): Fix call of
8681 targetm.have_memory_blockage.
8683 2017-10-14 Jakub Jelinek <jakub@redhat.com>
8686 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
8687 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
8688 objects to extra_objs instead of overwriting it.
8690 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8692 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
8693 Use any_fp_register_operand as operand[3] predicate. Simplify
8694 equality test for operands[2] and operands[4] memory location.
8695 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
8696 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
8697 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
8698 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
8699 any_fp_register_operand as operand[1] predicate. Simplify
8700 equality test for operands[0] and operands[3] memory location.
8701 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
8702 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
8703 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
8705 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8707 * target-insns.def: Add memory_blockage.
8708 * optabs.c (expand_memory_blockage): New function.
8709 (expand_asm_memory_barrier): Rename ...
8710 (expand_asm_memory_blockage): ... to this.
8711 (expand_mem_thread_fence): Call expand_memory_blockage
8712 instead of expand_asm_memory_barrier.
8713 (expand_mem_singnal_fence): Ditto.
8714 (expand_atomic_load): Ditto.
8715 (expand_atomic_store): Ditto.
8716 * doc/md.texi (Standard Pattern Names For Generation):
8717 Document memory_blockage instruction pattern.
8719 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
8721 * config/rl78/rl78.c (rl78_emit_libcall): New function.
8722 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
8723 * config/rl78/rl78.md: New define_expand "adddi3".
8725 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
8727 * cfghooks.c (verify_flow_info): Disable check that all probabilities
8730 2017-10-13 Jeff Law <law@redhat.com>
8732 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
8734 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8737 * internal-fn.c (expand_mul_overflow): If both operands have
8738 the same highpart of -1 or 0 and the topmost bit of lowpart
8739 is different, overflow is if res <= 0 rather than res < 0.
8741 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
8743 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
8744 TARGET_P9_VECTOR code for unaligned_load case.
8746 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
8748 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
8750 2017-10-13 Nathan Sidwell <nathan@acm.org>
8752 * tree-core.h (tree_contains_struct): Make bool.
8753 * tree.c (tree_contains_struct): Likewise.
8754 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
8755 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
8756 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
8757 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
8758 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
8760 2017-10-13 Richard Biener <rguenther@suse.de>
8762 * graphite-isl-ast-to-gimple.c
8763 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
8764 parameters and dominance check.
8765 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
8766 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
8767 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
8768 Do not update SSA form here or do intermediate IL verification.
8769 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
8770 (graphite_initialize): Remove check on the number of loops in
8771 the function and inline into graphite_transform_loops.
8772 (graphite_finalize): Inline into graphite_transform_loops.
8773 (graphite_transform_loops): Perform SSA update and IL verification
8775 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
8777 2017-10-13 Richard Biener <rguenther@suse.de>
8779 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
8780 graphite_expression_type_precision): Avoid global constructor
8782 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
8783 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
8784 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
8785 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
8786 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
8787 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
8789 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
8792 * config/i386/i386.h (ix86_red_zone_size): New.
8793 * config/i386/i386.md (push peephole2s): Replace
8794 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
8796 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8797 Alan Hayward <alan.hayward@arm.com>
8798 David Sherwood <david.sherwood@arm.com>
8800 * combine.c (can_change_dest_mode): Reject changes in
8801 REGMODE_NATURAL_SIZE.
8803 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8804 Alan Hayward <alan.hayward@arm.com>
8805 David Sherwood <david.sherwood@arm.com>
8807 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
8808 (expand_debug_source_expr): Likewise.
8809 * combine.c (combine_simplify_rtx): Likewise.
8810 * cse.c (fold_rtx): Likewise.
8811 * fwprop.c (canonicalize_address): Likewise.
8812 * targhooks.c (default_shift_truncation_mask): Likewise.
8814 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8815 Alan Hayward <alan.hayward@arm.com>
8816 David Sherwood <david.sherwood@arm.com>
8818 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
8819 (widened_mode): Likewise.
8820 (expand_unop): Likewise.
8821 * ree.c (transform_ifelse): Likewise.
8822 (merge_def_and_ext): Likewise.
8823 (combine_reaching_defs): Likewise.
8824 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8826 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8827 Alan Hayward <alan.hayward@arm.com>
8828 David Sherwood <david.sherwood@arm.com>
8830 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
8831 * combine.c (gen_lowpart_for_combine): Likewise.
8832 * dwarf2out.c (rtl_for_decl_location): Likewise.
8833 * final.c (alter_subreg): Likewise.
8834 * rtlhooks.c (gen_lowpart_general): Likewise.
8835 (gen_lowpart_if_possible): Likewise.
8837 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8838 Alan Hayward <alan.hayward@arm.com>
8839 David Sherwood <david.sherwood@arm.com>
8841 * calls.c (expand_call): Use subreg_lowpart_offset.
8842 * cse.c (cse_insn): Likewise.
8843 * regcprop.c (copy_value): Likewise.
8844 (copyprop_hardreg_forward_1): Likewise.
8846 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8849 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
8850 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
8851 =Q constraints to +Q and into insn condition add check
8852 that operands[0] and operands[1] are equal.
8853 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
8854 =Q constraints to +Q and into insn condition add check
8855 that operands[0] is equal to either operands[1] or operands[2].
8858 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
8859 instead of handling MINUS_EXPR twice (once for each argument),
8860 canonicalize operand order and handle just once, use rtype where
8861 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
8864 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
8865 any values of __C while still being pattern recognizable as a simple
8868 2017-10-13 Richard Biener <rguenther@suse.de>
8870 PR tree-optimization/82451
8872 2017-10-02 Richard Biener <rguenther@suse.de>
8874 PR tree-optimization/82355
8875 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
8876 a mapping for the enclosing loop but avoid generating one for
8878 (copy_bb_and_scalar_dependences): Remove premature codegen
8879 error on PHIs in blocks duplicated into multiple places.
8880 * graphite-scop-detection.c
8881 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
8882 in the region use it as loop and nest to analyze the DR in.
8883 (try_generate_gimple_bb): Likewise.
8884 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
8885 (add_loop_constraints): For blocks in a loop not in the region
8886 create a dimension with a single iteration.
8887 * sese.h (gbb_loop_at_index): Remove assert.
8889 * cfgloop.c (loop_preheader_edge): For the loop tree root
8890 return the single successor of the entry block.
8891 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8892 Reset the SCEV hashtable and niters.
8893 * graphite-scop-detection.c
8894 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
8895 assert that we only have POLYNOMIAL_CHREC that vary in loops
8896 contained in the region.
8897 (scop_detection::graphite_can_represent_expr): Adjust.
8898 (scop_detection::stmt_has_simple_data_refs_p): For loops
8899 not in the region set loop to NULL. The nest is now the
8900 entry edge to the region.
8901 (try_generate_gimple_bb): Likewise.
8902 * sese.c (scalar_evolution_in_region): Adjust for
8903 instantiate_scev change.
8904 * tree-data-ref.h (graphite_find_data_references_in_stmt):
8905 Make nest parameter the edge into the region.
8906 (create_data_ref): Likewise.
8907 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
8908 entry edge into a region and adjust instantiate_scev calls.
8909 (create_data_ref): Likewise.
8910 (graphite_find_data_references_in_stmt): Likewise.
8911 (find_data_references_in_stmt): Pass the loop preheader edge
8912 from the nest argument.
8913 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
8914 parameter the edge into the region.
8915 (instantiate_parameters): Use the loop preheader edge as entry.
8916 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
8918 (get_instantiated_value_entry): Make instantiate_below parameter
8919 the edge into the region.
8920 (instantiate_scev_name): Likewise. Adjust dominance checks,
8921 when we cannot use loop-based instantiation instantiate by
8922 walking use-def chains.
8923 (instantiate_scev_poly): Adjust.
8924 (instantiate_scev_binary): Likewise.
8925 (instantiate_scev_convert): Likewise.
8926 (instantiate_scev_not): Likewise.
8927 (instantiate_array_ref): Remove.
8928 (instantiate_scev_3): Likewise.
8929 (instantiate_scev_2): Likewise.
8930 (instantiate_scev_1): Likewise.
8931 (instantiate_scev_r): Do not blindly handle N-operand trees.
8932 Do not instantiate array-refs. Handle all constants and invariants.
8933 (instantiate_scev): Make instantiate_below parameter
8934 the edge into the region.
8935 (resolve_mixers): Use the loop preheader edge for the region
8936 parameter to instantiate_scev_r.
8937 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
8939 2017-10-13 Richard Biener <rguenther@suse.de>
8941 PR tree-optimization/82525
8942 * graphite-isl-ast-to-gimple.c
8943 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
8945 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
8946 Fail code generation when we cannot represent the isl integer.
8947 (binary_op_to_tree): Elide modulo operations that are no-ops
8948 in the type we code generate. Remove now superfluous code
8951 2017-10-13 Richard Biener <rguenther@suse.de>
8953 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
8954 (scop_detection::harmful_loop_in_region): Remove premature
8955 IV type restriction.
8956 (scop_detection::graphite_can_represent_scev): We can handle
8957 pointer IVs just fine.
8959 2017-10-13 Alan Modra <amodra@gmail.com>
8961 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
8962 "Clobbers and Scratch Registers". Add paragraph on
8963 alternative to clobbers for scratch registers and OpenBLAS
8966 2017-10-13 Alan Modra <amodra@gmail.com>
8968 * doc/extend.texi (Clobbers): Correct vax example. Delete old
8969 example of a memory input for a string of known length. Move
8970 commentary out of table. Add a number of new examples
8971 covering array memory inputs.
8973 2017-10-12 Martin Liska <mliska@suse.cz>
8975 PR tree-optimization/82493
8976 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
8977 (test_range_functions): New function.
8978 (sbitmap_c_tests): Likewise.
8979 * selftest-run-tests.c (selftest::run_tests): Run new tests.
8980 * selftest.h (sbitmap_c_tests): New function.
8982 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
8984 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
8986 * config/rs6000/amo.h: Fix spacing issue.
8988 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8991 * config/i386/i386.md (*ashl<mode>3_mask_1,
8992 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
8993 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
8996 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
8998 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
8999 (profile_probability): Set max_probability
9000 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
9002 * profile-count.c (profile_probability::differs_from_p): Do not
9003 rely on max_probaiblity == 10000
9005 2017-10-12 Jeff Law <law@redhat.com>
9007 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
9010 2017-10-12 Martin Sebor <msebor@redhat.com>
9014 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
9015 (handle_alias_pairs): Call it.
9016 * common.opt (-Wattribute-alias): New option.
9017 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
9018 * doc/invoke.texi (-Wattribute-alias): Document.
9020 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
9023 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9025 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9027 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9028 (make_hard_regno_born, make_hard_regno_dead): Update
9029 bb_killed_pseudos and bb_gen_pseudos.
9031 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
9033 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
9035 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
9037 * config/alpha/alpha.c (alpha_split_conditional_move):
9038 Use std::swap instead of manually swapping.
9039 (alpha_stdarg_optimize_hook): Ditto.
9040 (alpha_canonicalize_comparison): Ditto.
9042 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9044 * tree-loop-distribution.c (struct builtin_info): New struct.
9045 (struct partition): Refactor fields into struct builtin_info.
9046 (partition_free): Free struct builtin_info.
9047 (build_size_arg_loc, build_addr_arg_loc): Delete.
9048 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
9049 information from struct builtin_info.
9050 (find_single_drs): New function refactored from classify_partition.
9051 Also moved builtin validity checks to this function.
9052 (compute_access_range, alloc_builtin): New functions.
9053 (classify_builtin_st, classify_builtin_ldst): New functions.
9054 (classify_partition): Refactor code into functions find_single_drs,
9055 classify_builtin_st and classify_builtin_ldst.
9056 (distribute_loop): Don't do runtime alias check when distributing
9058 (find_seed_stmts_for_distribution): New function.
9059 (pass_loop_distribution::execute): Refactor code finding seed
9060 stmts into above function. Support distribution for the innermost
9061 two-level loop nest. Adjust dump information.
9063 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9065 * tree-loop-distribution.c: Adjust the general comment.
9066 (NUM_PARTITION_THRESHOLD): New macro.
9067 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
9068 (classify_partition): Skip builtin pattern of loop nest's inner loop.
9069 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
9070 in call to build_partition_graph.
9071 (finalize_partitions): New parameter. Make loop distribution more
9072 conservative by fusing more partitions.
9073 (distribute_loop): Don't do runtime alias check in case of loop nest
9075 (find_seed_stmts_for_distribution): New function.
9076 (prepare_perfect_loop_nest): New function.
9077 (pass_loop_distribution::execute): Refactor code finding seed stmts
9078 and loop nest into above functions. Support loop nest distribution.
9079 Adjust dump information accordingly.
9081 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9083 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
9084 and set PTYPE_SEQUENTIAL for merged partition.
9086 2017-10-12 Richard Biener <rguenther@suse.de>
9088 PR tree-optimization/69728
9090 2017-09-19 Richard Biener <rguenther@suse.de>
9092 PR tree-optimization/69728
9093 * graphite-sese-to-poly.c (schedule_error): New global.
9094 (add_loop_schedule): Handle empty domain by failing the
9096 (build_original_schedule): Handle schedule_error.
9098 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
9099 domain by returning an unchanged schedule.
9101 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9103 * genrecog.c (validate_pattern): For VEC_SELECT verify that
9104 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
9106 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
9108 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
9111 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9114 * expr.c (store_field): Don't optimize away bitsize == 0 store
9115 from CALL_EXPR with addressable return type.
9117 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9119 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
9120 * config/rs6000/rs6000.md (sel): Delete mode attribute.
9121 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
9122 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
9123 TARGET_ISEL instead of TARGET_ISEL<sel>.
9125 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
9127 * config/rs6000/rs6000.c
9128 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
9130 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9132 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
9134 (reg_or_cint_operand): Fix comment.
9135 (reg_or_zero_operand): New predicate.
9136 * config/rs6000/rs6000-protos.h (output_isel): Delete.
9137 * config/rs6000/rs6000.c (output_isel): Delete.
9138 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
9139 instead of reg_or_cint_operand. Output instruction directly (not via
9141 (isel_unsigned_<mode>): Ditto.
9142 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
9143 gpc_reg_operand. Add an instruction alternative for this. Output
9144 instruction directly.
9145 (*isel_reversed_unsigned_<mode>): Ditto.
9147 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9149 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
9150 (TARGET_CANONICALIZE_COMPARISON): Define.
9152 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
9155 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
9156 Check whether the dest is REG before adding REG_EQUIV note.
9158 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9161 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9163 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9164 (make_hard_regno_born, make_hard_regno_dead): Update
9165 bb_killed_pseudos and bb_gen_pseudos.
9167 2017-10-11 Nathan Sidwell <nathan@acm.org>
9169 * incpath.h (enum incpath_kind): Name enum, prefix values.
9170 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
9171 * incpath.c (heads, tails): Use INC_MAX.
9172 (add_env_var_paths, add_standard_paths): Use incpath_kind.
9173 (merge_include_chains, split_quote_chain,
9174 register_include_chains): Update incpath_kind names.
9175 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
9176 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
9178 (add_framework_path, darwin_register_objc_includes): Likewise.
9179 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
9181 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9183 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
9184 Do not use float_operator operator predicate.
9185 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
9186 * config/i386/predicates.md (float_operator): Remove predicate.
9188 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9190 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
9191 (*jccxf_i387): Ditto.
9192 (*jcc<mode>_i387): Ditto.
9193 (*jccu<mode>_i387): Ditto.
9194 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
9195 (*jcc_*_i387 splitters): Remove.
9196 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
9197 * config/i386/i386.c (ix86_split_fp_branch): Remove.
9198 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
9201 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9203 * profile-count.h (slow_safe_scale_64bit): New function.
9204 (safe_scale_64bit): New inline.
9205 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
9206 * profile-count.c: Include wide-int.h
9207 (slow_safe_scale_64bit): New.
9209 2017-10-11 Nathan Sidwell <nathan@acm.org>
9211 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
9212 HAS_DECL_ASSEMBLER_NAME_P.
9213 * gimple-expr.c (gimple_decl_printable_name: Check
9214 HAS_DECL_ASSEMBLER_NAME_P too.
9215 * ipa-utils.h (type_in_anonymous_namespace_p): Check
9216 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
9217 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
9218 * passes.c (rest_of_decl_compilation): Check
9219 HAS_DECL_ASSEMBLER_NAME_P too.
9220 * recog.c (verify_changes): Likewise.
9221 * tree-pretty-print.c (dump_decl_name): Likewise.
9222 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
9224 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
9225 (SET_DECL_ASSEMBLER_NAME): Use it.
9226 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
9227 (COPY_DECL_ASSEMBLER_NAME): Likewise.
9228 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
9230 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9232 * config.gcc (i386, x86_64): Add extra objects.
9233 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
9234 (ix86_min_insn_size): Declare.
9235 (ix86_issue_rate): Declare.
9236 (ix86_adjust_cost): Declare.
9237 (ia32_multipass_dfa_lookahead): Declare.
9238 (ix86_macro_fusion_p): Declare.
9239 (ix86_macro_fusion_pair_p): Declare.
9240 (ix86_bd_has_dispatch): Declare.
9241 (ix86_bd_do_dispatch): Declare.
9242 (ix86_core2i7_init_hooks): Declare.
9243 (ix86_atom_sched_reorder): Declare.
9244 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
9245 (COSTS_N_BYTES): Move to x86-tune-costs.h.
9246 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
9247 (rip_relative_addr_p): Rename to ...
9248 (ix86_rip_relative_addr_p): ... this one; export.
9249 (memory_address_length): Update.
9250 (ix86_issue_rate): Move to x86-tune-sched.c.
9251 (ix86_flags_dependent): Move to x86-tune-sched.c.
9252 (ix86_agi_dependent): Move to x86-tune-sched.c.
9253 (exact_dependency_1): Move to x86-tune-sched.c.
9254 (exact_store_load_dependency): Move to x86-tune-sched.c.
9255 (ix86_adjust_cost): Move to x86-tune-sched.c.
9256 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
9257 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
9258 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
9259 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
9260 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
9261 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
9262 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
9263 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
9264 (min_insn_size): Rename to ...
9265 (ix86_min_insn_size): ... this one; export.
9266 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
9267 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
9268 (core2i7_first_cycle_multipass_backtrack): Move to
9269 x86-tune-sched-core.c.
9270 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
9271 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
9272 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
9273 (ix86_avoid_jump_mispredicts): Update.
9274 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
9275 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
9276 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
9277 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
9278 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
9279 (MAX_INSN): Move to ix86-tune-sched-bd.c.
9280 (MAX_IMM): Move to ix86-tune-sched-bd.c.
9281 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
9282 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
9283 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
9284 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
9285 (MAX_STORE): Move to ix86-tune-sched-bd.c.
9286 (BIG): Move to ix86-tune-sched-bd.c.
9287 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
9288 (enum insn_path): Move to ix86-tune-sched-bd.c.
9289 (get_mem_group): Move to ix86-tune-sched-bd.c.
9290 (is_cmp): Move to ix86-tune-sched-bd.c.
9291 (dispatch_violation): Move to ix86-tune-sched-bd.c.
9292 (is_branch): Move to ix86-tune-sched-bd.c.
9293 (is_prefetch): Move to ix86-tune-sched-bd.c.
9294 (init_window): Move to ix86-tune-sched-bd.c.
9295 (allocate_window): Move to ix86-tune-sched-bd.c.
9296 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
9297 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
9298 (process_end_window): Move to ix86-tune-sched-bd.c.
9299 (allocate_next_window): Move to ix86-tune-sched-bd.c.
9300 (find_constant): Move to ix86-tune-sched-bd.c.
9301 (get_num_immediates): Move to ix86-tune-sched-bd.c.
9302 (has_immediate): Move to ix86-tune-sched-bd.c.
9303 (get_insn_path): Move to ix86-tune-sched-bd.c.
9304 (get_insn_group): Move to ix86-tune-sched-bd.c.
9305 (count_num_restricted): Move to ix86-tune-sched-bd.c.
9306 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
9307 (add_insn_window): Move to ix86-tune-sched-bd.c.
9308 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
9309 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
9310 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
9311 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
9312 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
9313 (do_dispatch): Move to ix86-tune-sched-bd.c.
9314 (has_dispatch): Move to ix86-tune-sched-bd.c.
9315 * config/i386/t-i386: Add new object files.
9316 * config/i386/x86-tune-costs.h: New file.
9317 * config/i386/x86-tune-sched-atom.c: New file.
9318 * config/i386/x86-tune-sched-bd.c: New file.
9319 * config/i386/x86-tune-sched-core.c: New file.
9320 * config/i386/x86-tune-sched.c: New file.
9322 2017-10-11 Liu Hao <lh_mouse@126.com>
9324 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
9325 the generic version below instead.
9326 (should_colorize): Recognize Windows consoles as terminals
9328 * pretty-print.c [__MINGW32__] (write_all): New function.
9329 [__MINGW32__] (find_esc_head): Likewise.
9330 [__MINGW32__] (find_esc_terminator): Likewise.
9331 [__MINGW32__] (eat_esc_sequence): Likewise.
9332 [__MINGW32__] (mingw_ansi_fputs): New function that handles
9334 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
9337 2017-10-11 Richard Biener <rguenther@suse.de>
9339 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
9340 Properly call analyze_scalar_evolution with the loop of the stmt.
9342 2017-10-11 Richard Biener <rguenther@suse.de>
9344 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
9345 * tree-core.h (tree_base): Add chrec_var union member.
9346 * tree.h (CHREC_VAR): Remove.
9347 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
9348 * tree-chrec.h (build_polynomial_chrec): Adjust.
9349 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
9350 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
9352 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
9354 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
9355 * match.pd: ... here.
9356 ((T) X == (T) Y): Relax condition.
9358 2017-10-11 Bin Cheng <bin.cheng@arm.com>
9360 PR tree-optimization/82472
9361 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
9363 (break_alias_scc_partitions): Update postorder number.
9365 2017-10-11 Martin Liska <mliska@suse.cz>
9368 * opts.c (parse_no_sanitize_attribute): Do not use error_value
9370 * opts.h (parse_no_sanitize_attribute): Remove last argument.
9372 2017-10-11 Martin Liska <mliska@suse.cz>
9374 * print-rtl.c (print_insn): Move declaration of idbuf
9375 to same scope as name.
9377 2017-10-11 Martin Liska <mliska@suse.cz>
9382 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9385 2017-10-11 Martin Liska <mliska@suse.cz>
9388 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9391 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
9393 * config.gcc (default_gnu_indirect_function): Default to yes for
9394 arm*-*-linux* with glibc.
9396 2017-10-11 Richard Biener <rguenther@suse.de>
9398 * tree-scalar-evolution.c (get_scalar_evolution): Handle
9399 default-defs and types we do not want to analyze.
9400 (interpret_loop_phi): Replace unreachable code with an assert.
9401 (compute_scalar_evolution_in_loop): Remove and inline ...
9402 (analyze_scalar_evolution_1): ... here, replacing condition with
9403 what makes the intent clearer. Remove handling of cases
9404 get_scalar_evolution now handles.
9406 2017-10-10 Jim Wilson <wilson@tuliptree.org>
9408 PR rtl-optimization/81434
9409 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
9410 comment for main loop. In sched_group_found if, also add checks for
9411 pass and min_cost_group.
9413 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
9415 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
9416 (rs6000_insn_cost): New function.
9417 * config/rs6000/rs6000.md (cost): New attribute.
9419 2017-10-10 Jakub Jelinek <jakub@redhat.com>
9420 H.J. Lu <hongjiu.lu@intel.com>
9424 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
9425 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
9426 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
9427 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
9428 to OPTION_MASK_ISA_AVX512VL - builtins that have both
9429 OPTION_MASK_ISA_MMX and some other bit set require both
9430 mmx and the ISAs without the mmx bit.
9431 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
9432 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
9433 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
9434 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
9435 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
9436 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
9437 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
9438 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
9439 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
9440 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
9441 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
9442 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
9443 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
9444 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
9445 Add OPTION_MASK_ISA_MMX.
9447 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
9449 * config.gcc (armv7*-*-freebsd*): New target.
9450 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
9452 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
9454 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
9455 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
9458 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
9460 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
9461 a template parameter.
9462 (WIDE_INT_REF_FOR): Update accordingly.
9463 * tree.h (wi::int_traits <const_tree>): Delete.
9464 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
9465 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
9466 (wi::tree_to_wide_ref): New typedef.
9467 (wi::to_wide): New function.
9468 * calls.c (get_size_range): Use wi::to_wide when operating on
9470 * cgraph.c (cgraph_node::create_thunk): Likewise.
9471 * config/i386/i386.c (ix86_data_alignment): Likewise.
9472 (ix86_local_alignment): Likewise.
9473 * dbxout.c (stabstr_O): Likewise.
9474 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
9475 * expr.c (const_vector_from_tree): Likewise.
9476 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
9477 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
9478 (fold_negate_expr_1, int_const_binop_1, const_binop)
9479 (fold_convert_const_int_from_real, optimize_bit_field_compare)
9480 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
9481 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
9482 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
9483 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
9484 (fold_not_const, round_up_loc): Likewise.
9485 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
9486 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
9487 (alloca_call_type): Likewise.
9488 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
9489 * godump.c (go_output_typedef): Likewise.
9490 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
9491 * internal-fn.c (get_min_precision): Likewise.
9492 * ipa-cp.c (ipcp_store_vr_results): Likewise.
9493 * ipa-polymorphic-call.c
9494 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
9495 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
9496 (ipa_modify_call_arguments): Likewise.
9497 * match.pd: Likewise.
9498 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
9499 * print-tree.c (print_node_brief, print_node): Likewise.
9500 * stmt.c (expand_case): Likewise.
9501 * stor-layout.c (layout_type): Likewise.
9502 * tree-affine.c (tree_to_aff_combination): Likewise.
9503 * tree-cfg.c (group_case_labels_stmt): Likewise.
9504 * tree-data-ref.c (dr_analyze_indices): Likewise.
9505 (prune_runtime_alias_test_list): Likewise.
9506 * tree-dump.c (dequeue_and_dump): Likewise.
9507 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
9508 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
9509 * tree-pretty-print.c (dump_generic_node): Likewise.
9510 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
9511 (simple_iv_with_niters): Likewise.
9512 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
9513 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
9514 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
9515 * tree-ssa-loop-niter.c (split_to_var_and_offset)
9516 (refine_value_range_using_guard, number_of_iterations_ne_max)
9517 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
9518 (get_cst_init_from_scev, record_nonwrapping_iv)
9519 (scev_var_range_cant_overflow): Likewise.
9520 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
9521 * tree-ssa-pre.c (compute_avail): Likewise.
9522 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
9523 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
9524 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
9525 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
9526 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
9527 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
9528 (dump_case_nodes, try_switch_expansion): Likewise.
9529 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
9530 (vect_do_peeling): Likewise.
9531 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
9532 * tree-vect-stmts.c (vectorizable_load): Likewise.
9533 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
9534 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
9535 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
9536 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
9537 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
9538 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
9539 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
9540 (evrp_dom_walker::before_dom_children): Likewise.
9541 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
9542 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
9543 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
9544 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
9545 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
9546 (get_range_pos_neg): Likewise.
9547 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
9548 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
9549 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
9550 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
9551 * config/avr/avr.c (avr_fold_builtin): Likewise.
9552 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
9553 * config/msp430/msp430.c (msp430_attr): Likewise.
9554 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
9555 * config/powerpcspe/powerpcspe-c.c
9556 (altivec_resolve_overloaded_builtin): Likewise.
9557 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
9558 (rs6000_expand_ternop_builtin): Likewise.
9559 * config/rs6000/rs6000-c.c
9560 (altivec_resolve_overloaded_builtin): Likewise.
9561 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
9562 (rs6000_expand_ternop_builtin): Likewise.
9563 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
9565 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9567 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
9568 when copying loop nest with only one inner loop.
9570 2017-10-10 Richard Biener <rguenther@suse.de>
9572 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
9573 blocks if SCEV is active.
9574 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
9576 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
9577 (scev_initialize): Assert we are not yet initialized.
9579 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9581 * tree-loop-distribution.c (generate_loops_for_partition): Remove
9582 inner loop's exit stmt by making it always exit the loop, otherwise
9583 we would generate an infinite empty loop.
9585 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9587 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
9588 renaming variables in new preheader if it's deleted.
9590 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9592 * tree-loop-distribution.c (struct partition): Remove unused field
9593 loops of the structure.
9594 (partition_alloc, partition_free): Ditto.
9595 (build_rdg_partition_for_vertex): Ditto.
9597 2017-10-09 Jeff Law <law@redhat.com>
9599 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
9600 return type to match prototype and documentation.
9602 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9604 * config/rs6000/rs6000.c (processor_costs): Move to ...
9605 * config/rs6000/rs6000.h: ... here.
9606 (rs6000_cost): Declare.
9608 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
9610 * except.c (setjmp_fn): New global variable.
9611 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
9612 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
9613 if DONT_USE_BUILTIN_SETJMP is defined.
9615 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9617 * target.def (insn_cost): New hook.
9618 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
9619 * doc/tm.texi: Regenerate.
9620 * rtlanal.c (insn_cost): Use the new hook.
9622 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9624 * combine.c (combine_validate_cost): Compute the new insn_cost,
9625 not just pattern_cost.
9626 (try_combine): Adjust comment.
9628 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9630 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
9632 * combine.c (uid_insn_cost): Adjust comment.
9633 (combine_validate_cost): Adjust comment. Use pattern_cost instead
9635 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
9636 * dse.c (find_shift_sequence): Ditto.
9637 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
9638 (bb_valid_for_noce_process_p): Use pattern_cost.
9639 * rtl.h (insn_rtx_cost): Delete.
9640 (pattern_cost): New prototype.
9641 (insn_cost): New prototype.
9642 * rtlanal.c (insn_rtx_cost): Rename to...
9643 (pattern_cost): ... this.
9646 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
9648 * config/i386/i386.md (*jcc_2): Remove insn pattern.
9649 (*jcc<mode>_0_r_i387): Ditto.
9650 (*jccxf_r_i387): Ditto.
9651 (*jcc<mode>_r_i387): Ditto.
9652 (*jccu<mode>_r_i387): Ditto.
9653 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
9654 (*jcc): Rename from *jcc_1.
9656 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9658 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
9659 deferred rescans after the lvx/stvx recombination pre-pass.
9661 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
9663 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
9664 memory operation instruction support.
9665 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
9666 (rs6000-ibm-aix[789]*): Likewise.
9667 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
9668 Document new functions.
9670 2017-10-09 Richard Biener <rguenther@suse.de>
9672 PR tree-optimization/82397
9673 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
9674 equality only for semantically equal trees.
9676 2017-10-09 Richard Biener <rguenther@suse.de>
9678 PR tree-optimization/82449
9679 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
9680 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
9681 allow constant addresses.
9682 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
9685 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9687 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
9690 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9693 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
9696 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9699 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
9701 2017-10-09 Jakub Jelinek <jakub@redhat.com>
9704 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
9705 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
9707 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
9709 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
9710 (WI_BINARY_PREDICATE_RESULT): Likewise.
9711 (wi::binary_traits::operator_result): New type.
9712 (wi::binary_traits::predicate_result): Likewise.
9713 (generic_wide_int::operator~, unary generic_wide_int::operator-)
9714 (generic_wide_int::operator==, generic_wide_int::operator!=)
9715 (generic_wide_int::operator&, generic_wide_int::and_not)
9716 (generic_wide_int::operator|, generic_wide_int::or_not)
9717 (generic_wide_int::operator^, generic_wide_int::operator+
9718 (binary generic_wide_int::operator-, generic_wide_int::operator*):
9720 (operator~, unary operator-, operator==, operator!=, operator&)
9721 (operator|, operator^, operator+, binary operator-, operator*): New
9723 * expr.c (get_inner_reference): Use wi::bit_and_not.
9724 * fold-const.c (fold_binary_loc): Likewise.
9725 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
9726 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
9727 (bit_value_binop): Likewise.
9728 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
9729 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
9730 (extract_range_from_binary_expr_1): Likewise.
9731 (masked_increment): Likewise.
9732 (simplify_bit_ops_using_ranges): Likewise.
9734 2017-10-09 Martin Jambor <mjambor@suse.cz>
9737 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
9738 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
9739 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
9741 (hsa_fixup_mov_insn_type): New function.
9742 (hsa_op_with_type::get_in_type): Use it.
9743 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
9744 immediates in an assert.
9745 (hsa_op_with_type::extend_int_to_32bit): New method.
9746 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
9747 types. Convert to dest type if necessary.
9748 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
9749 (reg_for_gimple_ssa): Pass false as min32int to
9750 hsa_type_for_scalar_tree_type.
9751 (gen_hsa_addr): Fixup type when creating addresable temporary.
9752 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
9753 (gen_hsa_unary_operation): Extend operands and convert to dest type if
9754 necessary. Call hsa_fixup_mov_insn_type.
9755 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
9756 extend operands and convert to dest type if necessary.
9757 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
9758 to dest type if necessary.
9759 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
9760 if conversion nt necessary and size matches.
9761 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
9762 to dest type if necessary.
9763 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
9764 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
9766 (gen_hsa_clrsb): Likewise.
9767 (gen_hsa_ffs): Likewise.
9768 (gen_hsa_divmod): Extend operands and convert to dest type if
9770 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
9772 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
9774 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
9775 Remove empty default arguments. Use a brace block as output
9777 (conditional return): Ditto.
9779 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
9780 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
9782 (group_ending_nop): Ditto.
9783 (doloop_end): Ditto.
9784 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
9785 (splitters for those): Ditto.
9787 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
9789 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
9790 a conditional jump (and the compare for it) so that pc_rtx is the
9792 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
9793 for the deleted and renamed ctr<mode>_internal[234] patterns.
9794 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
9795 Delete second conditional return pattern.
9796 (ctr<mode>_internal2): Delete this second bdnz pattern.
9797 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
9798 (ctr<mode>_internal4): Delete this second bdz pattern.
9800 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9802 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
9803 (always_initialized_rtx_for_ssa_name_p): New predicate.
9804 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
9805 (finish_out_of_ssa): Free new field of SA.
9806 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
9807 * tree-ssa-coalesce.c: Include tree-ssa.h.
9808 (get_parm_default_def_partitions): Remove extern keyword.
9809 (get_undefined_value_partitions): New function.
9810 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
9811 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
9813 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
9815 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9817 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
9819 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9821 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
9822 for targets that preffer 128bit.
9824 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9826 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
9828 2017-10-08 Olivier Hainque <hainque@adacore.com>
9830 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
9831 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
9833 (thumb_set_return_address): Likewise.
9835 2017-10-08 Olivier Hainque <hainque@adacore.com>
9837 * common/config/arm/arm-common.c (arm_except_unwind_info):
9838 Handle DWARF2_UNWIND_INFO.
9840 2017-10-07 Michael Collison <michael.collison@arm.com>
9842 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
9845 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
9847 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
9848 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
9849 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
9850 defined, force the creation of a new block for a dispatch label.
9852 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
9854 * invoke.texi (Wsuggest-attribute=cold): Document.
9855 * common.opt (Wsuggest-attribute=cold): New
9856 * ipa-pure-const.c (warn_function_cold): New function.
9857 * predict.c (compute_function_frequency): Use it.
9858 * predict.h (warn_function_cold): Declare.
9860 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
9862 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
9865 2017-10-06 Martin Liska <mliska@suse.cz>
9867 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
9868 keyword for member functions.
9869 (struct sanopt_tree_couple): New struct.
9870 (struct sanopt_tree_couple_hash): New function.
9871 (struct sanopt_ctx): Add new hash_map.
9872 (has_dominating_ubsan_ptr_check): New function.
9873 (record_ubsan_ptr_check_stmt): Likewise.
9874 (maybe_optimize_ubsan_ptr_ifn): Likewise.
9875 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
9876 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
9878 2017-10-06 Sudakshina Das <sudi.das@arm.com>
9881 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
9882 aarch64_simd_valid_immediate on CONST_VECTORs.
9883 (aarch64_reg_or_bic_imm): Likewise.
9885 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
9887 PR rtl-optimization/82396
9888 * haifa-sched.c (ready_sort_real): Disable qsort checking.
9890 2017-10-06 Sebastian Pop <sebpop@gmail.com>
9892 * graphite-dependences.c (scop_get_reads): Move code to...
9893 (scop_get_must_writes): Move code to...
9894 (scop_get_may_writes): Move code to...
9895 (scop_get_reads_and_writes): ... here.
9896 (scop_get_dependences): Call scop_get_reads_and_writes.
9898 2017-10-06 Jakub Jelinek <jakub@redhat.com>
9900 PR tree-optimization/82434
9901 * fold-const.h (can_native_encode_type_p,
9902 can_native_encode_string_p): Remove.
9903 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
9904 don't encode anything, just return what would be otherwise returned.
9905 (native_encode_fixed, native_encode_complex, native_encode_vector):
9907 (native_encode_string): Likewise. Inline by hand
9908 can_native_encode_string_p.
9909 (can_native_encode_type_p): Remove.
9910 (can_native_encode_string_p): Remove.
9911 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
9912 STRING_CSTs using can_native_encode_string_p, test all
9913 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
9914 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
9915 argument from native_encode_expr.
9916 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
9917 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
9920 2017-10-06 Richard Biener <rguenther@suse.de>
9922 PR tree-optimization/82397
9923 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
9924 operand_equal_p but rely on data_ref_compare_tree for detecting
9926 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
9927 to match up with dr_group_sort_cmp.
9929 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9932 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
9934 * config/s390/s390-builtin-types.def: Regenerate.
9936 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9939 * config/s390/s390-builtin-types.def: Regenerate.
9940 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
9941 Change flag from B_VXE to B_VX.
9942 (s390_vec_min_dbl): Remove B_VXE flag.
9944 2017-10-06 Richard Biener <rguenther@suse.de>
9946 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
9947 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
9948 translate_isl_ast_to_gimple::is_valid_rename,
9949 translate_isl_ast_to_gimple::get_rename,
9950 translate_isl_ast_to_gimple::get_def_bb_for_const,
9951 translate_isl_ast_to_gimple::get_new_name,
9952 translate_isl_ast_to_gimple::collect_all_ssa_names,
9953 translate_isl_ast_to_gimple::copy_loop_phi_args,
9954 translate_isl_ast_to_gimple::collect_all_ssa_names,
9955 translate_isl_ast_to_gimple::copy_loop_phi_args,
9956 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
9957 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
9958 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
9959 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
9960 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
9961 translate_isl_ast_to_gimple::copy_cond_phi_args,
9962 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
9963 translate_isl_ast_to_gimple::edge_for_new_close_phis,
9964 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
9965 translate_isl_ast_to_gimple::rename_uses,
9966 translate_isl_ast_to_gimple::rename_all_uses): Remove.
9967 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
9968 (set_rename_for_each_def): Likewise.
9969 (graphite_copy_stmts_from_block): Handle debug stmt resetting
9970 here. Handle rewriting SCEV analyzable uses here.
9971 (copy_bb_and_scalar_dependences): Generate code for PHI
9973 (graphite_regenerate_ast_isl): Adjust.
9974 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
9975 (add_write, add_read): New functions.
9976 (build_cross_bb_scalars_def): Use it and simplify.
9977 (build_cross_bb_scalars_use): Likewise.
9978 (graphite_find_cross_bb_scalar_vars): Inline into...
9979 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
9980 simulating out-of-SSA. Compute liveout and add dependencies.
9981 (build_scops): Force an empty entry block.
9982 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
9984 (sese_build_liveouts): Declare.
9985 (sese_trivially_empty_bb_p): Likewise.
9986 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
9987 compute liveout and debug_liveout.
9988 (sese_bad_liveouts_use): Remove.
9989 (sese_reset_debug_liveouts_bb): Likewise.
9990 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
9991 (sese_build_liveouts): Build liveout and debug_liveout and store
9993 (new_sese_info): Adjust.
9994 (free_sese_info): Likewise.
9995 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
9996 do not build liveout here.
9997 (move_sese_in_condition): Adjust region entry.
9998 (scev_analyzable_p): Match up with chrec_apply requirements.
9999 (sese_trivially_empty_bb_p): New.
10000 * tree-into-ssa.c (get_reaching_def): Properly support generating
10001 default-defs for incremental rewrite of anonymous names.
10003 2017-10-06 Richard Biener <rguenther@suse.de>
10005 * graphite-sese-to-poly.c (extract_affine): For casts increasing
10006 precision do not perform modulo reduction.
10008 2017-10-06 Richard Biener <rguenther@suse.de>
10010 PR tree-optimization/82436
10011 * tree-vect-slp.c (vect_supported_load_permutation_p): More
10012 conservatively choose the vectorization factor when checking
10013 whether we can perform the required load permutation.
10014 (vect_transform_slp_perm_load): Assert when we may not fail.
10016 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
10018 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
10019 message for incompatible -msdata=* and -mcall-* options.
10021 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10023 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
10024 rate for post-reload scheduling.
10026 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10028 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
10030 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10032 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
10033 to improve monte carlo in scimark.
10035 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10037 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
10038 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
10039 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
10040 pentium4_cost, nocona_cost): Set reassociation width to 1.
10041 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
10042 width to 2 for fp operations and 1 otherwise.
10043 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
10045 (atom_cost): Set reassociation width to 2.
10046 (slm_cost, generic_cost): Set fp reassociation width
10047 to 2 and 1 otherwise.
10048 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
10049 (core_cost): Set fp reassociation width to 4 and vector to 2.
10050 (ix86_reassociation_width): Rewrite using cost table; special case
10051 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
10052 and TARGET_AVX128_OPTIMAL.
10053 * config/i386/i386.h (processor_costs): Add
10054 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
10055 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
10056 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
10057 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
10058 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
10059 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
10061 2017-10-05 Nathan Sidwell <nathan@acm.org>
10063 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
10065 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10067 * config/arm/arm.c (arm_test_fpu_data): New.
10068 (arm_run_selftests): Call arm_test_fpu_data.
10070 2017-10-04 Nathan Sidwell <nathan@acm.org>
10072 * toplev.c (toplev::main): Remove excess parens on pretty_printer
10074 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
10076 2017-10-04 Sudakshina Das <sudi.das@arm.com>
10078 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
10079 check type for aarch64_simd_valid_immediate.
10080 (aarch64_output_simd_mov_immediate): Update prototype.
10081 (aarch64_simd_valid_immediate): Update prototype.
10082 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
10083 support for ORR-immediate.
10084 (and<mode>3): modified pattern to add support for BIC-immediate.
10085 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
10086 now checks for valid immediate for BIC and ORR based on new enum
10088 (aarch64_output_simd_mov_immediate): Function now used to output
10089 BIC/ORR imm as well based on new enum argument.
10090 * config/aarch64/constraints.md (Do): New vector immediate constraint.
10092 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
10093 (aarch64_reg_or_bic_imm): Likewise.
10095 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10097 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
10098 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
10100 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10104 PR rtl-optimization/82396
10105 * haifa-sched.c (autopref_multipass_init): Simplify
10107 (autopref_rank_data): Simplify sort order.
10108 * sched-int.h (autopref_multipass_data_): Remove
10109 multi_mem_insn_p, min_offset and max_offset.
10111 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10113 * doc/sourcebuild.texi: Document vect_peeling_profitable.
10115 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10117 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
10118 vect_doubleint_cvt.
10120 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10122 * doc/sourcebuild.texi: Document vect_long_mult.
10124 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
10126 PR tree-optimization/82413
10127 * fold-const.c (build_range_check): Use widest_int when comparing
10128 the maximum ETYPE value with HIGH.
10130 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10132 PR rtl-optimization/82396
10133 * haifa-sched.c (autopref_multipass_init): Simplify
10135 (autopref_rank_data): Simplify sort order.
10136 * sched-int.h (autopref_multipass_data_): Remove
10137 multi_mem_insn_p, min_offset and max_offset.
10139 2017-10-04 Jakub Jelinek <jakub@redhat.com>
10141 PR tree-optimization/82381
10142 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
10143 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
10146 PR tree-optimization/82374
10147 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
10148 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
10149 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
10150 current_function_decl to the new decl.
10152 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
10154 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
10155 helper macro for IEEE float128 hardware built-in functions.
10156 (SQRTF128_ODD): Add built-in functions with the round-to-odd
10158 (TRUNCF128_ODD): Likewise.
10159 (ADDF128_ODD): Likewise.
10160 (SUBF128_ODD): Likewise.
10161 (MULF128_ODD): Likewise.
10162 (DIVF128_ODD): Likewise.
10163 (FMAF128_ODD): Likewise.
10164 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
10165 UNSPEC_TRUNC_ROUND_TO_ODD.
10166 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
10167 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
10168 floating point round to odd instructions.
10169 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
10170 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
10171 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
10172 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
10173 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
10174 (trunc<mode>sf2_hw): Change the truncate with round to odd
10175 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
10176 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
10177 to odd hardware instructions.
10178 (sub<mode>3_odd): Likewise.
10179 (mul<mode>3_odd): Likewise.
10180 (div<mode>3_odd): Likewise.
10181 (sqrt<mode>2_odd): Likewise.
10182 (fma<mode>4_odd): Likewise.
10183 (fms<mode>4_odd): Likewise.
10184 (nfma<mode>4_odd): Likewise.
10185 (nfms<mode>4_odd): Likewise.
10186 (trunc<mode>df2_odd): Change the truncate with round to odd
10187 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
10189 * doc/extend.texi (PowerPC built-in functions): Update documentation
10190 for existing IEEE float128-bit built-in functions. Add built-in
10191 functions that generate the IEEE 128-bit floating point round to
10194 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
10196 PR rtl-optimization/77729
10197 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
10198 to (X&(C1&~C2))|C2 transformations.
10200 2017-10-03 Martin Jambor <mjambor@suse.cz>
10202 PR tree-optimization/82363
10203 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
10204 mismatch, mark lacc written regardless of racc.
10206 2017-10-03 Jakub Jelinek <jakub@redhat.com>
10208 PR tree-optimization/82381
10209 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
10210 stmt_to_insert nor wheather SSA_NAMEs are default defs.
10211 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
10212 fallthrough into reassoc_stmt_dominates_stmt_p.
10215 * combine.c (combine_instructions): Don't combine in unreachable
10218 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
10221 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
10222 function to not use the have_cpu variable. Do not set cpu_index,
10223 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
10224 or the default cpu.
10225 (rs6000_valid_attribute_p): Remove duplicate initializations of
10226 old_optimize and func_optimize.
10227 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
10228 (rs6000_activate_target_options): Make global.
10229 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
10232 2017-10-02 Jakub Jelinek <jakub@redhat.com>
10234 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
10235 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
10236 Set *poffset to 0 and *psize and *pmax_size to -1 if
10237 *poffset + *psize overflows in HOST_WIDE_INT.
10239 PR tree-optimization/82387
10240 PR tree-optimization/82388
10241 PR tree-optimization/82389
10242 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
10243 instead of live_bytes non-NULL.
10245 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
10248 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
10251 2017-10-02 Richard Biener <rguenther@suse.de>
10253 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
10254 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
10255 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
10257 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10259 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
10260 requested precision matches the type's.
10261 * calls.c (alloc_max_size): Calculate the new candidate size as
10262 a widest_int and use wi::to_widest when comparing it with the
10263 current candidate size.
10264 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
10265 zero rather than integer_zero_node.
10266 * match.pd: Check for a no-op conversion before using wi::add
10267 rather than after. Use tree_to_uhwi when summing small shift
10268 counts into an unsigned int.
10270 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10271 Alan Hayward <alan.hayward@arm.com>
10272 David Sherwood <david.sherwood@arm.com>
10275 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
10276 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
10277 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
10278 POINTER_AND_FP_REGS.
10280 2017-10-02 Richard Biener <rguenther@suse.de>
10282 PR tree-optimization/82355
10283 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
10284 a mapping for the enclosing loop but avoid generating one for
10285 the loop tree root.
10286 (copy_bb_and_scalar_dependences): Remove premature codegen
10287 error on PHIs in blocks duplicated into multiple places.
10288 * graphite-scop-detection.c
10289 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
10290 in the region use it as loop and nest to analyze the DR in.
10291 (try_generate_gimple_bb): Likewise.
10292 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
10293 (add_loop_constraints): For blocks in a loop not in the region
10294 create a dimension with a single iteration.
10295 * sese.h (gbb_loop_at_index): Remove assert.
10297 2017-10-01 Kevin Buettner <kevinb@redhat.com>
10299 * omp-expand.c (adjust_context_scope): New function.
10300 (expand_parallel_call): Call adjust_context_scope.
10302 2017-10-01 Jeff Law <law@redhat.com>
10304 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
10305 dom_opt_dom_walker class with direct access to private members.
10306 Add comments. Call test_for_singularity.
10307 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
10308 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
10309 m_dummy_cond anymore.
10310 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
10312 (pass_dominator:execute): Build the dummy_cond here and pass it
10313 to the dom_opt_dom_walker ctor.
10314 (test_for_singularity): New function.
10316 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
10317 Maya Rashish <coypu@sdf.org>
10319 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
10320 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
10321 (alpha*-*-netbsd*) Use nbsd_tm_file.
10322 (arm*-*-netbsdelf*) Likewise.
10323 (i[34567]86-*-netbsdelf*) Likewise.
10324 (x86_64-*-netbsd*) Likewise.
10325 (mips*-*-netbsd*) Likewise.
10326 (powerpc-*-netbsd*) Likewise.
10327 (sh*-*-netbsd*) Likewise.
10328 (sparc-*-netbsdelf*) Likewise.
10329 (sparc64-*-netbsd*) Likewise.
10330 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
10332 (vax-*-netbsdelf*) Likewise.
10333 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
10334 (UINT_FAST8_TYPE) Likewise.
10335 (INT_FAST16_TYPE) Check CHAR_FAST16.
10336 (UINT_FAST16_TYPE) Likewise.
10338 2017-09-30 Jakub Jelinek <jakub@redhat.com>
10341 * config/i386/i386.md
10342 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
10343 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
10344 *divmodsi4_zext_2): New define_insn_and_split.
10345 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
10346 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
10347 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
10348 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
10349 New define_insn_and_split.
10350 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
10351 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
10352 operands[1] having DImode when mode is SImode.
10354 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
10355 always SImode for DIV and MOD in REG_EQUAL notes.
10357 2017-09-29 Yury Gribov <tetra2005@gmail.com>
10359 PR middle-end/82319
10360 * match.pd: Fix handling of NaNs in pattern.
10362 2017-09-29 Jeff Law <law@redhat.com>
10364 * sbitmap.c (bitmap_bit_in_range_p): New function.
10365 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
10366 * tree-ssa-dse.c (live_bytes_read): New function.
10367 (dse_classify_store): Ignore reads of dead bytes.
10369 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
10370 typos and whitespace errors.
10371 * config/i386/predicates.md (address_no_seg_operand): Likewise.
10372 * config/s390/s390.c (s390_emit_prologue): Likewise.
10374 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10377 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
10378 with a symbol for LRA.
10380 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10382 PR rtl-optimization/82338
10383 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
10385 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
10387 * genmodes.c (calc_wider_mode): Suppress qsort macro.
10388 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
10389 (qsort_chk): Declare.
10390 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
10391 (qsort_chk): New function.
10393 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10395 PR tree-optimization/82337
10396 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
10397 phi definition if the PHI result appears in an abnormal PHI.
10398 (find_basis_for_base_expr): Don't record a basis if the LHS of the
10399 basis appears in an abnormal PHI.
10401 2017-09-29 Richard Biener <rguenther@suse.de>
10403 * graphite-isl-ast-to-gimple.c
10404 (translate_isl_ast_to_gimple::set_codegen_error): New function.
10405 (binary_op_to_tree): Use it.
10406 (get_rename_from_scev): Likewise.
10407 (copy_loop_phi_nodes): Likewise.
10408 (copy_bb_and_scalar_dependences): Likewise.
10409 (translate_pending_phi_nodes): Likewise.
10411 2017-09-29 Jakub Jelinek <jakub@redhat.com>
10414 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
10415 for movabsq $(i32 << shift), r64.
10417 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
10419 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
10420 index when encoding %esp as %rsp to avoid 0x67 prefix.
10422 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
10424 * config/i386/i386.md (*movsf_internal, *movdf_internal):
10425 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
10427 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
10429 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
10430 Extensions with more than 16 double VFP registers.
10431 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
10432 to_clear_mask and all code related to it. Replace the remaining
10433 entry by a sbitmap and adapt code accordingly.
10435 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
10437 * brig-builtins.def: Change pure attributes to const.
10439 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10441 * config.gcc (default_gnu_indirect_function): Default to yes for
10442 sparc*-*-linux* with glibc.
10444 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10446 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
10447 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
10448 when creating .init_array and .fini_array sections with priority
10451 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
10454 * config/aarch64/aarch64.c
10455 (aarch64_builtin_support_vector_misalignment): Always return false
10456 when misalignment is unknown.
10458 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
10460 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
10461 this function to return false if the definition used by the swap
10462 instruction is artificial, or if the memory address from which the
10463 constant value is loaded is not represented by a base address held
10464 in a register or if the base address register is a frame or stack
10465 pointer. Additionally, return false if the base address of the
10466 loaded constant is a SYMBOL_REF but is not considered to be a
10468 (replace_swapped_load_constant): New function.
10469 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
10470 loaded constant vector with a load of a swapped constant vector.
10472 2017-09-27 Carl Love <cel@us.ibm.com>
10474 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
10475 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
10476 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
10479 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
10481 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
10482 first, always call autopref_rank_data otherwise.
10484 2017-09-27 Richard Biener <rguenther@suse.de>
10486 * graphite-scop-detection.c (find_scop_parameters): Move
10487 loop bound handling ...
10488 (gather_bbs::before_dom_children): ... here, avoiding the need
10489 to build scop_info->loop_nest.
10490 (record_loop_in_sese): Remove.
10491 * sese.h (sese_info_t::loop_nest): Remove.
10492 * sese.c (new_sese_info): Do not allocate loop_nest.
10493 (free_sese_info): Do not free loop_nest.
10495 2017-09-27 Jakub Jelinek <jakub@redhat.com>
10498 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
10499 lhs from calls if the lhs has addressable type.
10501 2017-09-27 Richard Biener <rguenther@suse.de>
10503 * graphite.h (scop::max_alias_set): New member.
10504 * graphite-scop-detection.c: Remove references to non-existing
10505 --param in comments.
10506 (build_alias_sets): Record the maximum alias set used for drs.
10507 (build_scops): Support zero as unlimited for
10508 --param graphite-max-arrays-per-scop.
10509 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
10510 and inline into ...
10511 (build_poly_sr_1): ... here. Compute alias set based on the
10512 maximum alias set used for drs rather than
10513 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
10515 2017-09-27 Richard Biener <rguenther@suse.de>
10517 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
10518 --param loop-block-tile-size=0 to disable tiling.
10520 2017-09-27 Richard Biener <rguenther@suse.de>
10522 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
10523 (graphite-max-nb-scop-params): Document special value zero.
10524 * domwalk.h (dom_walker::STOP): New symbolical constant.
10525 (dom_walker::dom_walker): Add optional parameter for bb to
10527 (dom_walker::~dom_walker): Declare.
10528 (dom_walker::before_dom_children): Document STOP return value.
10529 (dom_walker::m_user_bb_to_rpo): New member.
10530 (dom_walker::m_bb_to_rpo): Likewise.
10531 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
10532 mapping here if not provided by the user.
10533 (dom_walker::~dom_walker): Free bb to RPO mapping if not
10534 provided by the user.
10535 (dom_walker::STOP): Define.
10536 (dom_walker::walk): Do not compute bb to RPO mapping here.
10537 Support STOP return value from before_dom_children to stop
10539 * graphite-optimize-isl.c (optimize_isl): If the schedule
10540 is the same still generate code if -fgraphite-identity
10541 or -floop-parallelize-all are given.
10542 * graphite-scop-detection.c: Include cfganal.h.
10543 (gather_bbs::gather_bbs): Get and pass through bb to RPO
10545 (gather_bbs::before_dom_children): Return STOP for BBs
10547 (build_scops): Compute bb to RPO mapping and pass it to
10548 the domwalk. Treat --param graphite-max-nb-scop-params=0
10549 as not limiting the number of params.
10550 * graphite.c (graphite_initialize): Remove limit on the
10551 number of basic-blocks in a function.
10552 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
10553 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
10554 default value of 10.
10556 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10558 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
10559 Adjust code to eliminate needing to do the shift right 32-bits
10560 operation after XSCVDPSPN.
10562 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10564 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
10565 ((X / Y) != 0 -> X >= Y): Likewise.
10567 2017-09-26 Carl Love <cel@us.ibm.com>
10569 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
10570 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
10571 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
10572 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
10573 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
10574 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
10575 definitions and overloading.
10576 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
10577 statement for P9V_BUILTIN_XST_LEN_R.
10578 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
10579 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
10580 define_expand and define_insn for the instructions and builtins.
10581 * doc/extend.texi: Update the built-in documentation file for the new
10582 built-in functions.
10583 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
10584 define_insn for the instructions
10586 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
10589 * gcc/config/netbsd-protos.h: New file.
10590 * gcc/config/netbsd.c: New file.
10591 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
10592 * gcc/config/t-netbsd: New file.
10593 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
10594 (tmake_file) Add t-netbsd.
10595 (extra_objs) Add netbsd.o.
10597 2017-09-26 Janus Weil <janus@gcc.gnu.org>
10601 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
10603 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10605 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
10606 sign extension from a vector register to a GPR by doing a 32-bit
10607 direct move and then an EXTSW.
10608 (extendsi<mode>2 splitter): Likewise.
10609 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
10610 right or vector extract after doing XSCVDPSPN. Use
10611 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
10613 (movdi_from_sf_zero_ext): Likewise.
10614 (reload_gpr_from_vsxsf): Likewise.
10615 (p8_mfvsrd_4_disf): Delete, no longer used.
10616 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
10617 then needing to move the SFmode to a GPR to use the XSCVDPSP
10618 instruction instead of FRSP and XSCVDPSPN.
10619 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
10620 it is adjacent to the other XSCVSPDP insns.
10621 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
10622 SFmode to be in traditional Altivec registers.
10623 (vsx_xscvdpspn): Eliminate useless alternative constraint.
10624 (vsx_xscvspdpn): Likewise.
10625 (vsx_xscvspdpn_scalar): Likewise.
10627 2017-09-26 Martin Jambor <mjambor@suse.cz>
10629 * tree-sra.c (compare_access_positions): Put integral types first,
10630 stabilize sorting of integral types, remove conditions putting
10631 non-full-precision integers last.
10632 (sort_and_splice_var_accesses): Disable scalarization if a
10633 non-integert would be represented by a non-full-precision integer.
10635 2017-09-26 Joseph Myers <joseph@codesourcery.com>
10637 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
10638 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
10639 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
10640 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
10641 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
10642 conditionals inside the function instead of around it. Call
10643 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
10644 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
10646 2017-09-26 Richard Biener <rguenther@suse.de>
10648 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
10650 (scop_detection::build_scop_breadth): ... this. Removed.
10651 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
10652 (scop_detection::harmful_stmt_in_bb): Likewise.
10653 (scop_detection::graphite_can_represent_stmt): Likewise.
10654 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
10655 (scop_detection::can_represent_loop): Remove recursion, fold in ...
10656 (scop_detection::can_represent_loop_1): ... this. Removed.
10657 (scop_detection::harmful_loop_in_region): Simplify after inlining
10658 the above and remove more quadraticness.
10659 (build_scops): Adjust.
10660 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
10663 2017-09-26 Jakub Jelinek <jakub@redhat.com>
10666 * config/i386/i386.c (ix86_print_operand_address_as): Only test
10667 REGNO (base) == SP_REG if base is a REG.
10669 PR middle-end/35691
10670 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
10671 if it is different SSA_NAME.
10672 (optimize_range_tests_cmp_bitwise): New function.
10673 (optimize_range_tests): Call it.
10675 2017-09-26 Richard Biener <rguenther@suse.de>
10677 PR tree-optimization/82321
10678 * graphite.c (canonicalize_loop_closed_ssa): Properly check
10679 for the def being inside the loop.
10681 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10683 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
10685 * config/s390/s390-builtins.def: Fix constraint on op4.
10687 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10689 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
10690 independent expanders.
10691 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
10692 ("vec_ordered", "vec_unordered"): New expanders.
10694 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10696 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
10699 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10701 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
10702 vec_unpacks_lo_v16qi.
10703 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
10705 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10707 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
10708 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
10709 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
10711 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10713 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
10715 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
10717 ("*vec_slb<mode>"): New insn pattern.
10718 ("vec_shr_<mode>"): New expander.
10719 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
10720 and force the shift count operand to V16QImode.
10721 ("vec_srb<mode>"): Set shift count mode to V16QI.
10723 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10725 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
10726 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
10727 ("vec_widen_smult_hi_<mode>"): New expander definitions.
10729 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
10732 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
10734 2017-09-26 Richard Biener <rguenther@suse.de>
10736 PR tree-optimization/82320
10737 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
10740 2017-09-25 Jeff Law <law@redhat.com>
10742 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
10743 prototype for new argument.
10744 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
10745 mostly extracted from rs6000_emit_allocate_stack.
10746 (rs6000_emit_probe_stack_range_stack_clash): New function.
10747 (rs6000_emit_allocate_stack): Call
10748 rs6000_emit_probe_stack_range_stack_clash as needed.
10749 (rs6000_emit_probe_stack_range): Add additional argument
10750 to call to gen_probe_stack_range{si,di}.
10751 (output_probe_stack_range): New.
10752 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
10753 (output_probe_stack_range_stack_clash): New.
10754 (rs6000_emit_prologue): Emit notes into dump file as requested.
10755 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
10756 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
10757 Add additional operand and pass it to output_probe_stack_range.
10759 2017-09-25 Bin Cheng <bin.cheng@arm.com>
10761 PR tree-optimization/82163
10762 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
10763 (checking_verify_loop_closed_ssa): New parameter.
10764 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
10765 (check_loop_closed_ssa_stmt): Delete.
10766 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
10767 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
10768 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
10771 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
10773 * brig-builtins.def: Treat HSAIL barrier builtins as
10774 setjmp/longjump style functions.
10776 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10778 * target.def (constant_alignment): New hook.
10779 * defaults.h (CONSTANT_ALIGNMENT): Delete.
10780 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
10781 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
10782 * doc/tm.texi: Regenerate.
10783 * targhooks.h (default_constant_alignment): Declare.
10784 (constant_alignment_word_strings): Likewise.
10785 * targhooks.c (default_constant_alignment): New function.
10786 (constant_alignment_word_strings): Likewise.
10787 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
10788 instead of CONSTANT_ALIGNMENT.
10789 * varasm.c (align_variable, get_variable_align, build_constant_desc)
10790 (force_const_mem): Likewise.
10791 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
10792 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
10793 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
10794 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10795 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
10797 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
10798 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10799 constant_alignment_word_strings.
10800 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
10801 (CONSTANT_ALIGNMENT): Likewise.
10802 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10803 (arm_constant_alignment): New function.
10804 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
10805 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10806 constant_alignment_word_strings.
10807 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
10808 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10809 constant_alignment_word_strings.
10810 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
10811 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10812 (cris_constant_alignment): New function.
10813 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
10814 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10815 (epiphany_constant_alignment): New function.
10816 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
10817 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10818 constant_alignment_word_strings.
10819 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
10820 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10821 constant_alignment_word_strings.
10822 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
10823 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10824 constant_alignment_word_strings.
10825 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
10826 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
10827 * config/i386/i386.c (ix86_constant_alignment): Make static.
10828 Use the same interface as the target hook.
10829 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10830 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
10831 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10832 constant_alignment_word_strings.
10833 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
10834 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
10835 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10836 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
10837 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10838 constant_alignment_word_strings.
10839 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
10840 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10841 constant_alignment_word_strings.
10842 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
10843 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10844 constant_alignment_word_strings.
10845 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
10846 * config/microblaze/microblaze.c (microblaze_constant_alignment):
10848 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10849 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
10850 * config/mips/mips.c (mips_constant_alignment): New function.
10851 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10852 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
10853 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
10854 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10855 (mmix_constant_alignment): Make static. Use the same interface
10856 as the target hook.
10857 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
10858 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10859 constant_alignment_word_strings.
10860 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
10861 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10862 constant_alignment_word_strings.
10863 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
10864 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10865 constant_alignment_word_strings.
10866 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
10867 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10868 (rs6000_constant_alignment): New function.
10869 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
10870 * config/riscv/riscv.c (riscv_constant_alignment): New function.
10871 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10872 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
10873 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10874 (rs6000_constant_alignment): New function.
10875 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
10876 * config/s390/s390.c (s390_constant_alignment): New function.
10877 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10878 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
10879 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10880 constant_alignment_word_strings.
10881 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
10882 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10883 (sparc_constant_alignment): New function.
10884 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
10885 * config/spu/spu.c (spu_constant_alignment): New function.
10886 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10887 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
10888 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10889 constant_alignment_word_strings.
10890 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
10891 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10892 constant_alignment_word_strings.
10893 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
10894 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10895 constant_alignment_word_strings.
10896 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
10897 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10898 (visium_constant_alignment): New function.
10899 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
10900 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10901 (xtensa_constant_alignment): New function.
10902 * system.h (CONSTANT_ALIGNMENT): Poison.
10904 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
10906 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10907 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
10908 (rs6000_builtin_valid_without_lhs): New helper function.
10909 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10910 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
10912 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10914 * target.h (vec_perm_indices): Use unsigned short rather than
10916 (auto_vec_perm_indices): Likewise.
10917 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
10918 Use unsigned int rather than unsigned char.
10919 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
10921 2017-09-25 Richard Biener <rguenther@suse.de>
10923 * cfgloop.h (sort_sibling_loops): Declare.
10924 * cfgloop.c (sort_sibling_loops_cmp): New helper.
10925 (sort_sibling_loops): New function sorting the sibling loop list
10927 * graphite.c (graphite_transform_loops): Sort sibling loops.
10929 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
10931 * target.def (vec_perm_const_ok): Change sel parameter to
10933 * optabs-query.c (can_vec_perm_p): Update accordingly.
10934 * doc/tm.texi: Regenerate.
10935 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
10936 auto_vec_perm_indices and remove separate nelt field.
10937 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
10938 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
10939 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
10940 (aarch64_expand_vec_perm_const): Update accordingly.
10941 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
10942 to vec_perm_indices.
10943 * config/arm/arm.c (expand_vec_perm_d): Change perm to
10944 auto_vec_perm_indices and remove separate nelt field.
10945 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
10946 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
10947 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
10949 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
10950 to vec_perm_indices.
10951 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
10952 sel to vec_perm_indices.
10953 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
10954 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
10955 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
10957 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
10960 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
10963 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
10964 on the FUNCTION_DECL function context if it has a DIE that is a
10967 2017-09-25 Richard Biener <rguenther@suse.de>
10969 PR tree-optimization/82285
10970 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
10973 2017-09-25 Tom de Vries <tom@codesourcery.com>
10977 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
10980 2017-09-25 Richard Biener <rguenther@suse.de>
10982 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
10983 ISL errors other than isl_error_quota happen. Dump if the
10984 schedule is the same.
10985 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
10986 errors instead of aborting inside ISL.
10988 2017-09-25 Iain Sandoe <iain@codesourcery.com>
10991 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
10992 of libgcc_eh for m64.
10993 * config/i386/darwin64.h: Likewise.
10995 2017-09-25 Richard Biener <rguenther@suse.de>
10997 PR middle-end/82144
10998 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
10999 attribute for incomplete types nor twice for complete ones.
11001 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
11004 * config/i386/i386.c (ix86_print_operand_address_as): Encode
11005 %esp as %rsp to avoid 0x67 prefix if there is no index or base
11008 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
11011 * config/i386/i386.opt (mprefer-avx256): Use
11012 ix86_target_flags variable.
11013 * config/i386/i386.c (ix86_target_string): Move
11014 -mprefer-avx256 to flag2_opts.
11016 2017-09-22 Jakub Jelinek <jakub@redhat.com>
11018 PR middle-end/35691
11019 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
11020 and x != -1 | y != -1 into (x & y) != -1.
11022 2017-09-22 Steve Ellcey <sellcey@cavium.com>
11024 * config.gcc: Add new case statement to set
11025 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
11026 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
11027 s390x-*-linux* case statements. Added aarch64 to the list of
11028 supported architectures.
11030 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11032 PR tree-optimization/82289
11033 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
11034 STMT_VINFO_RELEVANT_P.
11036 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11037 Alan Hayward <alan.hayward@arm.com>
11038 David Sherwood <david.sherwood@arm.com>
11040 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
11041 for VR_RANGE only; don't allow VR_ANTI_RANGE.
11042 (extract_range_from_binary_expr_1): Don't call
11043 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
11045 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11046 Alan Hayward <alan.hayward@arm.com>
11047 David Sherwood <david.sherwood@arm.com>
11049 * target.def (preferred_vector_alignment): New hook.
11050 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
11052 * doc/tm.texi: Regenerate.
11053 * targhooks.h (default_preferred_vector_alignment): Declare.
11054 * targhooks.c (default_preferred_vector_alignment): New function.
11055 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
11057 (DR_TARGET_ALIGNMENT): New macro.
11058 (aligned_access_p): Update commentary.
11059 (vect_known_alignment_in_bytes): New function.
11060 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
11062 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
11063 Calculate the misalignment based on the target alignment rather than
11065 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
11066 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
11067 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
11068 the target alignment, rather than masking the element misalignment
11069 with the number of elements in a vector. Also use the target
11070 alignment when calculating the maximum number of peels.
11071 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
11072 instead of TYPE_ALIGN_UNIT.
11073 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
11074 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
11075 (vect_create_addr_base_for_vector_ref): Update call accordingly.
11076 (vect_create_data_ref_ptr): Likewise.
11077 (vect_setup_realignment): Realign by ANDing with
11078 -DR_TARGET_MISALIGNMENT.
11079 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
11080 the number of peels based on DR_TARGET_ALIGNMENT.
11081 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
11082 with the guaranteed alignment boundary when deciding whether
11084 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
11085 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
11086 (ensure_base_align): Remove stmt_info parameter. Get the
11087 target base alignment from DR_TARGET_ALIGNMENT.
11088 (vectorizable_store): Update call accordingly. Interpret
11089 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
11091 (vectorizable_load): Likewise.
11093 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11094 Alan Hayward <alan.hayward@arm.com>
11095 David Sherwood <david.sherwood@arm.com>
11097 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
11098 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
11099 (vect_enhance_data_refs_alignment): Likewise.
11101 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11103 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
11104 error. Only quit immediately if parsing is complete.
11105 (BEGIN): Initialize fatal_err and parse_done.
11106 (begin fpu, end fpu): Check number of arguments.
11107 (begin arch, end arch): Likewise.
11108 (begin cpu, end cpu): Likewise.
11109 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
11110 (optalias): Likewise.
11112 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11114 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
11115 * config/arm/arm-isa.h: Delete. Move definitions to ...
11116 * arm-cpus.in: ... here. Use new feature and fgroup values.
11117 * config/arm/arm.c (arm_option_override): Use lower case for feature
11119 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
11120 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
11121 * config/arm/parsecpu.awk (END): Add new command 'isa'.
11123 (print_isa_bits_for): New function.
11124 (gen_isa): New function.
11125 (gen_comm_data): Use print_isa_bits_for.
11126 (define feature): New keyword.
11127 (define fgroup): New keyword.
11128 * config/arm/t-arm (TM_H): Remove.
11129 (GTM_H): Add arm-isa.h.
11130 (arm-isa.h): Add rule to generate file.
11131 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
11132 case for feature bit names.
11134 2017-09-22 Richard Biener <rguenther@suse.de>
11136 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
11138 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
11139 print of no dependency loops ...
11140 * graphite.c (graphite_transform_loops): ... here.
11141 (canonicalize_loop_closed_ssa_form): Work from inner to outer
11143 (same_close_phi_node, remove_duplicate_close_phi,
11144 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
11145 (canonicalize_loop_closed_ssa): ... here and simplify.
11146 * graphite-optimize-isl.c: Include tree-vectorizer.h.
11147 (optimize_isl): Use dump_printf_loc to tell when we stopped
11148 optimizing because of an ISL timeout.
11150 2017-09-22 Richard Biener <rguenther@suse.de>
11152 PR tree-optimization/82291
11153 * tree-if-conv.c (predicate_mem_writes): Make sure to
11154 remove writes in blocks predicated with false.
11156 2017-09-22 Richard Biener <rguenther@suse.de>
11158 * sese.c: Include cfganal.h.
11159 (if_region_set_false_region): Remove.
11160 (create_if_region_on_edge): Likewise.
11161 (move_sese_in_condition): Re-implement without destroying
11164 2017-09-22 Richard Biener <rguenther@suse.de>
11166 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
11167 Verify both BBs contain loop PHI nodes before dispatching to
11168 copy_loop_phi_args.
11169 (graphite_regenerate_ast_isl): Do not recompute dominators,
11170 do not verify three times. Restructure for clarity.
11171 * graphite-scop-detection.c (same_close_phi_node,
11172 remove_duplicate_close_phi, make_close_phi_nodes_unique,
11173 defined_in_loop_p, canonicalize_loop_closed_ssa,
11174 canonicalize_loop_closed_ssa_form): Simplify, remove excess
11175 checking and SSA rewrite, move to ...
11176 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
11177 (graphite_initialize): Do not pass in ctx, do not reset the
11178 SCEV cache, compute only dominators.
11179 (graphite_transform_loops): Allocate ISL ctx after
11180 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
11181 Maintain post-dominators only around build_scops.
11182 * sese.c (if_region_set_false_region): Make static. Free
11183 and recompute dominators.
11184 (move_sese_in_condition): Assert we don't get called with
11185 post-dominators computed.
11186 * sese.h (if_region_set_false_region): Remove.
11188 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
11190 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
11191 mode attribute for TARGET_AVX512VL.
11193 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
11195 * config/i386/i386.opt (mprefer-avx256): New option.
11196 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
11198 (ix86_preferred_simd_mode): Return 256-bit AVX modes
11199 for TARGET_PREFER_AVX256.
11200 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
11202 2017-09-21 Jeff Law <law@redhat.com>
11204 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
11205 Fix dump output if the only stack space is for pushed registers.
11207 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11209 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
11212 2017-09-21 Martin Sebor <msebor@redhat.com>
11215 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
11216 code (in C++) or code that triggers warnings.
11218 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
11220 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
11222 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
11224 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
11225 * sched-rgn.c: Ditto.
11226 * sel-sched-ir.c: Ditto.
11228 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
11230 * toplev.h (set_random_seed): Adjust return type.
11231 * toplev.c (init_local_tick): Move eager initialization of random_seed
11232 to get_random_seed. Adjust comment.
11233 (init_random_seed): Inline to get_random_seed, delete.
11234 (get_random_seed): Initialize random_seed lazily.
11235 (set_random_seed): Do not return previous value.
11236 (print_switch_value): Do not call get_random_seed.
11238 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
11240 * cgraph.c (delete_function_version): New, broken out from...
11241 (cgraph_node::delete_function_version): ...here. Rename to
11242 cgraph_node::delete_function_version_by_decl. Update all uses.
11243 (cgraph_node::remove): Call delete_function_version.
11245 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11248 * tree-inline.c (expand_call_inline): Emit clobber stmts for
11249 VAR_DECLs to which addressable non-volatile parameters are mapped
11250 and for id->retvar after the return value assignment. Clear
11251 id->retval and id->retbnd after inlining.
11253 2017-09-21 Richard Biener <rguenther@suse.de>
11255 PR tree-optimization/82276
11256 PR tree-optimization/82244
11257 * tree-vrp.c (build_assert_expr_for): Set
11258 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
11260 (remove_range_assertions): Revert earlier change.
11262 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
11265 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
11267 2017-09-21 Richard Biener <rguenther@suse.de>
11269 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11270 Restore valid IL after code generation errors.
11271 * graphite.c (graphite_transform_loops): Diagnose code
11272 generation issues as MSG_MISSED_OPTIMIZATION and continue
11273 with processing SCOPs.
11275 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11276 Alan Hayward <alan.hayward@arm.com>
11277 David Sherwood <david.sherwood@arm.com>
11279 * calls.c (compute_argument_addresses): Use simplify_gen_binary
11280 rather than choosing between plus_constant and gen_rtx_<CODE>.
11281 * expr.c (emit_push_insn): Likewise.
11282 (expand_expr_real_2): Likewise.
11284 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11285 Alan Hayward <alan.hayward@arm.com>
11286 David Sherwood <david.sherwood@arm.com>
11288 * loop-unroll.c (split_iv): Call copy_rtx on the step.
11290 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11291 Alan Hayward <alan.hayward@arm.com>
11292 David Sherwood <david.sherwood@arm.com>
11294 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
11295 calling tree_to_uhwi.
11297 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11298 Alan Hayward <alan.hayward@arm.com>
11299 David Sherwood <david.sherwood@arm.com>
11301 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
11302 INTEGER_CST rather than a negative test for ADDR_EXPR.
11304 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11305 Alan Hayward <alan.hayward@arm.com>
11306 David Sherwood <david.sherwood@arm.com>
11308 * tree-vrp.c (extract_range_from_binary_expr_1): Check
11309 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
11311 2017-09-21 Richard Biener <rguenther@suse.de>
11313 PR tree-optimization/71351
11314 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
11315 graphite_create_new_loop_guard): Remove, fold remaining parts
11317 (translate_isl_ast_node_for): ... here and simplify.
11319 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11322 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
11323 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
11324 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
11325 alternative always use QI mode, for -Os imov (=R,R) alternative
11326 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
11329 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11330 Jeff Law <law@redhat.com>
11332 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
11333 (allocate_stack_space): New function, partially extracted from
11334 s390_emit_prologue.
11335 (s390_emit_prologue): Track offset to most recent stack probe.
11336 Code to allocate space moved into allocate_stack_space.
11337 Dump actions when no stack is allocated.
11338 (s390_prologue_plus_offset): New function.
11339 (s390_emit_stack_probe): Likewise.
11341 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
11343 * common.opt (Wa, Wl, Wp, g, gz=): Add
11345 (gno-column-info): Remove.
11346 (gcolumn-info): Drop RejectNegative.
11347 (gno-): New prefix.
11348 (gno-record-gcc-switches): Remove.
11349 (grecord-gcc-switches): Drop RejectNegative.
11350 (gno-split-dwarf): Remove.
11351 (gsplit-dwarf): Drop RejectNegative.
11352 (gno-strict-dwarf): Remove.
11353 (gstrict-dwarf): Drop RejectNegative.
11354 * config/darwin.opt (gfull, gused): Add RejectNegative.
11355 * dwarf2out.c (gen_producer_string): Drop
11356 gno-record-gcc-switches handler.
11357 * optc-gen.awk: Add g to prefixes with negative forms.
11358 * opts-common.c (remapping_prefix_p): New.
11359 (find_opt): Check it.
11360 (generate_canonical_option): Test g prefix.
11361 (option_map): Add -gno- mapping.
11362 (add_misspelling_candidates): Check remapping_prefix_p.
11364 2017-09-20 Jeff Law <law@redhat.com>
11366 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
11367 thinko in stack clash protection support.
11369 * explow.c (compute_stack_clash_protection_loop_data): Use
11370 CONST_INT_P instead of explicit test. Verify object is a
11371 CONST_INT_P before looking at INTVAL.
11372 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
11373 instead of explicit test.
11375 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
11378 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
11379 address instead of to r1 and r11.
11381 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
11383 * config.gcc: Support "knm".
11384 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
11385 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
11387 * config/i386/i386.c (m_KNM): Define.
11388 (processor_target_table): Add "knm".
11390 (ix86_option_override_internal): Add "knm".
11391 (ix86_issue_rate): Add PROCESSOR_KNM.
11392 (ix86_adjust_cost): Ditto.
11393 (ia32_multipass_dfa_lookahead): Ditto.
11394 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
11395 (fold_builtin_cpu): Add M_INTEL_KNM.
11396 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
11397 (processor_type): Add PROCESSOR_KNM.
11398 * config/i386/x86-tune.def: Add m_KNM.
11399 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
11401 2017-09-20 Richard Biener <rguenther@suse.de>
11403 PR tree-optimization/80213
11404 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
11405 are allowed in empty BBs as well.
11406 (canonicalize_loop_closed_ssa): Also look for other complex
11408 (scop_detection::get_sese): Include the loop-closed PHI block
11410 (scop_detection::merge_sese): Remove code adding extra blocks.
11411 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
11412 (build_scops): Assert the final returned scop is invalid.
11414 2017-09-20 Richard Biener <rguenther@suse.de>
11416 PR tree-optimization/82264
11417 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
11419 (vn_phi_lookup): Likewise.
11420 (vn_phi_insert): Likewise.
11422 2017-09-20 Jakub Jelinek <jakub@redhat.com>
11424 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
11425 that fits into uhwi or shwi, add DW_AT_const_value regardless
11426 of early_dwarf without going through RTL, using add_AT_unsigned
11429 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
11430 (DEBUG_LTO_ABBREV_SECTION): Likewise.
11431 (DEBUG_LTO_MACINFO_SECTION): Likewise.
11432 (DEBUG_MACRO_SECTION): Likewise.
11433 (DEBUG_LTO_MACRO_SECTION): Likewise.
11434 (DEBUG_STR_DWO_SECTION): Likewise.
11435 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
11436 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
11437 (DEBUG_LTO_DWO_LINE_SECTION): Define.
11438 (DEBUG_LTO_LINE_STR_SECTION): Define.
11439 (init_sections_and_labels): Initialize debug_line_str_section
11440 variable. Initialize debug_loc_section for -gdwarf-5 to
11441 DEBUG_LOCLISTS_SECTION. Formatting fixes.
11443 2017-09-20 Richard Biener <rguenther@suse.de>
11445 * graphite-sese-to-poly.c (extract_affine): Properly handle
11446 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
11448 2017-09-20 Richard Biener <rguenther@suse.de>
11450 PR tree-optimization/81373
11451 * graphite-scop-detection.c (build_cross_bb_scalars_def):
11452 Force SESE live-out defs to be handled even if they are
11455 2017-09-19 Jeff Law <law@redhat.com>
11457 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
11458 nothing for stack adjustments with REG_STACK_CHECK.
11459 * sched-deps.c (parse_add_or_inc): Reject insns with
11460 REG_STACK_CHECK from dependency breaking.
11461 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
11462 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
11463 * reg-notes.def (STACK_CHECK): New note.
11465 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
11466 (ix86_expand_prologue): Dump stack clash info as needed.
11467 Call ix86_adjust_stack_and_probe_stack_clash as needed.
11469 * function.c (dump_stack_clash_frame_info): New function.
11470 * function.h (dump_stack_clash_frame_info): Prototype.
11471 (enum stack_clash_probes): New enum.
11473 * config/alpha/alpha.c (alpha_expand_prologue): Also check
11474 flag_stack_clash_protection.
11475 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
11476 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
11477 (arm_frame_pointer_required): Likewise.
11478 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
11479 (ia64_expand_prologue): Likewise.
11480 * config/mips/mips.c (mips_expand_prologue): Likewise.
11481 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
11482 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11483 (sparc_flat_expand_prologue): Likewise.
11484 * config/spu/spu.c (spu_expand_prologue): Likewise.
11486 * explow.c: Include "params.h".
11487 (anti_adjust_stack_and_probe_stack_clash): New function.
11488 (get_stack_check_protect): Likewise.
11489 (compute_stack_clash_protection_loop_data): Likewise.
11490 (emit_stack_clash_protection_loop_start): Likewise.
11491 (emit_stack_clash_protection_loop_end): Likewise.
11492 (allocate_dynamic_stack_space): Use get_stack_check_protect.
11493 Use anti_adjust_stack_and_probe_stack_clash.
11494 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
11495 (emit_stack_clash_protection_loop_start): Likewise.
11496 (emit_stack_clash_protection_loop_end): Likewise.
11497 * rtl.h (get_stack_check_protect): Prototype.
11498 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
11499 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
11500 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
11502 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
11504 * doc/tm.texi: Rebuilt.
11505 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
11506 get_stack_check_protect.
11507 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
11508 * config/arm/arm.c (arm_expand_prologue): Likewise.
11509 (arm_frame_pointer_required): Likewise.
11510 * config/i386/i386.c (ix86_expand_prologue): Likewise.
11511 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
11512 * config/mips/mips.c (mips_expand_prologue): Likewise.
11513 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
11514 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
11515 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11516 (sparc_flat_expand_prologue): Likewise.
11518 * common.opt (-fstack-clash-protection): New option.
11519 * flag-types.h (enum stack_check_type): Note difference between
11520 -fstack-check= and -fstack-clash-protection.
11521 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
11522 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
11523 * toplev.c (process_options): Issue warnings/errors for cases
11524 not handled with -fstack-clash-protection.
11525 * doc/invoke.texi (-fstack-clash-protection): Document new option.
11526 (-fstack-check): Note additional problem with -fstack-check=generic.
11527 Note that -fstack-check is primarily for Ada and refer users
11528 to -fstack-clash-protection for stack-clash-protection.
11529 Document new params for stack clash protection.
11531 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
11533 * config/i386/i386.c (ix86_split_long_move): Do not handle
11534 address used for LEA in a special way.
11536 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
11538 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
11540 2017-09-19 Martin Sebor <msebor@redhat.com>
11543 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
11544 of incompatible types.
11546 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
11548 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11549 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
11550 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11551 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
11553 2017-09-19 Richard Biener <rguenther@suse.de>
11555 PR tree-optimization/82244
11556 * tree-vrp.c (remove_range_assertions): Do not propagate
11557 a constant to abnormals but replace the assert with a copy.
11559 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
11561 PR rtl-optimization/57878
11562 PR rtl-optimization/68988
11563 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
11564 avoidance test involving non_reload_pseudos. Move frequency test
11565 below the general fragmentation avoidance test.
11567 2017-09-19 Richard Biener <rguenther@suse.de>
11569 PR tree-optimization/69728
11570 * graphite-sese-to-poly.c (schedule_error): New global.
11571 (add_loop_schedule): Handle empty domain by failing the
11573 (build_original_schedule): Handle schedule_error.
11575 2017-09-19 Richard Biener <rguenther@suse.de>
11577 * graphite-scop-detection.c (scop_detection::can_represent_loop):
11578 Do not iterate to sibling loops but only to siblings of inner
11581 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
11584 * config/m68k/m68k.md (moveq feeding equality comparison): Check
11585 that the registers are different.
11587 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
11589 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
11590 to processor_model and "amdfam17h" to arch_names_table.
11591 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
11593 2017-09-18 Jakub Jelinek <jakub@redhat.com>
11596 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
11598 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11599 Alan Hayward <alan.hayward@arm.com>
11600 David Sherwood <david.sherwood@arm.com>
11602 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
11604 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
11606 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
11607 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
11608 vect_schedule_slp_instance.
11609 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
11610 Update call to vect_slp_analyze_node_operations. Simplify return
11612 (vect_slp_analyze_bb_1): Update call accordingly.
11613 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
11614 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
11615 (vect_schedule_slp): Update call accordingly.
11617 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11618 Alan Hayward <alan.hayward@arm.com>
11619 David Sherwood <david.sherwood@arm.com>
11621 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
11622 with types that aren't in fact scalar.
11624 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11626 * tree-vect-slp.c (vect_record_max_nunits): New function,
11628 (vect_build_slp_tree_1): ...here.
11629 (vect_build_slp_tree_2): Call it for phis too.
11631 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11633 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
11634 to vect_get_vec_def_for_operand when getting the mask operand.
11636 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11637 Alan Hayward <alan.hayward@arm.com>
11638 David Sherwood <david.sherwood@arm.com>
11640 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
11643 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11644 Alan Hayward <alan.hayward@arm.com>
11645 David Sherwood <david.sherwood@arm.com>
11647 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
11648 calculation for vector booleans.
11650 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11651 Alan Hayward <alan.hayward@arm.com>
11652 David Sherwood <david.sherwood@arm.com>
11654 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
11656 (vect_transform_stmt): ...here.
11657 (vect_analyze_stmt): Use it instead of calling
11658 vectorizable_live_operation directly.
11660 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
11662 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
11663 non-SIMT targets in acc vector loops.
11665 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
11667 * configure.ac: Add arc and check if assembler supports gdwarf2.
11668 * configure: Regenerate.
11670 2017-09-18 Richard Biener <rguenther@suse.de>
11672 PR tree-optimization/82220
11673 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
11674 epilogue niters from the min_profitable_iters compute.
11676 2017-09-18 Jakub Jelinek <jakub@redhat.com>
11679 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
11680 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
11681 (ix86_init_pic_reg): Revert 2017-09-01 changes.
11683 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
11686 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
11687 switching to a new text section.
11689 2017-09-18 Richard Biener <rguenther@suse.de>
11691 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
11693 (build_alias_set): Reject aliases with no access function.
11695 2017-09-18 Richard Biener <rguenther@suse.de>
11697 PR tree-optimization/79622
11698 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
11700 (build_cross_bb_scalars_use): Likewise.
11702 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
11704 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
11706 2017-09-18 Alan Modra <amodra@gmail.com>
11709 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
11710 stack_pointer_rtx for count 0. Update comments. Break up
11711 large rtl expression.
11713 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
11715 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
11716 Increase to 20 bytes.
11717 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
11718 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
11719 or avx version of the stub.
11721 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
11724 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
11725 compute the minimum stack alignment. Also update preferred stack
11726 boundary for leaf functions.
11728 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
11730 PR tree-optimization/82228
11731 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
11734 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
11736 * common/config/nds32/nds32-common.c
11737 (nds32_option_optimization_table): Refine formatting.
11738 (nds32_option_optimization_table): Use -fsched-pressure and
11739 -fomit-frame-pointer for specific optimization level.
11741 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
11743 * config/nds32/nds32.c: Refine formatting and comments.
11744 * config/nds32/nds32.h: Likewise.
11745 * config/nds32/nds32.md: Likewise.
11746 * config/nds32/nds32-cost.c: Likewise.
11747 * config/nds32/nds32-isr.c: Likewise.
11748 * config/nds32/nds32-md-auxiliary.c: Likewise.
11749 * config/nds32/nds32-multiple.md: Likewise.
11750 * config/nds32/nds32-predicates.c: Likewise.
11752 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
11753 Jakub Jelinek <jakub@redhat.com>
11755 Add support for -std=c++2a.
11756 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
11758 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
11760 2017-09-15 Steve Ellcey <sellcey@cavium.com>
11763 * doc/extend.texi (Common Function Attributes): Add
11764 references to ARM, AArch64, and S/390 specific attributes.
11765 (Function Specific Option Pragmas): Add AArch64 and S/390
11766 to list of back ends that support the target pragma.
11768 2017-09-15 Nathan Sidwell <nathan@acm.org>
11770 * doc/standards.texi: Fix C++17 description. Update URLs for
11773 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
11775 * common.opt (Wcast-align=strict): New warning option.
11776 * doc/invoke.texi: Document -Wcast-align=strict.
11778 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
11780 * cgraph.h (cgraph_thunk_info): Add comments.
11781 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
11782 assert for VIRTUAL_* arguments stricter.
11784 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
11786 PR tree-optimization/71026
11787 * match.pd: Move RDIV patterns from fold-const.c
11788 * fold-const.c (distribute_real_division): Removed.
11789 (fold_binary_loc): Remove calls to distribute_real_divison.
11791 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11793 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
11794 c++1z and gnu++1z as deprecated. Change other references to
11795 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
11796 Change -Wc++1z-compat to -Wc++17-compat.
11797 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
11798 * dwarf2out.c (highest_c_language): Handle C++17.
11799 (gen_compile_unit_die): Likewise.
11801 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11803 PR rtl-optimization/82192
11804 * combine.c (make_extraction): Don't look through non-paradoxical
11805 SUBREGs or TRUNCATE if pos + len is or might be bigger than
11808 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11809 Alan Hayward <alan.hayward@arm.com>
11810 David Sherwood <david.sherwood@arm.com>
11812 * target.def (function_arg_offset): New hook.
11813 * targhooks.h (default_function_arg_offset): Declare.
11814 * targhooks.c (default_function_arg_offset): New function.
11815 * function.c (locate_and_pad_parm): Use
11816 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
11817 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
11818 (TARGET_FUNCTION_ARG_OFFSET): ...this.
11819 * doc/tm.texi: Regenerate.
11820 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
11821 * config/spu/spu.c (spu_function_arg_offset): New function.
11822 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
11823 * system.h (FUNCTION_ARG_OFFSET): Poison.
11825 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11826 Alan Hayard <alan.hayward@arm.com>
11827 David Sherwood <david.sherwood@arm.com>
11829 * target.def (truly_noop_truncation): New hook.
11830 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
11831 than TRULY_NOOP_TRUNCATION.
11832 * hooks.h (hook_bool_uint_uint_true): Declare.
11833 * hooks.c (hook_bool_uint_uint_true): New function.
11834 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
11835 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
11836 * doc/tm.texi: Regenerate.
11837 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
11838 rather than TRULY_NOOP_TRUNCATION in comments.
11839 (simplify_comparison): Likewise.
11840 (record_truncated_value): Likewise.
11841 * expmed.c (extract_bit_field_1): Likewise.
11842 (extract_split_bit_field): Likewise.
11843 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
11844 instead of TRULY_NOOP_TRUNCATION.
11845 * function.c (assign_parm_setup_block): Likewise.
11846 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
11847 * rtlhooks.c: Include target.h.
11848 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
11849 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
11850 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
11851 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
11852 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
11853 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
11854 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
11855 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
11856 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
11857 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
11858 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
11859 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
11860 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
11861 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
11862 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
11863 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
11864 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
11865 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
11866 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
11867 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
11868 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
11869 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
11870 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
11871 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
11872 * config/mips/mips.c (mips_truly_noop_truncation): New function.
11873 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11874 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11875 rather than TRULY_NOOP_TRUNCATION in comments.
11876 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
11877 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
11878 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
11879 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
11880 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
11881 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
11882 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
11883 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
11884 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
11885 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
11886 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
11887 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11888 rather than TRULY_NOOP_TRUNCATION in comments.
11889 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
11890 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
11891 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
11892 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
11893 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
11894 TRULY_NOOP_TRUNCATION condition.
11895 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
11896 (TRULY_NOOP_TRUNCATION): Delete.
11897 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
11898 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
11899 * config/spu/spu.c (spu_truly_noop_truncation): New function.
11900 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11901 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
11902 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
11903 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
11904 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11905 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11906 rather than TRULY_NOOP_TRUNCATION in comments.
11907 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
11908 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
11909 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
11910 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
11911 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
11912 * system.h (TRULY_NOOP_TRUNCATION): Poison.
11914 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
11917 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
11918 (*cmp_ior): Likewise.
11919 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
11920 (*ior_scc_scc_cmp): Likewise.
11921 (*and_scc_scc): Likewise.
11922 (*and_scc_scc_cmp): Likewise.
11924 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11925 Alan Hayard <alan.hayward@arm.com>
11926 David Sherwood <david.sherwood@arm.com>
11928 * target.def (can_change_mode_class): New hook.
11929 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
11930 (hard_regno_nregs): Likewise.
11931 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
11932 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
11933 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
11934 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
11935 (LOAD_EXTEND_OP): Update accordingly.
11936 * doc/tm.texi: Regenerate.
11937 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
11938 CANNOT_CHANGE_MODE_CLASS.
11939 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
11940 (REG_CAN_CHANGE_MODE_P): ...this new macro.
11941 * combine.c (simplify_set): Update accordingly.
11942 * emit-rtl.c (validate_subreg): Likewise.
11943 * recog.c (general_operand): Likewise.
11944 * regcprop.c (mode_change_ok): Likewise.
11945 * reload1.c (choose_reload_regs): Likewise.
11946 (inherit_piecemeal_p): Likewise.
11947 * rtlanal.c (simplify_subreg_regno): Likewise.
11948 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
11949 instead of CANNOT_CHANGE_MODE_CLASS.
11950 (reload_cse_simplify_operands): Likewise.
11951 * reload.c (push_reload): Use targetm.can_change_mode_class
11952 instead of CANNOT_CHANGE_MODE_CLASS.
11953 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
11954 REG_CANNOT_CHANGE_MODE_P.
11955 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11956 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
11957 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11958 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11959 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11960 (arm_can_change_mode_class): New function.
11961 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11962 than CANNOT_CHANGE_MODE_CLASS in comments.
11963 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11964 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
11965 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
11966 (ix86_can_change_mode_class): ...this new function, inverting the
11967 sense of the return value.
11968 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11969 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11970 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11971 (ia64_can_change_mode_class): New function.
11972 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11973 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
11974 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
11975 (m32c_can_change_mode_class): ...this new function, inverting the
11976 sense of the return value.
11977 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11978 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11979 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
11980 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
11981 (mips_can_change_mode_class): ...this new function, inverting the
11982 sense of the return value.
11983 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11984 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11985 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11986 (msp430_can_change_mode_class): New function.
11987 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11988 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
11989 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11990 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11991 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11992 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
11993 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11994 (pa_cannot_change_mode_class): Replace with...
11995 (pa_can_change_mode_class): ...this new function, inverting the
11996 sense of the return value.
11997 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11998 than CANNOT_CHANGE_MODE_CLASS in comments.
11999 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12000 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
12001 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12002 (pdp11_cannot_change_mode_class): Replace with...
12003 (pdp11_can_change_mode_class): ...this new function, inverting the
12004 sense of the return value.
12005 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12006 * config/powerpcspe/powerpcspe-protos.h
12007 (rs6000_cannot_change_mode_class_ptr): Delete.
12008 * config/powerpcspe/powerpcspe.c
12009 (rs6000_cannot_change_mode_class_ptr): Delete.
12010 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12011 (rs6000_option_override_internal): Assign to
12012 targetm.can_change_mode_class instead of
12013 rs6000_cannot_change_mode_class_ptr.
12014 (rs6000_cannot_change_mode_class): Replace with...
12015 (rs6000_can_change_mode_class): ...this new function, inverting the
12016 sense of the return value.
12017 (rs6000_debug_cannot_change_mode_class): Replace with...
12018 (rs6000_debug_can_change_mode_class): ...this new function.
12019 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12020 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
12021 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12022 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12023 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
12025 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
12026 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12027 (rs6000_option_override_internal): Assign to
12028 targetm.can_change_mode_class instead of
12029 rs6000_cannot_change_mode_class_ptr.
12030 (rs6000_cannot_change_mode_class): Replace with...
12031 (rs6000_can_change_mode_class): ...this new function, inverting the
12032 sense of the return value.
12033 (rs6000_debug_cannot_change_mode_class): Replace with...
12034 (rs6000_debug_can_change_mode_class): ...this new function.
12035 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12036 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
12037 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
12038 (s390_can_change_mode_class): ...this new function, inverting the
12039 sense of the return value.
12040 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12041 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12042 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
12043 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12044 (sh_cannot_change_mode_class): Replace with...
12045 (sh_can_change_mode_class): ...this new function, inverting the
12046 sense of the return value.
12047 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12048 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12049 (sparc_can_change_mode_class): New function.
12050 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12051 * config/spu/spu.c (spu_can_change_mode_class): New function.
12052 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12053 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12054 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12055 (visium_can_change_mode_class): New function.
12056 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
12058 2017-09-15 Richard Biener <rguenther@suse.de>
12060 PR tree-optimization/82217
12061 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
12062 but not undefined case.
12064 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12067 * postreload.c (reload_cse_simplify_operands): Skip
12068 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
12070 2017-09-15 Richard Biener <rguenther@suse.de>
12072 PR tree-optimization/68823
12073 * graphite-scop-detection.c (build_alias_set): If we have a
12074 possible dependence check whether we can handle them by just
12075 looking at the DRs DR_ACCESS_FNs.
12076 (build_scops): If build_alias_set fails, fail the SCOP.
12078 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
12080 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
12081 to support float128 built-in functions that require the ISA 3.0
12083 (BU_FLOAT128_3_HW): Likewise.
12084 (SQRTF128): Add support for the IEEE 128-bit square root and fma
12085 built-in functions.
12086 (FMAF128): Likewise.
12088 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
12089 support for built-in functions that need the ISA 3.0 IEEE 128-bit
12090 floating point instructions.
12091 (rs6000_invalid_builtin): Likewise.
12092 (rs6000_builtin_mask_names): Likewise.
12093 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
12094 (RS6000_BTM_FLOAT128_HW): Likewise.
12095 (RS6000_BTM_COMMON): Likewise.
12096 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
12098 * doc/extend.texi (RS/6000 built-in functions): Document the
12099 IEEE 128-bit floating point square root and fused multiply-add
12100 built-in functions.
12102 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
12104 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
12105 reg (r2) isn't in the set of registers defined in the prologue.
12107 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12108 Alan Hayward <alan.hayward@arm.com>
12109 David Sherwood <david.sherwood@arm.com>
12111 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
12112 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
12113 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
12114 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
12115 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
12117 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
12118 max_vectorization_factor.
12119 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
12121 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12122 Alan Hayward <alan.hayward@arm.com>
12123 David Sherwood <david.sherwood@arm.com>
12125 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
12126 (vect_worthwhile_without_simd_p): Declare.
12127 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
12128 (vectorizable_reduction): Use it.
12129 * tree-vect-stmts.c (vectorizable_shift): Likewise.
12130 (vectorizable_operation): Likewise.
12132 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12133 Alan Hayward <alan.hayward@arm.com>
12134 David Sherwood <david.sherwood@arm.com>
12136 * tree-vectorizer.h (vect_get_num_copies): New function.
12137 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
12138 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12139 (vectorizable_induction): Likewise.
12140 (vectorizable_live_operation): Likewise.
12141 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12142 (vectorizable_bswap): Likewise.
12143 (vectorizable_call): Likewise.
12144 (vectorizable_conversion): Likewise.
12145 (vectorizable_assignment): Likewise.
12146 (vectorizable_shift): Likewise.
12147 (vectorizable_operation): Likewise.
12148 (vectorizable_store): Likewise.
12149 (vectorizable_load): Likewise.
12150 (vectorizable_condition): Likewise.
12151 (vectorizable_comparison): Likewise.
12152 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
12154 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12155 Alan Hayward <alan.hayward@arm.com>
12156 David Sherwood <david.sherwood@arm.com>
12158 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
12159 of vect_init_vector.
12161 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12162 Alan Hayward <alan.hayward@arm.com>
12163 David Sherwood <david.sherwood@arm.com>
12165 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
12166 an inline wrapper that provides a location.
12167 (gimple_build_vector): Likewise.
12168 * gimple-fold.c (gimple_build_vector_from_val): New function.
12169 (gimple_build_vector): Likewise.
12170 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
12171 functions to build the initial value. Always return a gimple value.
12172 (get_initial_defs_for_reduction): Likewise. Only compute
12174 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
12175 vect_init_vector on the results from get_initial_def(s)_for_reduction.
12176 (vectorizable_induction): Use gimple_build_vector rather than
12179 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12180 Alan Hayward <alan.hayward@arm.com>
12181 David Sherwood <david.sherwood@arm.com>
12183 * target.h (vec_perm_indices): New typedef.
12184 (auto_vec_perm_indices): Likewise.
12185 * optabs-query.h: Include target.h
12186 (can_vec_perm_p): Take a vec_perm_indices *.
12187 * optabs-query.c (can_vec_perm_p): Likewise.
12188 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
12189 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12190 * tree-vect-generic.c (lower_vec_perm): Likewise.
12191 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
12192 (vect_grouped_load_supported): Likewise.
12193 (vect_shift_permute_load_chain): Likewise.
12194 (vect_permute_store_chain): Use auto_vec_perm_indices.
12195 (vect_permute_load_chain): Likewise.
12196 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
12197 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
12198 Update uses of can_vec_perm_p.
12199 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
12200 mode with a number of elements. Take a vec_perm_indices *.
12201 (vect_create_epilog_for_reduction): Update accordingly.
12202 Use auto_vec_perm_indices.
12203 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
12204 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
12205 (vect_transform_slp_perm_load): Likewise.
12206 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
12207 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
12208 (vect_gen_perm_mask_checked): Likewise.
12209 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
12210 (vect_gen_perm_mask_checked): Likewise.
12211 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
12212 (vectorizable_store): Likewise.
12213 (vectorizable_load): Likewise.
12214 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
12215 (vectorizable_bswap): Likewise.
12217 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12218 Alan Hayward <alan.hayward@arm.com>
12219 David Sherwood <david.sherwood@arm.com>
12221 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
12222 * tree.c (build_vector): Likewise.
12223 (build_vector_from_ctor): Update accordingly.
12224 (build_vector_from_val): Likewise.
12225 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
12226 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12227 * tree-vect-generic.c (add_rshift): Likewise.
12228 (expand_vector_divmod): Likewise.
12229 (optimize_vector_constructor): Likewise.
12230 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
12231 (vect_transform_slp_perm_load): Likewise.
12232 (vect_schedule_slp_instance): Likewise.
12233 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
12234 (vectorizable_call): Likewise.
12235 (vect_gen_perm_mask_any): Likewise. Add elements in order.
12236 * expmed.c (make_tree): Likewise.
12237 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
12238 a vector passed to build_vector.
12239 (fold_convert_const): Likewise.
12240 (exact_inverse): Likewise.
12241 (fold_ternary_loc): Likewise.
12242 (fold_relational_const): Likewise.
12243 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
12244 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
12245 (const_unop): Likewise. Store the reduction accumulator in a
12246 variable rather than an array.
12247 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
12248 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
12249 the new vector, rather than constructing it after the input arrays.
12250 (native_interpret_vector): Use auto_vec<tree> when building
12251 a vector passed to build_vector. Add elements in order.
12252 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
12253 auto_vec<tree> when building a vector passed to build_vector.
12254 (vect_create_epilog_for_reduction): Likewise.
12255 (vectorizable_induction): Likewise.
12256 (get_initial_def_for_reduction): Likewise. Fix indentation of
12258 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
12260 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
12261 passed to build_vector.
12263 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12264 Alan Hayward <alan.hayward@arm.com>
12265 David Sherwood <david.sherwood@arm.com>
12267 * tree-core.h (tree_base::u): Add an "nelts" field.
12268 (tree_vector): Use VECTOR_CST_NELTS as the length.
12269 * tree.c (tree_size): Likewise.
12270 (make_vector): Initialize VECTOR_CST_NELTS.
12271 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
12272 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
12273 TYPE_VECTOR_SUBPARTS.
12274 * expr.c (const_vector_mask_from_tree): Consistently use "units"
12275 as the number of units, setting it from VECTOR_CST_NELTS.
12276 (const_vector_from_tree): Likewise.
12277 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
12278 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
12279 (fold_negate_expr_1): Likewise.
12280 (fold_convert_const): Likewise.
12281 (const_binop): Likewise. Differentiate the number of output and
12283 (const_unop): Likewise.
12284 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
12285 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
12286 in cases that did the opposite.
12288 2017-09-14 Richard Biener <rguenther@suse.de>
12290 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
12293 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
12295 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
12297 2017-09-14 Jakub Jelinek <jakub@redhat.com>
12300 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
12301 if and where to split a bb, except for splitting before debug insn
12302 sequences followed by non-label real insn. Delete debug insns
12303 in between basic blocks.
12305 * combine.c (make_compound_operation_int): Formatting fixes.
12307 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
12308 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12309 * config/netbsd.h (LINK_EH_SPEC): Likewise.
12310 * config/sol2.h (LINK_EH_SPEC): Likewise.
12311 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12312 * config/s390/linux.h (LINK_SPEC): Likewise.
12313 * config/freebsd.h (LINK_EH_SPEC): Likewise.
12314 * config/openbsd.h (LINK_EH_SPEC): Likewise.
12315 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12316 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
12317 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
12318 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12319 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
12320 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
12322 2017-09-13 Jakub Jelinek <jakub@redhat.com>
12324 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
12326 (ENDFILE_LINUX_SPEC): Likewise.
12327 (LINK_EH_SPEC): Likewise.
12328 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
12329 (LINK_OS_LINUX_SPEC32): Likewise.
12330 (LINK_OS_LINUX_SPEC64): Likewise.
12331 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
12332 (LINK_OS_LINUX_SPEC): Likewise.
12334 2017-09-13 Martin Liska <mliska@suse.cz>
12336 PR middle-end/82154
12337 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
12338 CASE_HIGH is NULL_TREE.
12340 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12341 Alan Hayward <alan.hayward@arm.com>
12342 David Sherwood <david.sherwood@arm.com>
12344 * target.def (secondary_memory_needed): New hook.
12345 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
12346 instead of SECONDARY_MEMORY_NEEDED.
12347 (secondary_memory_needed_mode): Likewise.
12348 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
12349 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
12350 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
12351 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
12352 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
12353 * doc/tm.texi: Regenerate.
12354 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
12355 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
12356 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12357 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
12358 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
12359 * config/i386/i386.c (inline_secondary_memory_needed): Put the
12360 mode argument first and change the reg_class arguments to reg_class_t.
12361 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
12362 Make static. Update the call to inline_secondary_memory_needed.
12363 (ix86_register_move_cost): Update the call to
12364 inline_secondary_memory_needed.
12365 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12366 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
12368 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
12369 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12371 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
12372 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
12373 * config/mips/mips.c (mips_secondary_memory_needed): Make static
12374 and match hook interface. Add comment from mips.h.
12375 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12376 * config/mmix/mmix.md (truncdfsf2): Refer to
12377 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12379 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
12380 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
12381 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12382 (pa_secondary_memory_needed): New function.
12383 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
12384 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
12385 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12386 (pdp11_secondary_memory_needed): Make static and match hook interface.
12387 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
12388 * config/powerpcspe/powerpcspe-protos.h
12389 (rs6000_secondary_memory_needed_ptr): Delete.
12390 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
12392 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12393 (rs6000_option_override_internal): Assign to
12394 targetm.secondary_memory_needed rather than
12395 rs6000_secondary_memory_needed_ptr.
12396 (rs6000_secondary_memory_needed): Match hook interface.
12397 (rs6000_debug_secondary_memory_needed): Likewise.
12398 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
12399 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
12400 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
12401 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12402 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
12403 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
12405 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
12406 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12407 (rs6000_option_override_internal): Assign to
12408 targetm.secondary_memory_needed rather than
12409 rs6000_secondary_memory_needed_ptr.
12410 (rs6000_secondary_memory_needed): Match hook interface.
12411 (rs6000_debug_secondary_memory_needed): Likewise.
12412 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
12413 * config/s390/s390.c (s390_secondary_memory_needed): New function.
12414 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12415 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
12416 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12417 (sparc_secondary_memory_needed): New function.
12418 * lra-constraints.c (check_and_process_move): Refer to
12419 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12421 (curr_insn_transform): Likewise.
12422 (process_alt_operands): Use targetm.secondary_memory_needed
12423 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12424 (check_secondary_memory_needed_p): Likewise.
12425 (choose_split_class): Likewise.
12426 * reload.c: Unconditionally include code that was previously
12427 conditional on SECONDARY_MEMORY_NEEDED.
12428 (push_secondary_reload): Use targetm.secondary_memory_needed
12429 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12430 (push_reload): Likewise.
12431 * reload1.c: Unconditionally include code that was previously
12432 conditional on SECONDARY_MEMORY_NEEDED.
12433 (choose_reload_regs): Use targetm.secondary_memory_needed
12434 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12435 (gen_reload): Likewise.
12436 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
12438 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12439 Alan Hayward <alan.hayward@arm.com>
12440 David Sherwood <david.sherwood@arm.com>
12442 * target.def (secondary_memory_needed_mode): New hook:
12443 * targhooks.c (default_secondary_memory_needed_mode): Declare.
12444 * targhooks.h (default_secondary_memory_needed_mode): New function.
12445 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
12446 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
12447 * doc/tm.texi: Regenerate.
12448 * lra-constraints.c (check_and_process_move): Use
12449 targetm.secondary_memory_needed_mode instead of
12450 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
12451 (curr_insn_transform): Likewise.
12452 * reload.c (get_secondary_mem): Likewise.
12453 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12454 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
12456 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12457 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12458 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
12459 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12460 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
12462 * config/powerpcspe/powerpcspe-protos.h
12463 (rs6000_secondary_memory_needed_mode): Delete.
12464 * config/powerpcspe/powerpcspe.c
12465 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12466 (rs6000_secondary_memory_needed_mode): Make static.
12467 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12468 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
12470 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12472 (rs6000_secondary_memory_needed_mode): Make static.
12473 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12474 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
12475 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12476 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12477 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12479 (sparc_secondary_memory_needed_mode): New function.
12480 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
12482 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
12484 * config/aarch64/constraints.md (Umq): New constraint.
12485 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
12487 (mov<mode>): Update condition.
12489 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12491 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
12492 when bitposition is the same.
12494 2017-09-13 Richard Biener <rguenther@suse.de>
12496 * dwarf2out.c (output_die_symbol): Remove.
12497 (output_die): Do not output a DIEs symbol.
12499 2017-09-13 Richard Biener <rguenther@suse.de>
12501 PR middle-end/82128
12502 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
12503 default-def to avoid breaking iterator update with the weird
12504 interaction with cgraph_update_edges_for_call_stmt_node.
12506 2017-09-13 Richard Biener <rguenther@suse.de>
12508 * tree-cfg.c (verify_gimple_assign_binary): Add verification
12509 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
12510 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
12511 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
12513 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
12515 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
12516 Disable pc relative literal load irrespective of
12517 TARGET_FIX_ERR_A53_84341 for default.
12519 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
12521 * config/sparc/sparc.c (output_return): Output the source location of
12522 the insn in the delay slot, if any.
12523 (output_sibcall): Likewise.
12525 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
12528 * common.opt (-static-pie): New alias.
12529 (shared): Negate static-pie.
12530 (-no-pie): Update help text.
12532 (static-pie): New option.
12533 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
12534 -static-pie support.
12535 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
12536 (LINK_EH_SPEC): Likewise.
12537 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12538 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
12539 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
12540 * gcc.c (LINK_COMMAND_SPEC): Likewise.
12541 (init_gcc_specs): Likewise.
12542 (init_spec): Likewise.
12543 (display_help): Update help message for -pie.
12544 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
12547 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
12549 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
12550 (movdi_aarch64): Likewise.
12551 (movti_aarch64): Likewise.
12553 2017-09-12 Simon Wright <simon@pushface.org>
12556 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
12557 calculation of the minor version, always output as 0.
12559 2017-09-12 Jakub Jelinek <jakub@redhat.com>
12562 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
12563 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
12564 on it early, rather than manual conversion late. For
12565 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
12566 instead of performing manual conversion.
12568 2017-09-12 Carl Love <cel@us.ibm.com>
12570 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
12571 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
12573 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12574 VMULOSW): Add definitions.
12575 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12576 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12577 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12578 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12579 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12581 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
12583 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
12585 (movti_aarch64): Likewise.
12586 (movdf_aarch64): Likewise.
12587 (movtf_aarch64): Likewise.
12588 (load_pairdi): Likewise.
12589 (store_pairdi): Likewise.
12590 (load_pairdf): Likewise.
12591 (store_pairdf): Likewise.
12592 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
12593 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
12594 (ldr_got_small_<mode>): Likewise.
12595 (ldr_got_small_28k_<mode>): Likewise.
12596 (ldr_got_tiny): Likewise.
12597 * config/aarch64/iterators.md (ldst_sz): New.
12598 (ldpstp_sz): Likewise.
12599 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
12601 (thunderx_load): Split load_8 to load_16.
12602 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
12604 (thunderx2t99_storepair_basic): Split store_8 to store_16.
12605 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
12606 (xgene1_store_pair): Split store_8 to store_16.
12607 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
12608 (falkor_st_0_st_sd): Split store_8 to store_16.
12610 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
12612 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
12613 and store1/2/3/4 to store_4/8/12/16.
12614 * config/aarch64/aarch64.md: Update for rename.
12615 * config/arm/arm.md: Likewise.: Likewise.
12616 * config/arm/arm.c: Likewise.
12617 * config/arm/thumb1.md: Likewise.
12618 * config/arm/thumb2.md: Likewise.
12619 * config/arm/vfp.md: Likewise.
12620 * config/arm/arm-generic.md: Likewise.
12621 * config/arm/arm1020e.md: Likewise.
12622 * config/arm/arm1026ejs.md: Likewise.
12623 * config/arm/arm1136jfs.md: Likewise.
12624 * config/arm/arm926ejs.md: Likewise.
12625 * config/arm/cortex-a15.md: Likewise.
12626 * config/arm/cortex-a17.md: Likewise.
12627 * config/arm/cortex-a5.md: Likewise.
12628 * config/arm/cortex-a53.md: Likewise.
12629 * config/arm/cortex-a57.md: Likewise.
12630 * config/arm/cortex-a7.md: Likewise.
12631 * config/arm/cortex-a8.md: Likewise.
12632 * config/arm/cortex-a9.md: Likewise.
12633 * config/arm/cortex-m4.md: Likewise.
12634 * config/arm/cortex-m7.md: Likewise.
12635 * config/arm/cortex-r4.md: Likewise.
12636 * config/arm/exynos-m1.md: Likewise.
12637 * config/arm/fa526.md: Likewise.
12638 * config/arm/fa606te.md: Likewise.
12639 * config/arm/fa626te.md: Likewise.
12640 * config/arm/fa726te.md: Likewise.
12641 * config/arm/fmp626.md: Likewise.
12642 * config/arm/iwmmxt.md: Likewise.
12643 * config/arm/ldmstm.md: Likewise.
12644 * config/arm/marvell-pj4.md: Likewise.
12645 * config/arm/xgene1.md: Likewise.
12646 * config/aarch64/thunderx.md: Likewise.
12647 * config/aarch64/thunderx2t99.md: Likewise.
12648 * config/aarch64/falkor.md: Likewise.
12650 2017-09-12 Martin Liska <mliska@suse.cz>
12652 * attribs.c (private_lookup_attribute): New function.
12653 * attribs.h (private_lookup_attribute): Declared here.
12654 (lookup_attribute): Called from this place.
12656 2017-09-12 Richard Biener <rguenther@suse.de>
12658 PR tree-optimization/82157
12659 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
12660 stmts with side-effects.
12662 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12663 Alan Hayward <alan.hayward@arm.com>
12664 David Sherwood <david.sherwood@arm.com>
12666 * target.def (hard_regno_nregs): New hook.
12667 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
12668 * targhooks.h (default_hard_regno_nregs): Declare.
12669 * targhooks.c (default_hard_regno_nregs): New function.
12670 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
12671 (TARGET_HARD_REGNO_NREGS): ...this hook.
12672 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
12673 (CLASS_MAX_NREGS): Likewise.
12674 * doc/tm.texi: Regenerate.
12675 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
12676 instead of HARD_REGNO_NREGS.
12677 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
12678 HARD_REGNO_NREGS in the comment.
12679 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
12680 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
12681 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
12682 Return an unsigned int.
12683 (TARGET_HARD_REGNO_NREGS): Redefine.
12684 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
12685 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
12686 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
12687 (arc_hard_regno_nregs): New function.
12688 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
12689 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
12690 (arm_hard_regno_nregs): New function.
12691 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
12692 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
12693 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
12694 (TARGET_HARD_REGNO_NREGS): Redefine.
12695 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
12696 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
12697 (HARD_REGNO_NREGS): Delete.
12698 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
12699 (cr16_hard_regno_nregs): New function.
12700 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
12701 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
12702 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
12703 (cris_hard_regno_nregs): New function.
12704 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
12705 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
12706 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
12707 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
12708 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
12709 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
12710 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
12711 (frv_hard_regno_nregs): Make static. Take and return an
12713 (frv_class_max_nregs): Remove outdated copy of documentation.
12714 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
12715 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
12716 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
12717 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
12718 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
12719 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
12720 (TARGET_HARD_REGNO_NREGS): Redefine.
12721 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
12722 (CLASS_MAX_NREGS): Update comment.
12723 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
12724 (ia64_hard_regno_nregs): New function.
12725 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
12726 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
12727 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
12728 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
12729 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
12731 (m32c_hard_regno_nregs): Likewise. Make static.
12732 (TARGET_HARD_REGNO_NREGS): Redefine.
12733 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
12734 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
12735 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
12736 (m68k_hard_regno_nregs): New function.
12737 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
12738 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
12739 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
12740 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
12741 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
12742 Take and return an unsigned int.
12743 (TARGET_HARD_REGNO_NREGS): Redefine.
12744 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
12745 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
12746 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
12747 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
12748 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
12749 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
12750 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
12751 (msp430_hard_regno_nregs): Make static. Take and return an
12753 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
12754 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
12755 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
12756 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
12757 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
12758 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
12759 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
12760 (TARGET_HARD_REGNO_NREGS): Redefine.
12761 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
12762 (PA_HARD_REGNO_NREGS): ...this.
12763 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
12764 (PA_HARD_REGNO_NREGS): ...this.
12765 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
12766 (pa_hard_regno_nregs): New function.
12767 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
12768 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
12769 (pdp11_hard_regno_nregs): New function.
12770 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
12771 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
12772 (rs6000_hard_regno_nregs_hook): New function.
12773 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
12774 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
12775 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
12776 Take and return an unsigned int. Move earlier in file.
12777 (TARGET_HARD_REGNO_NREGS): Redefine.
12778 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
12779 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
12780 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
12781 (rl78_hard_regno_nregs): Make static. Take and return an
12783 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
12784 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
12785 (rs6000_hard_regno_nregs_hook): New function.
12786 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
12787 * config/rx/rx.c (rx_hard_regno_nregs): New function.
12788 (TARGET_HARD_REGNO_NREGS): Redefine.
12789 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
12790 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
12791 instead of HARD_REGNO_NREGS.
12792 (s390_hard_regno_nregs): New function.
12793 (s390_hard_regno_mode_ok): Add comment from s390.h.
12794 (TARGET_HARD_REGNO_NREGS): Redefine.
12795 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
12796 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
12797 (sh_hard_regno_nregs): New function.
12798 (sh_pass_in_reg_p): Use it.
12799 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
12800 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
12801 (sparc_hard_regno_nregs): New function.
12802 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
12803 * config/spu/spu.c (spu_hard_regno_nregs): New function.
12804 (spu_function_arg_advance): Use it, supplying a valid register number.
12805 (TARGET_HARD_REGNO_NREGS): Redefine.
12806 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
12807 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
12808 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
12809 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
12810 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
12811 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
12812 (CLASS_MAX_NREGS): Remove copy of old documentation.
12813 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
12814 (visium_hard_regno_nregs): New function.
12815 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
12816 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
12817 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
12818 xtensa_hard_regno_nregs): New function.
12819 * system.h (HARD_REGNO_NREGS): Poison.
12821 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12823 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
12824 hard_regno_nregs instead of HARD_REGNO_NREGS.
12825 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
12826 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
12827 (c6x_expand_epilogue): Likewise.
12828 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
12829 (frv_read_iacc_argument): Likewise.
12830 * config/sh/sh.c: Include regs.h.
12831 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12832 (regs_used): Likewise.
12833 (output_stack_adjust): Likewise.
12834 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
12835 * expmed.c: Include regs.h.
12836 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12837 * ree.c: Include regs.h.
12838 (combine_reaching_defs): Use hard_regno_nregs instead of
12840 (add_removable_extension): Likewise.
12842 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12844 * regs.h (hard_regno_nregs): Turn into a function.
12845 (end_hard_regno): Update accordingly.
12846 * caller-save.c (setup_save_areas): Likewise.
12847 (save_call_clobbered_regs): Likewise.
12848 (replace_reg_with_saved_mem): Likewise.
12849 (insert_restore): Likewise.
12850 (insert_save): Likewise.
12851 * combine.c (can_change_dest_mode): Likewise.
12852 (move_deaths): Likewise.
12853 (distribute_notes): Likewise.
12854 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
12855 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
12856 (rs6000_split_multireg_move): Likewise.
12857 (rs6000_register_move_cost): Likewise.
12858 (rs6000_memory_move_cost): Likewise.
12859 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
12860 (rs6000_split_multireg_move): Likewise.
12861 (rs6000_register_move_cost): Likewise.
12862 (rs6000_memory_move_cost): Likewise.
12863 * cselib.c (cselib_reset_table): Likewise.
12864 (cselib_lookup_1): Likewise.
12865 * emit-rtl.c (set_mode_and_regno): Likewise.
12866 * function.c (aggregate_value_p): Likewise.
12867 * ira-color.c (setup_profitable_hard_regs): Likewise.
12868 (check_hard_reg_p): Likewise.
12869 (calculate_saved_nregs): Likewise.
12870 (assign_hard_reg): Likewise.
12871 (improve_allocation): Likewise.
12872 (calculate_spill_cost): Likewise.
12873 * ira-emit.c (modify_move_list): Likewise.
12874 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
12875 (ira_hard_reg_in_set_p): Likewise.
12876 * ira.c (setup_reg_mode_hard_regset): Likewise.
12877 (clarify_prohibited_class_mode_regs): Likewise.
12878 (check_allocation): Likewise.
12879 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12880 (lra_setup_reg_renumber): Likewise.
12881 (setup_try_hard_regno_pseudos): Likewise.
12882 (spill_for): Likewise.
12883 (assign_hard_regno): Likewise.
12884 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
12885 * lra-constraints.c (in_class_p): Likewise.
12886 (lra_constraint_offset): Likewise.
12887 (simplify_operand_subreg): Likewise.
12888 (lra_constraints): Likewise.
12889 (split_reg): Likewise.
12890 (split_if_necessary): Likewise.
12891 (invariant_p): Likewise.
12892 (inherit_in_ebb): Likewise.
12893 * lra-lives.c (process_bb_lives): Likewise.
12894 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
12895 (get_hard_regs): Likewise.
12896 (do_remat): Likewise.
12897 * lra-spills.c (assign_spill_hard_regs): Likewise.
12898 * mode-switching.c (create_pre_exit): Likewise.
12899 * postreload.c (reload_combine_recognize_pattern): Likewise.
12900 * recog.c (peep2_find_free_register): Likewise.
12901 * regcprop.c (kill_value_regno): Likewise.
12902 (set_value_regno): Likewise.
12903 (copy_value): Likewise.
12904 (maybe_mode_change): Likewise.
12905 (find_oldest_value_reg): Likewise.
12906 (copyprop_hardreg_forward_1): Likewise.
12907 * regrename.c (check_new_reg_p): Likewise.
12908 (regrename_do_replace): Likewise.
12909 * reload.c (push_reload): Likewise.
12910 (combine_reloads): Likewise.
12911 (find_dummy_reload): Likewise.
12912 (operands_match_p): Likewise.
12913 (find_reloads): Likewise.
12914 (find_equiv_reg): Likewise.
12915 (reload_adjust_reg_for_mode): Likewise.
12916 * reload1.c (count_pseudo): Likewise.
12917 (count_spilled_pseudo): Likewise.
12918 (find_reg): Likewise.
12919 (clear_reload_reg_in_use): Likewise.
12920 (free_for_value_p): Likewise.
12921 (allocate_reload_reg): Likewise.
12922 (choose_reload_regs): Likewise.
12923 (reload_adjust_reg_for_temp): Likewise.
12924 (emit_reload_insns): Likewise.
12925 (delete_output_reload): Likewise.
12926 * rtlanal.c (subreg_get_info): Likewise.
12927 * sched-deps.c (sched_analyze_reg): Likewise.
12928 * sel-sched.c (init_regs_for_mode): Likewise.
12929 (mark_unavailable_hard_regs): Likewise.
12930 (choose_best_reg_1): Likewise.
12931 (verify_target_availability): Likewise.
12932 * valtrack.c (dead_debug_insert_temp): Likewise.
12933 * var-tracking.c (track_loc_p): Likewise.
12934 (emit_note_insn_var_location): Likewise.
12935 * varasm.c (make_decl_rtl): Likewise.
12936 * reginfo.c (choose_hard_reg_mode): Likewise.
12937 (init_reg_modes_target): Refer directly to
12938 this_target_regs->x_hard_regno_nregs.
12940 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12942 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
12943 instead of hard_regno_nregs.
12945 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12947 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
12948 end_hard_regno instead of hard_regno_nregs.
12949 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
12950 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
12951 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
12952 * ira-color.c (improve_allocation): Likewise.
12953 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12954 * lra-lives.c (mark_regno_live): Likewise.
12955 (mark_regno_dead): Likewise.
12956 * lra-remat.c (operand_to_remat): Likewise.
12957 * lra.c (collect_non_operand_hard_regs): Likewise.
12958 * postreload.c (reload_combine_note_store): Likewise.
12959 (move2add_valid_value_p): Likewise.
12960 * reload.c (regno_clobbered_p): Likewise.
12962 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12964 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
12966 * config/v850/v850.c (v850_reorg): Likewise.
12967 * reload.c (refers_to_regno_for_reload_p): Likewise.
12968 (find_equiv_reg): Likewise.
12969 * reload1.c (reload_reg_reaches_end_p): Likewise.
12971 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12973 * caller-save.c (add_used_regs): Use REG_NREGS instead of
12975 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
12976 * config/arm/arm.c (output_move_neon): Likewise.
12977 (arm_attr_length_move_neon): Likewise.
12978 (neon_split_vcombine): Likewise.
12979 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
12980 (c6x_mark_reg_written): Likewise.
12981 (c6x_dwarf_register_span): Likewise.
12982 * config/i386/i386.c (ix86_save_reg): Likewise.
12983 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
12984 (rws_access_reg): Likewise.
12985 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
12986 * mode-switching.c (create_pre_exit): Likewise.
12987 * ree.c (combine_reaching_defs): Likewise.
12988 (add_removable_extension): Likewise.
12989 * regcprop.c (find_oldest_value_reg): Likewise.
12990 (copyprop_hardreg_forward_1): Likewise.
12991 * reload.c (reload_inner_reg_of_subreg): Likewise.
12992 (push_reload): Likewise.
12993 (combine_reloads): Likewise.
12994 (find_dummy_reload): Likewise.
12995 (reload_adjust_reg_for_mode): Likewise.
12996 * reload1.c (find_reload_regs): Likewise.
12997 (forget_old_reloads_1): Likewise.
12998 (reload_reg_free_for_value_p): Likewise.
12999 (reload_adjust_reg_for_temp): Likewise.
13000 (emit_reload_insns): Likewise.
13001 (delete_output_reload): Likewise.
13002 * sel-sched.c (choose_best_reg_1): Likewise.
13003 (choose_best_pseudo_reg): Likewise.
13005 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13006 Alan Hayward <alan.hayward@arm.com>
13007 David Sherwood <david.sherwood@arm.com>
13009 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
13010 * target.def (slow_unaligned_access): New hook.
13011 * targhooks.h (default_slow_unaligned_access): Declare.
13012 * targhooks.c (default_slow_unaligned_access): New function.
13013 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
13014 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
13015 * doc/tm.texi: Regenerate.
13016 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
13017 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
13018 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
13020 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
13021 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
13023 (rs6000_slow_unaligned_access): New function.
13024 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
13025 (expand_block_compare): Likewise.
13026 (expand_strn_compare): Likewise.
13027 (rs6000_rtx_costs): Likewise.
13028 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
13029 (riscv_slow_unaligned_access): Likewise.
13030 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
13031 (riscv_slow_unaligned_access_p): ...this and make static.
13032 (riscv_option_override): Update accordingly.
13033 (riscv_slow_unaligned_access): New function.
13034 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
13035 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
13036 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
13037 (rs6000_slow_unaligned_access): New function.
13038 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
13039 (rs6000_rtx_costs): Likewise.
13040 * config/rs6000/rs6000-string.c (expand_block_compare)
13041 (expand_strn_compare): Use targetm.slow_unaligned_access instead
13042 of SLOW_UNALIGNED_ACCESS.
13043 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
13044 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
13045 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
13046 of SLOW_UNALIGNED_ACCESS.
13047 * expmed.c (simple_mem_bitfield_p): Likewise.
13048 * expr.c (alignment_for_piecewise_move): Likewise.
13049 (emit_group_load_1): Likewise.
13050 (emit_group_store): Likewise.
13051 (copy_blkmode_from_reg): Likewise.
13052 (emit_push_insn): Likewise.
13053 (expand_assignment): Likewise.
13054 (store_field): Likewise.
13055 (expand_expr_real_1): Likewise.
13056 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
13057 * lra-constraints.c (simplify_operand_subreg): Likewise.
13058 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13059 * gimple-ssa-store-merging.c: Likewise in block comment at start
13061 * tree-ssa-strlen.c: Include target.h.
13062 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
13063 of SLOW_UNALIGNED_ACCESS.
13064 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
13066 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13068 PR rtl-optimization/82185
13069 * expmed.c (emit_store_flag_int): Only test tem if it has been
13072 2017-09-12 Richard Biener <rguenther@suse.de>
13074 PR middle-end/82149
13075 * match.pd ((FTYPE) N CMP CST): Fix typo.
13077 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
13079 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
13081 (mips_near_type_p): Add 'short_call' attribute as a synonym
13083 * doc/extend.texi (short_call): Document new function attribute.
13085 2017-09-12 Jakub Jelinek <jakub@redhat.com>
13088 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
13089 assertion check that in the condition.
13090 (get_atomic_generic_size): Likewise. Before testing if parameter
13091 has pointer type, if it has array type, call for C++
13092 default_conversion to perform array-to-pointer conversion.
13094 2017-09-12 Richard Biener <rguenther@suse.de>
13096 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
13097 for operations we cannot scalarize.
13099 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
13101 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
13102 vectors heap vectors. Clean up comments.
13103 Make visited_bbs a reference.
13104 (profitable_jump_thread_path): Make GC
13105 vectors heap vectors. Clean up comments.
13107 (convert_and_register_jump_thread_path): Make GC vectors heap
13109 (check_subpath_and_update_thread_path): Same. Clean up comments.
13110 Make visited_bbs a reference.
13111 (handle_phi): Abstract common code to to
13112 register_jump_thread_path_if_profitable.
13113 Rename VAR_BB to DEF_BB.
13115 Make GC vectors heap vectors.
13116 Make visited_bbs a reference.
13117 (handle_assignment): Same.
13118 (register_jump_thread_path_if_profitable): New.
13119 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
13121 Make GC vectors heap vectors. Clean up comments.
13122 Make visited_bbs a reference.
13123 (find_jump_threads_backwards): Make visited_bbs live in the stack.
13124 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
13127 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
13130 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
13131 words of E_DImode object are reachable by xtensa_uimm8x4 access.
13133 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
13135 Revert r251800 and r251799.
13137 2017-09-11 Martin Jambor <mjambor@suse.cz>
13140 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
13141 arguments in advance.
13142 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
13143 use it to find predecessor edges.
13144 (naive_outof_ssa): Collect vector of predecessors.
13146 2017-09-08 Jason Merrill <jason@redhat.com>
13148 PR c++/70029 - ICE with ref-qualifier and -flto
13149 * langhooks.h (struct lang_hooks_for_types): Add
13150 copy_lang_qualifiers.
13151 * attribs.c (build_type_attribute_qual_variant): Use it.
13152 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
13154 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
13155 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
13157 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
13160 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
13161 (*mulsi3_sp64): New instruction.
13162 (mulsi3): New expander.
13164 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
13166 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
13168 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13170 * sancov.c: Include memmodel.h.
13172 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
13175 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
13178 2017-09-07 Carl Love <cel@us.ibm.com>
13180 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
13181 the sldi instruction.
13183 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
13185 * sancov.c: Include tm_p.h.
13187 2017-09-07 Jakub Jelinek <jakub@redhat.com>
13190 * output.h (switch_to_other_text_partition): New declaration.
13191 * varasm.c (switch_to_other_text_partition): New function.
13192 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
13193 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
13194 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
13195 to the other text partition before emitting LCL label and switch back
13196 after emitting the word after it.
13198 2017-09-07 Richard Biener <rguenther@suse.de>
13200 * passes.def (pass_split_crit_edges): Remove instance before PRE.
13201 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
13202 critical edges here, after loop init.
13203 (pass_data_pre): Remove PROP_no_crit_edges flags.
13204 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
13205 for valueization of call args to avoid leaking VN_TOP.
13206 (visit_use): Assert we do not visit default defs.
13207 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
13208 Use error_mark_node to more easily detect leaking VN_TOP.
13209 All default-defs are varying, not VN_TOP. Mark them visited.
13210 (run_scc_vn): Make code match comment.
13212 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13214 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
13215 OPTION_MASK_FLOAT128_KEYWORD.
13216 (POWERPC_MASKS): Likewise.
13217 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
13218 support for the -mfloat128-type option, and make -mfloat128
13219 default on PowerPC Linux systems. Define or undefine
13220 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
13221 Define __float128 to be __ieee128 if IEEE 128-bit support is
13222 enabled, or undefine it.
13223 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
13224 Delete defining __FLOAT128_TYPE__.
13225 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
13226 -mfloat128-type option and make -mfloat128 default on PowerPC
13228 (TARGET_FLOAT128_TYPE): Likewise.
13229 (-mfloat128-type): Likewise.
13230 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13231 Delete the -mfloat128-type option and make -mfloat128 default on
13232 PowerPC Linux systems. Always use __ieee128 to be the keyword for
13233 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
13234 128-bit floating point is enabled. Change tests from using
13235 -mfloat128-type to -mfloat128.
13236 (rs6000_mangle_type): Use the correct mangling for the __float128
13237 type even if normal long double is restricted to 64-bits.
13238 (floatn_mode): Enable the _Float128 type by default on VSX Linux
13240 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
13241 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
13242 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
13244 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
13245 documentation for -mfloat128.
13247 2017-09-06 Olivier Hainque <hainque@adacore.com>
13249 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
13251 2017-09-06 Wish Wu <wishwu007@gmail.com>
13252 Jakub Jelinek <jakub@redhat.com>
13254 * asan.c (initialize_sanitizer_builtins): Add
13255 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
13256 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
13257 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
13258 BT_FN_VOID_UINT64_PTR variables.
13259 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
13260 (BT_FN_VOID_UINT16_UINT16): Likewise.
13261 (BT_FN_VOID_UINT32_UINT32): Likewise.
13262 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
13263 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
13264 (BT_FN_VOID_UINT64_PTR): Likewise.
13265 * common.opt (flag_sanitize_coverage): New variable.
13266 (fsanitize-coverage=trace-pc): Remove.
13267 (fsanitize-coverage=): Add.
13268 * flag-types.h (enum sanitize_coverage_code): New enum.
13269 * fold-const.c (fold_range_test): Disable non-short-circuit
13270 optimization if flag_sanitize_coverage.
13271 (fold_truth_andor): Likewise.
13272 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
13273 * opts.c (COVERAGE_SANITIZER_OPT): Define.
13274 (coverage_sanitizer_opts): New array.
13275 (get_closest_sanitizer_option): Add OPTS argument, handle also
13276 OPT_fsanitize_coverage_.
13277 (parse_sanitizer_options): Adjusted to also handle
13278 OPT_fsanitize_coverage_.
13279 (common_handle_option): Add OPT_fsanitize_coverage_.
13280 * sancov.c (instrument_comparison, instrument_switch): New function.
13281 (sancov_pass): Add trace-cmp support.
13282 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
13283 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
13284 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
13285 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
13286 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
13287 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
13288 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
13289 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
13290 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
13291 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
13293 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13295 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13296 error. Only quit immediately if parsing is complete.
13297 (BEGIN): Initialize fatal_err and parse_done.
13298 (begin fpu, end fpu): Check number of arguments.
13299 (begin arch, end arch): Likewise.
13300 (begin cpu, end cpu): Likewise.
13301 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13302 (optalias): Likewise.
13304 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13306 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13307 * config/arm/arm-isa.h: Delete. Move definitions to ...
13308 * arm-cpus.in: ... here. Use new feature and fgroup values.
13309 * config/arm/arm.c (arm_option_override): Use lower case for feature
13311 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13312 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13313 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13315 (print_isa_bits_for): New function.
13316 (gen_isa): New function.
13317 (gen_comm_data): Use print_isa_bits_for.
13318 (define feature): New keyword.
13319 (define fgroup): New keyword.
13320 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
13321 (arm-isa.h): Add rule to generate file.
13322 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13323 case for feature bit names.
13325 2017-09-06 Richard Biener <rguenther@suse.de>
13327 * tree-ssa-pre.c (NECESSARY): Remove.
13328 (create_expression_by_pieces): Do not touch pass-local flags.
13329 (insert_into_preds_of_block): Likewise.
13330 (do_pre_regular_insertion): Likewise.
13331 (eliminate_insert): Likewise.
13332 (eliminate_dom_walker::before_dom_children): Likewise.
13333 (fini_eliminate): Do not look at inserted_exprs.
13334 (mark_operand_necessary): Remove.
13335 (remove_dead_inserted_code): Replace with simple work-list
13336 algorithm based on inserted_exprs and SSA uses.
13337 (pass_pre::execute): Re-order fini_eliminate and
13338 remove_dead_inserted_code.
13340 2017-09-06 Olivier Hainque <hainque@adacore.com>
13342 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
13343 for VxWorks 7. Adjust surrounding comments.
13345 2017-09-06 Richard Biener <rguenther@suse.de>
13347 * gimple-ssa-strength-reduction.c
13348 (find_candidates_dom_walker::before_dom_children): Also allow
13351 2017-09-06 Richard Biener <rguenther@suse.de>
13353 PR tree-optimization/82108
13354 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
13355 for gap in the non-permutation SLP case.
13357 2017-09-06 Martin Jambor <mjambor@suse.cz>
13359 PR tree-optimization/82078
13360 * tree-sra.c (sort_and_splice_var_accesses): Move call to
13361 add_access_to_work_queue...
13362 (build_accesses_from_assign): ...here.
13363 (propagate_all_subaccesses): Make sure racc is the group
13364 representative, if there is one.
13366 2017-09-06 Jakub Jelinek <jakub@redhat.com>
13368 PR middle-end/82095
13369 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
13372 2017-09-06 Richard Biener <rguenther@suse.de>
13374 * gimple-ssa-strength-reduction.c
13375 (find_candidates_dom_walker::before_doom_children): Use a
13376 type and not a mode check.
13378 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
13381 * config/arm/predicates.md (arm_general_adddi_operand): Create new
13383 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
13385 2017-09-05 Jeff Law <law@redhat.com>
13387 PR tree-optimization/64910
13388 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
13389 cases where we have 3 or more operands.
13391 2017-09-05 Jakub Jelinek <jakub@redhat.com>
13393 PR middle-end/81768
13394 * omp-low.c (lower_omp_for): Recompute tree invariant if
13395 gimple_omp_for_initial/final is ADDR_EXPR.
13397 PR middle-end/81768
13398 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
13399 into gimple val before gimplification fo the COND_EXPR.
13401 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
13403 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
13404 REGION_COPY argument.
13405 (thread_through_all_blocks): Remove unused argument to
13406 duplicate_thread_path.
13408 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13409 Alan Hayward <alan.hayward@arm.com>
13410 David Sherwood <david.sherwood@arm.com>
13412 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
13413 Take a scalar_mode rather than a machine_mode.
13414 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13415 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
13416 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13417 (aarch64_gen_adjusted_ldpstp): Likewise.
13418 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
13420 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13421 Alan Hayward <alan.hayward@arm.com>
13422 David Sherwood <david.sherwood@arm.com>
13424 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
13425 Take a scalar_int_mode instead of a machine_mode.
13426 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13427 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13428 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13429 (aarch64_simd_attr_length_rglist): Delete.
13430 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
13431 a scalar_int_mode instead of a machine_mode.
13432 (aarch64_add_offset): Likewise.
13433 (aarch64_internal_mov_immediate): Likewise
13434 (aarch64_add_constant_internal): Likewise.
13435 (aarch64_add_constant): Likewise.
13436 (aarch64_movw_imm): Likewise.
13437 (aarch64_rtx_arith_op_extract_p): Likewise.
13438 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13439 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13440 Remove assert that the mode isn't a vector.
13441 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13442 (aarch64_expand_mov_immediate): Update calls after above changes.
13443 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
13444 (aarch64_and_bitmask_imm): Check for scalar integer modes.
13445 (aarch64_move_imm): Likewise.
13446 (aarch64_can_const_movi_rtx_p): Likewise.
13447 (aarch64_strip_extend): Likewise.
13448 (aarch64_extr_rtx_p): Likewise.
13449 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
13450 a CONST_INT when the mode parameter is VOIDmode.
13451 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
13453 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13455 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
13456 * stor-layout.c (bitwise_mode_for_mode): Likewise.
13457 (bitwise_type_for_mode): Update accordingly.
13459 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13461 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
13462 * stor-layout.c (mode_for_size_tree): Likewise.
13463 (mode_for_array): Update accordingly.
13464 (layout_decl): Likewise.
13465 (compute_record_mode): Likewise. Only set the mode once.
13467 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13469 * target.def (get_mask_mode): Change return type to opt_mode.
13471 * doc/tm.texi: Regenerate.
13472 * targhooks.h (default_get_mask_mode): Return an opt_mode.
13473 * targhooks.c (default_get_mask_mode): Likewise.
13474 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
13475 * optabs-query.c (can_vec_mask_load_store_p): Update use of
13476 targetm.get_mask_mode.
13477 * tree.c (build_truth_vector_type): Likewise.
13479 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13481 * machmode.h (mode_for_vector): Return an opt_mode.
13482 * stor-layout.c (mode_for_vector): Likewise.
13483 (mode_for_int_vector): Update accordingly.
13484 (layout_type): Likewise.
13485 * config/i386/i386.c (emit_memmov): Likewise.
13486 (ix86_expand_set_or_movmem): Likewise.
13487 (ix86_expand_vector_init): Likewise.
13488 (ix86_get_mask_mode): Likewise.
13489 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
13491 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
13492 * expmed.c (extract_bit_field_1): Likewise.
13493 * expr.c (expand_expr_real_2): Likewise.
13494 * optabs-query.c (can_vec_perm_p): Likewise.
13495 (can_vec_mask_load_store_p): Likewise.
13496 * optabs.c (expand_vec_perm): Likewise.
13497 * targhooks.c (default_get_mask_mode): Likewise.
13498 * tree-vect-stmts.c (vectorizable_store): Likewise.
13499 (vectorizable_load): Likewise.
13500 (get_vectype_for_scalar_type_and_size): Likewise.
13502 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13504 * machmode.h (mode_for_int_vector): New function.
13505 * stor-layout.c (mode_for_int_vector): Likewise.
13506 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
13507 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
13508 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
13509 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
13510 (s390_expand_vcond): Likewise.
13512 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13514 * machmode.h (opt_machine_mode): New type.
13515 (opt_mode<T>): Allow construction from anything that can be
13517 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
13518 (mode_for_size): Return an opt_machine_mode.
13519 * stor-layout.c (mode_for_size): Likewise.
13520 (mode_for_size_tree): Update call accordingly.
13521 (bitwise_mode_for_mode): Likewise.
13522 (make_fract_type): Likewise.
13523 (make_accum_type): Likewise.
13524 * caller-save.c (replace_reg_with_saved_mem): Update call
13526 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13527 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13528 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13529 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13530 * expmed.c (extract_bit_field_1): Likewise.
13531 * reload.c (get_secondary_mem): Likewise.
13532 * varasm.c (assemble_integer): Likewise.
13533 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
13536 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13538 * machmode.h (decimal_float_mode_for_size): New function.
13539 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
13540 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
13541 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
13542 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
13543 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
13544 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
13546 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13548 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
13549 (get_builtin_sync_mode): Likewise.
13550 (expand_ifn_atomic_compare_exchange): Likewise.
13551 (expand_builtin_atomic_clear): Likewise.
13552 (expand_builtin_atomic_test_and_set): Likewise.
13553 (fold_builtin_atomic_always_lock_free): Likewise.
13554 * calls.c (compute_argument_addresses): Likewise.
13555 (emit_library_call_value_1): Likewise.
13556 (store_one_arg): Likewise.
13557 * combine.c (combine_instructions): Likewise.
13558 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
13559 * config/arm/arm.c (arm_function_value): Likewise.
13560 (aapcs_allocate_return_reg): Likewise.
13561 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
13562 * config/i386/i386.c (construct_container): Likewise.
13563 (ix86_gimplify_va_arg): Likewise.
13564 (ix86_expand_sse_cmp): Likewise.
13565 (emit_memmov): Likewise.
13566 (emit_memset): Likewise.
13567 (expand_small_movmem_or_setmem): Likewise.
13568 (ix86_expand_pextr): Likewise.
13569 (ix86_expand_pinsr): Likewise.
13570 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
13571 * config/microblaze/microblaze.c (microblaze_block_move_straight):
13573 * config/mips/mips.c (mips_function_value_1) Likewise.
13574 (mips_block_move_straight): Likewise.
13575 (mips_expand_ins_as_unaligned_store): Likewise.
13576 * config/powerpcspe/powerpcspe.c
13577 (rs6000_darwin64_record_arg_advance_flush): Likewise.
13578 (rs6000_darwin64_record_arg_flush): Likewise.
13579 * config/rs6000/rs6000.c
13580 (rs6000_darwin64_record_arg_advance_flush): Likewise.
13581 (rs6000_darwin64_record_arg_flush): Likewise.
13582 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
13583 (sparc_function_value_1): Likewise.
13584 * config/spu/spu.c (adjust_operand): Likewise.
13585 (spu_emit_branch_or_set): Likewise.
13586 (arith_immediate_p): Likewise.
13587 * emit-rtl.c (gen_lowpart_common): Likewise.
13588 * expr.c (expand_expr_real_1): Likewise.
13589 * function.c (assign_parm_setup_block): Likewise.
13590 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
13591 * reload1.c (alter_reg): Likewise.
13592 * stor-layout.c (mode_for_vector): Likewise.
13593 (layout_type): Likewise.
13595 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13597 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
13598 (spu_convert_move): Likewise.
13599 * lower-subreg.c (resolve_simple_move): Likewise.
13601 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13604 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
13605 define_insn to a define_expand.
13606 (altivec_vsum2sws_direct): New define_insn.
13607 (altivec_vsumsws): Convert from a define_insn to a define_expand.
13609 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
13611 * config/arm/arm.c (arm_option_params_internal): Improve setting of
13614 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
13619 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
13620 realign stack if stack alignment needed is less than incoming
13623 2017-09-05 Marek Polacek <polacek@redhat.com>
13626 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
13629 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
13631 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
13633 2017-09-05 Richard Biener <rguenther@suse.de>
13635 PR tree-optimization/82084
13636 * fold-const.c (can_native_encode_string_p): Handle wide characters.
13638 2017-09-05 Richard Biener <rguenther@suse.de>
13640 PR tree-optimization/82102
13641 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
13643 2017-09-05 Martin Liska <mliska@suse.cz>
13645 PR tree-optimization/82032
13646 * tree-cfg.c (generate_range_test): New function.
13647 * tree-cfg.h (generate_range_test): Declared here.
13648 * tree-cfgcleanup.c (convert_single_case_switch): New function.
13649 (cleanup_control_expr_graph): Use it.
13650 * tree-switch-conversion.c (try_switch_expansion): Remove
13652 (emit_case_nodes): Use generate_range_test.
13654 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
13657 * config/i386/i386.md (*<btsc><mode>_mask): Add
13658 TARGET_USE_BT to insn constraint.
13659 (*btr<mode>_mask): Ditto.
13661 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
13663 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
13664 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
13666 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
13669 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
13670 TARGET_NEON and TARGET_IWMMXT.
13671 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
13672 TARGET_NEON and TARGET_IWMMXT.
13673 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
13675 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
13677 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
13678 (ix86_rewrite_tls_address): Ditto.
13679 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
13680 (ix86_rewrite_tls_address_1): Ditto.
13681 (ix86_rewrite_tls_address): Ditto.
13682 * config/i386/predicates.md (tls_address_pattern): New predicate.
13683 * config/i386/i386.md (TLS address splitter): New splitter.
13685 2017-09-04 Richard Biener <rguenther@suse.de>
13687 PR tree-optimization/82084
13688 * fold-const.h (can_native_encode_string_p): Declare.
13689 * fold-const.c (can_native_encode_string_p): Factor out from ...
13690 (native_encode_string): ... here.
13691 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
13692 vectorizing stores from constants we later cannot handle.
13694 2017-09-04 Marek Polacek <polacek@redhat.com>
13697 * doc/invoke.texi: Update -Wtautological-compare documentation.
13699 2017-09-04 Jeff Law <law@redhat.com>
13701 PR tree-optimization/64910
13702 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
13703 swap the first and last operand if the last is a constant.
13705 2017-09-04 Marek Polacek <polacek@redhat.com>
13708 * convert.c (do_narrow): When sanitizing signed integer overflows,
13709 bail out for signed types.
13710 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
13712 2017-09-04 Richard Biener <rguenther@suse.de>
13714 PR tree-optimization/82060
13715 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13716 Move devirtualization after stmt folding and before EH/AB/noreturn
13717 cleanup to get the stmt refs canonicalized. Use a bool instead
13718 of gimple_modified_p since that doesn't work for NOPs. Schedule
13719 NOPs generated by folding for removal.
13721 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13722 Alan Hayward <alan.hayward@arm.com>
13723 David Sherwood <david.sherwood@arm.com>
13725 * coretypes.h (pad_direction): New enum.
13726 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
13727 (FUNCTION_ARG_PADDING): Likewise.
13728 * target.def (function_arg_padding): New hook.
13729 * targhooks.h (default_function_arg_padding): Declare.
13730 * targhooks.c (default_function_arg_padding): New function.
13731 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
13732 (TARGET_FUNCTION_ARG_PADDING): ...this.
13733 * doc/tm.texi: Regenerate.
13734 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
13735 instead of direction.
13736 (compute_argument_addresses): Likewise.
13737 (load_register_parameters): Likewise.
13738 (emit_library_call_value_1): Likewise.
13739 (store_one_arg): Use targetm.calls.function_arg_padding instead
13740 of FUNCTION_ARG_PADDING.
13741 (must_pass_in_stack_var_size_or_pad): Likewise.
13742 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
13743 (emit_group_store): Likewise.
13744 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
13745 instead of FUNCTION_ARG_PADDING.
13746 (emit_push_insn): Likewise, and propagate enum change throughout
13748 * function.h (direction): Delete.
13749 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
13751 * function.c (assign_parm_find_stack_rtl): Likewise.
13752 (assign_parm_setup_block_p): Likewise.
13753 (assign_parm_setup_block): Likewise.
13754 (gimplify_parameters): Likewise.
13755 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
13756 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
13758 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
13759 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13760 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
13761 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
13762 (aarch64_function_arg_padding): ...this new function.
13763 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
13764 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13765 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
13766 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13767 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
13768 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13769 (arm_pad_arg_upward): Replace with...
13770 (arm_function_arg_padding): ...this new function.
13771 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
13773 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
13774 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
13775 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13776 (ia64_hpux_function_arg_padding): Replace with...
13777 (ia64_function_arg_padding): ...this new function. Use pad_direction
13778 instead of direction. Check for TARGET_HPUX.
13779 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
13780 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13781 (iq2000_function_arg_padding): New function.
13782 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
13783 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
13784 (mips_function_arg_padding): ...this new function.
13785 (mips_pad_reg_upward): Update accordingly.
13786 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13787 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
13788 targetm.calls.function_arg_padding.
13789 (FUNCTION_ARG_PADDING): Delete.
13790 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13791 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
13792 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13793 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
13794 (nios2_block_reg_padding): Return pad_direction instead of direction.
13795 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
13796 instead of direction.
13797 (nios2_function_arg_padding): Likewise. Make static.
13798 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13799 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
13800 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
13801 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
13802 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13803 (pa_function_arg_padding): Make static. Return pad_direction instead
13805 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
13806 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13807 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
13808 instead of direction. Use targetm.calls.function_arg_padding.
13809 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
13810 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13811 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
13812 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
13813 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
13815 (function_arg_padding): Rename to...
13816 (rs6000_function_arg_padding): ...this. Make static. Return
13817 pad_direction instead of direction.
13818 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13819 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
13820 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13821 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
13822 instead of direction. Use targetm.calls.function_arg_padding.
13823 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
13824 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13825 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
13826 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
13827 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13828 (function_arg_padding): Rename to...
13829 (rs6000_function_arg_padding): ...this. Make static. Return
13830 pad_direction instead of direction.
13831 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13832 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
13833 * config/s390/s390.c (s390_function_arg_padding): New function.
13834 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13835 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
13836 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
13837 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13838 (function_arg_padding): Rename to...
13839 (sparc_function_arg_padding): ...this. Make static. Return
13840 pad_direction instead of direction.
13841 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
13842 * config/spu/spu.c (spu_function_arg_padding): New function.
13843 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13844 * system.h (FUNCTION_ARG_PADDING): Poison.
13846 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13847 Alan Hayward <alan.hayward@arm.com>
13848 David Sherwood <david.sherwood@arm.com>
13850 * target.def (modes_tieable_p): New hook.
13851 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
13852 (TARGET_MODES_TIEABLE_P): ...this.
13853 * doc/tm.texi.in: Regenerate.
13854 * hooks.h (hook_bool_mode_mode_true): Declare.
13855 * hooks.c (hook_bool_mode_mode_true): New function.
13856 * combine.c (subst): Use targetm.modes_tieable_p instead of
13858 * dse.c (find_shift_sequence): Likewise.
13859 * expmed.c (extract_low_bits): Likewise.
13860 * lower-subreg.c: Include target.h.
13861 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
13863 * rtlanal.c (rtx_cost): Likewise.
13864 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
13865 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
13866 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
13867 (TARGET_MODES_TIEABLE_P): Redefine.
13868 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
13869 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
13870 (TARGET_MODES_TIEABLE_P): Redefine.
13871 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
13872 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
13873 (arc_modes_tieable_p): New function.
13874 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
13875 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
13876 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
13877 (arm_modes_tieable_p): Make static.
13878 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
13879 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
13880 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
13881 (TARGET_MODES_TIEABLE_P): Redefine.
13882 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
13883 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
13884 (TARGET_MODES_TIEABLE_P): Redefine.
13885 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
13886 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
13887 (cr16_modes_tieable_p): New function.
13888 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
13889 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
13890 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
13891 (TRULY_NOOP_TRUNCATION): Update comment.
13892 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
13893 (TRULY_NOOP_TRUNCATION): Update comment.
13894 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
13895 (frv_modes_tieable_p): New function.
13896 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
13897 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
13898 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
13899 (TARGET_MODES_TIEABLE_P): Redefine.
13900 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
13901 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
13902 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
13903 (TARGET_MODES_TIEABLE_P): Redefine.
13904 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
13905 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
13906 (ia64_modes_tieable_p): New function.
13907 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
13908 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
13909 (iq2000_modes_tieable_p): New function.
13910 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
13911 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
13912 (lm32_modes_tieable_p): New function.
13913 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
13914 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
13915 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
13916 (TARGET_MODES_TIEABLE_P): Redefine.
13917 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
13918 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
13919 (m32r_modes_tieable_p): New function.
13920 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
13921 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
13922 (m68k_modes_tieable_p): New function.
13923 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
13924 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
13925 (mcore_modes_tieable_p): New function.
13926 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
13927 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
13929 (TARGET_MODES_TIEABLE_P): Redefine.
13930 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
13931 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
13932 * config/mips/mips.c (mips_modes_tieable_p): Make static.
13933 (TARGET_MODES_TIEABLE_P): Redefine.
13934 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
13935 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
13936 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
13937 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
13938 (mn10300_modes_tieable_p): ...this and make static.
13939 (TARGET_MODES_TIEABLE_P): Redefine.
13940 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
13941 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
13942 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
13943 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
13944 (msp430_modes_tieable_p): Make static.
13945 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
13946 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
13947 (TARGET_MODES_TIEABLE_P): Redefine.
13948 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
13949 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
13950 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
13951 (TARGET_MODES_TIEABLE_P): Redefine.
13952 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
13953 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
13954 * config/pa/pa.c (pa_modes_tieable_p): Make static.
13955 (TARGET_MODES_TIEABLE_P): Redefine.
13956 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
13957 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
13958 (pdp11_modes_tieable_p): New function.
13959 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
13960 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
13961 (rs6000_modes_tieable_p): New function.
13962 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
13963 * config/powerpcspe/powerpcspe.md: Update comment.
13964 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
13965 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
13966 (TARGET_MODES_TIEABLE_P): Redefine.
13967 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
13968 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
13969 (rl78_modes_tieable_p): New function.
13970 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
13971 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
13972 (rs6000_modes_tieable_p): New function.
13973 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
13974 * config/rs6000/rs6000.md: Update comment.
13975 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
13976 * config/rx/rx.c (rx_modes_tieable_p): New function.
13977 (TARGET_MODES_TIEABLE_P): Redefine.
13978 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
13979 * config/s390/s390.c (s390_modes_tieable_p): New function.
13980 (TARGET_MODES_TIEABLE_P): Redefine.
13981 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
13982 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
13983 (sh_modes_tieable_p): New function.
13984 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
13985 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
13986 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
13987 (sparc_modes_tieable_p): Make static.
13988 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
13989 * config/spu/spu.c (spu_modes_tieable_p): New function.
13990 (TARGET_MODES_TIEABLE_P): Redefine.
13991 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
13992 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
13993 (TARGET_MODES_TIEABLE_P): Redefine.
13994 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
13995 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
13996 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
13997 * config/v850/v850.c (v850_modes_tieable_p): New function.
13998 (TARGET_MODES_TIEABLE_P): Redefine.
13999 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
14000 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
14001 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
14002 (visium_modes_tieable_p): New function.
14003 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
14004 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
14005 (xtensa_modes_tieable_p): New function.
14006 * system.h (MODES_TIEABLE_P): Poison.
14008 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14009 Alan Hayward <alan.hayward@arm.com>
14010 David Sherwood <david.sherwood@arm.com>
14012 * target.def (hard_regno_mode_ok): New hook.
14013 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
14014 (TARGET_HARD_REGNO_MODE_OK): ...this.
14015 * doc/tm.texi.in: Regenerate.
14016 * hooks.h (hook_bool_uint_mode_true): Declare.
14017 * hooks.c (hook_bool_uint_mode_true): New function.
14018 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
14019 HARD_REGNO_MODE_OK.
14020 * genpreds.c (write_insn_preds_c): Add an include of target.h.
14021 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
14022 instead of HARD_REGNO_MODE_OK.
14023 * caller-save.c: Include target.h.
14024 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
14025 HARD_REGNO_MODE_OK.
14026 * combine.c (can_combine_p): Likewise.
14027 (combinable_i3pat): Likewise.
14028 (can_change_dest_mode): Likewise.
14029 * expr.c (init_expr_target): Likewise.
14030 (convert_move): Likewise.
14031 (convert_modes): Likewise.
14032 * ira.c (setup_prohibited_class_mode_regs): Likewise.
14033 (setup_prohibited_mode_move_regs): Likewise.
14034 * ira.h (target_ira): Likewise.
14035 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14036 * lra-constraints.c (process_alt_operands): Likewise.
14037 (split_reg): Likewise.
14038 * recog.c (peep2_find_free_register): Likewise.
14039 * ree.c (combine_reaching_defs): Likewise.
14040 * regcprop.c (maybe_mode_change): Likewise.
14041 * reginfo.c (init_reg_sets_1): Likewise.
14042 (choose_hard_reg_mode): Likewise.
14043 (simplifiable_subregs): Likewise.
14044 * regrename.c (check_new_reg_p): Likewise.
14045 * reload.c (find_valid_class): Likewise.
14046 (find_valid_class_1): Likewise.
14047 (reload_inner_reg_of_subreg): Likewise.
14048 (push_reload): Likewise.
14049 (combine_reloads): Likewise.
14050 (find_dummy_reload): Likewise.
14051 (find_reloads): Likewise.
14052 * reload1.c (find_reg): Likewise.
14053 (set_reload_reg): Likewise.
14054 (allocate_reload_reg): Likewise.
14055 (choose_reload_regs): Likewise.
14056 (reload_adjust_reg_for_temp): Likewise.
14057 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
14058 (simplify_subreg_regno): Likewise.
14059 * sel-sched.c (init_regs_for_mode): Likewise.
14060 * varasm.c (make_decl_rtl): Likewise.
14061 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
14062 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
14063 HARD_REGNO_MODE_OK.
14064 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
14065 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
14066 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14067 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
14068 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
14069 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14070 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
14071 (arc_mode_class): Delete.
14072 (HARD_REGNO_MODE_OK): Delete.
14073 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14074 (arc_hard_regno_mode_ok): Rename old array to...
14075 (arc_hard_regno_mode_ok_modes): ...this.
14076 (arc_conditional_register_usage): Update accordingly.
14077 (arc_mode_class): Make static.
14078 (arc_hard_regno_mode_ok): New function.
14079 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
14080 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
14081 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14082 (arm_hard_regno_mode_ok): Make static.
14083 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
14084 HARD_REGNO_MODE_OK.
14085 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
14086 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
14087 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
14089 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14090 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
14091 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
14092 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
14093 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
14094 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14095 * config/bfin/predicates.md (valid_reg_operand): Use
14096 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
14097 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
14098 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
14099 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14100 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
14101 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
14102 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14103 (cr16_hard_regno_mode_ok): Make static and return a bool.
14104 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
14105 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14106 (cris_hard_regno_mode_ok): New function.
14107 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
14108 (epiphany_mode_class): Delete.
14109 (HARD_REGNO_MODE_OK): Delete.
14110 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
14111 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14112 (hard_regno_mode_ok): Rename to...
14113 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
14114 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
14115 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
14116 HARD_REGNO_MODE_OK.
14117 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
14118 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
14119 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14120 (frv_hard_regno_mode_ok): Make static and return a bool.
14121 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
14122 HARD_REGNO_MODE_OK.
14123 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
14124 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
14125 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
14126 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
14128 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14129 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
14130 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
14131 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
14133 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14134 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
14135 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14136 (ia64_hard_regno_mode_ok): New function.
14137 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
14138 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14139 (iq2000_hard_regno_mode_ok): New function.
14140 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
14141 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14142 (lm32_hard_regno_mode_ok): New function.
14143 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
14144 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
14145 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
14146 instead of HARD_REGNO_MODE_OK.
14147 (m32c_hard_regno_ok): Rename to...
14148 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
14149 (m32c_cannot_change_mode_class): Update accordingly.
14150 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14151 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
14152 (m32r_mode_class): Delete.
14153 (HARD_REGNO_MODE_OK): Delete.
14154 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14155 (m32r_hard_regno_mode_ok): Rename to...
14156 (m32r_hard_regno_modes): ...this.
14157 (m32r_mode_class): Make static.
14158 (m32r_hard_regno_mode_ok): New function.
14159 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
14160 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
14161 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14162 (m68k_hard_regno_mode_ok): Make static.
14163 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
14164 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14165 (mcore_hard_regno_mode_ok): New function.
14166 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
14167 (HARD_REGNO_MODE_OK): Delete.
14168 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
14170 (microblaze_hard_regno_mode_ok_p): ...this and make static.
14171 (microblaze_hard_regno_mode_ok): New function.
14172 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14173 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
14174 (mips_hard_regno_mode_ok): Delete.
14175 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
14176 (mips_hard_regno_mode_ok_p): ...this and make static.
14177 (mips_hard_regno_mode_ok_p): Rename to...
14178 (mips_hard_regno_mode_ok_uncached): ...this.
14179 (mips_hard_regno_mode_ok): New function.
14180 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
14181 of HARD_REGNO_MODE_OK.
14182 (mips_option_override): Update after above name changes.
14183 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14184 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
14185 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
14186 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
14187 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
14188 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14189 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
14190 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
14191 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
14192 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14193 (msp430_hard_regno_mode_ok): Make static and return a bool.
14194 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
14195 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
14196 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
14198 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14199 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
14200 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
14201 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
14202 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
14203 (PA_HARD_REGNO_MODE_OK): ...this
14204 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
14205 (PA_HARD_REGNO_MODE_OK): ...this.
14206 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14207 (pa_hard_regno_mode_ok): New function.
14208 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
14209 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14210 (pdp11_hard_regno_mode_ok): New function.
14211 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
14212 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
14214 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
14216 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14217 (rs6000_hard_regno_mode_ok): Rename to...
14218 (rs6000_hard_regno_mode_ok_uncached): ...this.
14219 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14220 (rs6000_hard_regno_mode_ok): New function.
14221 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
14222 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
14223 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
14224 (riscv_hard_regno_mode_ok): ...this and make static.
14225 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14226 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
14227 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
14228 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14229 (rl78_hard_regno_mode_ok): Make static and return bool.
14230 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
14231 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
14233 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
14234 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14235 (rs6000_hard_regno_mode_ok): Rename to...
14236 (rs6000_hard_regno_mode_ok_uncached): ...this.
14237 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14238 (rs6000_hard_regno_mode_ok): New function.
14239 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
14240 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
14241 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14242 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
14243 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
14244 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
14245 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14246 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
14247 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
14248 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14249 (sh_hard_regno_mode_ok): Make static.
14250 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
14251 instead of HARD_REGNO_MODE_OK.
14252 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
14253 (sparc_mode_class): Delete.
14254 (HARD_REGNO_MODE_OK): Delete.
14255 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14256 (hard_regno_mode_classes): Make static.
14257 (sparc_mode_class): Likewise.
14258 (sparc_hard_regno_mode_ok): New function.
14259 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
14260 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
14261 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
14263 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14264 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
14265 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
14266 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
14267 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
14268 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14269 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
14270 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
14271 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14272 (visium_hard_regno_mode_ok): New function.
14273 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
14274 instead of HARD_REGNO_MODE_OK.
14275 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
14276 (HARD_REGNO_MODE_OK): Delete.
14277 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
14278 (xtensa_hard_regno_mode_ok_p): ...this and make static.
14279 (xtensa_option_override): Update accordingly.
14280 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14281 (xtensa_hard_regno_mode_ok): New function.
14282 * system.h (HARD_REGNO_MODE_OK): Poison.
14284 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14285 Alan Hayward <alan.hayward@arm.com>
14286 David Sherwood <david.sherwood@arm.com>
14288 * target.def (hard_regno_call_part_clobbered): New hook.
14289 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
14290 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
14291 * doc/tm.texi: Regenerate.
14292 * hooks.h (hook_bool_uint_mode_false): Declare.
14293 * hooks.c (hook_bool_uint_mode_false): New function.
14294 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14295 * cselib.c (cselib_process_insn): Use
14296 targetm.hard_regno_call_part_clobbered instead of
14297 HARD_REGNO_CALL_PART_CLOBBERED.
14298 * ira-conflicts.c (ira_build_conflicts): Likewise.
14299 * ira-costs.c (ira_tune_allocno_costs): Likewise.
14300 * lra-constraints.c (need_for_call_save_p): Likewise.
14301 * lra-lives.c: Include target.h.
14302 (check_pseudos_live_through_calls): Use
14303 targetm.hard_regno_call_part_clobbered instead of
14304 HARD_REGNO_CALL_PART_CLOBBERED.
14305 * regcprop.c: Include target.h.
14306 (copyprop_hardreg_forward_1): Use
14307 targetm.hard_regno_call_part_clobbered instead of
14308 HARD_REGNO_CALL_PART_CLOBBERED.
14309 * reginfo.c (choose_hard_reg_mode): Likewise.
14310 * regrename.c (check_new_reg_p): Likewise.
14311 * reload.c (find_equiv_reg): Likewise.
14312 * reload1.c (emit_reload_insns): Likewise.
14313 * sched-deps.c (deps_analyze_insn): Likewise.
14314 * sel-sched.c (init_regs_for_mode): Likewise.
14315 (mark_unavailable_hard_regs): Likewise.
14316 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
14317 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14318 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
14320 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14321 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14322 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
14324 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
14326 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14327 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14328 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
14330 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14331 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14332 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
14334 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14335 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
14337 * config/powerpcspe/powerpcspe.c
14338 (rs6000_hard_regno_call_part_clobbered): New function.
14339 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14340 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14341 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
14343 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14344 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14345 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
14347 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14348 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14349 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
14351 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14352 Alan Hayward <alan.hayward@arm.com>
14353 David Sherwood <david.sherwood@arm.com>
14355 * rtl.h (subreg_memory_offset): Declare.
14356 * emit-rtl.c (subreg_memory_offset): New function.
14357 * expmed.c (store_bit_field_1): Use it.
14358 * expr.c (undefined_operand_subword_p): Likewise.
14359 * simplify-rtx.c (simplify_subreg): Likewise.
14361 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
14363 PR rtl-optimization/57448
14366 * optabs.c (expand_atomic_load): Place compiler memory barriers if
14367 using atomic_load pattern.
14368 (expand_atomic_store): Likewise.
14370 2017-09-04 Jakub Jelinek <jakub@redhat.com>
14373 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
14374 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
14375 handling. Use replace_call_with_value with NULL instead of
14376 gsi_replace, unlink_stmt_vdef and release_defs.
14378 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
14381 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
14383 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14386 * rtl.h (emit_library_call_value_1): Declare.
14387 (emit_library_call): Replace declaration with a series of overloads.
14388 Remove the parameter count argument.
14389 (emit_library_call_value): Likewise.
14390 * calls.c (emit_library_call_value_1): Make global. Replace varargs
14391 with an "rtx_mode_t *".
14392 (emit_library_call_value): Delete.
14393 (emit_library_call): Likewise.
14394 * asan.c (asan_emit_stack_protection): Update calls accordingly.
14395 (asan_emit_allocas_unpoison): Likewise.
14396 * builtins.c (expand_builtin_powi): Likewise.
14397 (expand_asan_emit_allocas_unpoison): Likewise.
14398 * cfgexpand.c (expand_main_function): Likewise.
14399 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
14400 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
14401 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
14402 * config/arm/arm.c (arm_trampoline_init): Likewise.
14403 (arm_call_tls_get_addr): Likewise.
14404 (arm_expand_divmod_libfunc): Likewise.
14405 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
14406 (smulsi3_highpart): Likewise.
14407 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
14408 (c6x_expand_compare): Likewise.
14409 (c6x_expand_movmem): Likewise.
14410 * config/frv/frv.c (frv_trampoline_init): Likewise.
14411 * config/i386/i386.c (ix86_trampoline_init): Likewise.
14412 (ix86_expand_divmod_libfunc): Likewise.
14413 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
14414 (ia64_expand_compare): Likewise.
14415 (ia64_profile_hook): Likewise.
14416 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
14417 (nonlocal_goto): Likewise.
14418 (restore_stack_nonlocal): Likewise.
14419 * config/m32r/m32r.c (block_move_call): Likewise.
14420 (m32r_trampoline_init): Likewise.
14421 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
14422 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
14423 (m68k_call_m68k_read_tp): Likewise.
14424 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
14425 (microblaze_expand_divide): Likewise.
14426 * config/mips/mips.h (mips_args): Likewise.
14427 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
14428 (MIPS_ICACHE_SYNC): Likewise.
14429 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
14430 (nios2_trampoline_init): Likewise.
14431 * config/pa/pa.c (hppa_tls_call): Likewise.
14432 (pa_trampoline_init): Likewise.
14433 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
14434 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
14435 (expand_strn_compare): Likewise.
14436 (rs6000_generate_compare): Likewise.
14437 (rs6000_expand_float128_convert): Likewise.
14438 (output_profile_hook): Likewise.
14439 (rs6000_trampoline_init): Likewise.
14440 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
14441 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
14442 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
14443 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
14444 (rs6000_generate_compare): Likewise.
14445 (rs6000_expand_float128_convert): Likewise.
14446 (output_profile_hook): Likewise.
14447 (rs6000_trampoline_init): Likewise.
14448 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
14449 * config/sh/sh.c (sh_trampoline_init): Likewise.
14450 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
14451 (sparc_emit_float_lib_cmp): Likewise.
14452 (sparc32_initialize_trampoline): Likewise.
14453 (sparc64_initialize_trampoline): Likewise.
14454 (sparc_profile_hook): Likewise.
14455 * config/spu/spu.c (ea_load_store): Likewise.
14456 * config/spu/spu.md (floatunssidf2): Likewise.
14457 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
14458 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
14459 * config/visium/visium.c (expand_block_move_4): Likewise.
14460 (expand_block_move_2): Likewise.
14461 (expand_block_move_1): Likewise.
14462 (expand_block_set_4): Likewise.
14463 (expand_block_set_2): Likewise.
14464 (expand_block_set_1): Likewise.
14465 (visium_trampoline_init): Likewise.
14466 (visium_profile_hook): Likewise.
14467 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
14468 (xtensa_setup_frame_addresses): Likewise.
14469 (xtensa_trampoline_init): Likewise.
14470 * except.c (sjlj_emit_function_enter): Likewise.
14471 (sjlj_emit_function_exit): Likewise.
14472 * explow.c (allocate_dynamic_stack_space): Likewise.
14473 (probe_stack_range): Likewise.
14474 * expr.c (convert_mode_scalar): Likewise.
14475 * optabs.c (expand_binop): Likewise.
14476 (expand_twoval_binop_libfunc): Likewise.
14477 (expand_unop): Likewise.
14478 (prepare_cmp_insn): Likewise.
14479 (prepare_float_lib_cmp): Likewise.
14480 (expand_float): Likewise.
14481 (expand_fix): Likewise.
14482 (expand_fixed_convert): Likewise.
14483 (maybe_emit_sync_lock_test_and_set): Likewise.
14484 (expand_atomic_compare_and_swap): Likewise.
14485 (expand_mem_thread_fence): Likewise.
14486 (expand_atomic_fetch_op): Likewise.
14488 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
14490 * doc/generic.texi (OpenACC): Adjust URL.
14491 * doc/invoke.texi (C Dialect Options): Ditto.
14493 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
14495 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
14496 predicate for operand 1. Add (m,<S>) constraint.
14497 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
14498 Prevent memory operand 1 with register operand 2.
14500 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
14502 PR rtl-optimization/82024
14503 * combine.c (try_combine): If the combination result is a PARALLEL,
14504 and we only need to retain the SET in there that would be placed
14505 at I2, check that we can place that at I3 instead, before doing so.
14507 2017-09-01 Jakub Jelinek <jakub@redhat.com>
14510 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
14512 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
14513 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
14516 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
14517 Jeff Law <law@redhat.com>
14519 * varasm.c (bss_initializer_p): Do not put constants into .bss
14520 (categorize_decl_for_section): Handle bss_initializer_p returning
14521 false when DECL_INITIAL is NULL.
14523 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14526 * config/s390/s390.c (s390_can_inline_p): New function.
14528 2017-09-01 Jeff Law <law@redhat.com>
14530 PR tree-optimization/82052
14531 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
14532 Always initialize the returned slot after a hash table miss
14533 when INSERT is true.
14535 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
14537 * config/s390/s390.md (mem_signal_fence): Remove.
14538 * doc/md.texi (mem_signal_fence): Remove.
14539 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
14541 * target-insns.def (mem_signal_fence): Remove.
14543 2017-09-01 Jakub Jelinek <jakub@redhat.com>
14546 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
14549 * asan.c (create_odr_indicator): Strip name encoding from assembler
14550 name before appending it after __odr_asan_.
14552 2017-09-01 Martin Liska <mliska@suse.cz>
14554 PR tree-optimization/82059
14555 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
14556 frequency only when an edge is redirected.
14558 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14560 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
14561 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
14562 (arc_conditional_register_usage): Remove ARC600 lp_count
14564 (arc_file_start): Emit Tag_ARC_CPU_variation.
14565 (arc_can_use_doloop_p): New conditions to use ZOLs.
14566 (hwloop_fail): New function.
14567 (hwloop_optimize): Likewise.
14568 (hwloop_pattern_reg): Likewise.
14569 (arc_doloop_hooks): New struct, to be used with reorg_loops.
14570 (arc_reorg_loops): New function, calls reorg_loops.
14571 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
14572 (arc600_corereg_hazard): Remove ZOL checking, case handled by
14574 (arc_loop_hazard): Remove function, functionality moved into
14576 (arc_hazard): Remove arc_loop_hazard call.
14577 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
14578 into hwloop_optimize.
14579 (arc_label_align): Remove ZOL handling.
14580 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
14581 * config/arc/arc.md (doloop_begin): Remove pattern.
14582 (doloop_begin_i): Likewise.
14583 (doloop_end_i): Likewise.
14584 (doloop_fallback): Likewise.
14585 (doloop_fallback_m): Likewise.
14586 (doloop_end): Reimplement expand.
14587 (arc_lp): New pattern for LP instruction.
14588 (loop_end): New pattern.
14589 (loop_fail): Likewise.
14590 (decrement_and_branch_until_zero): Likewise.
14591 * config/arc/arc.opt (mlpc-width): New option.
14592 * doc/invoke.texi (mlpc-width): Document option.
14594 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14596 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
14597 (arc_ccfsm_advance): Fix checking for delay slots.
14598 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
14600 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14602 * config/arc/arc.md (movqi_insn): Add stores to save constant long
14604 (movhi_insn): Update store instruction constraint which are saving
14605 6-bit short immediates.
14606 (movsi_insn): Consider also short scaled load operations.
14607 (zero_extendhisi2_i): Use Usd constraint instead of T.
14608 (extendhisi2_i): Add q constraint.
14609 (arc_clzsi2): Add type and length attributes.
14610 (arc_ctzsi2): Likewise.
14611 * config/arc/constraints.md (Usc): Update constraint, the
14612 assembler can parse two relocations for a single instruction.
14614 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14616 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
14617 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
14619 2017-08-31 Olivier Hainque <hainque@adacore.com>
14621 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
14622 match as powerpc-wrs-vxworks*.
14624 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
14626 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
14627 register constraint for by-element operand.
14628 (aarch64_mls_elt_merge<mode>): Likewise.
14630 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14632 * config/arc/arc.c (arc_can_follow_jump): Check for short
14635 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14637 * config.gcc: Use g.opt for arc.
14638 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
14639 functionality moved to ...
14640 (legitimate_scaled_address_p): New function, ...here.
14641 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
14642 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
14643 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
14645 (arc_override_options): Handle G option.
14646 (arc_output_pic_addr_const): Correct function definition.
14647 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
14648 (arc_decl_anon_ns_mem_p): Delete.
14649 (arc_in_small_data_p): Overhaul this function to take into
14650 consideration the value given via G option.
14651 (arc_rewrite_small_data_1): Renamed and corrected old
14652 arc_rewrite_small_data function.
14653 (arc_rewrite_small_data): New function.
14654 (small_data_pattern): Don't use pic_offset_table_rtx.
14655 * config/arc/arc.h (CC1_SPEC): Recognize G option.
14656 * config/arc/simdext.md (movmisalignv2hi): Use
14657 prepare_move_operands function.
14659 (movmisalign*): Likewise.
14660 * doc/invoke.texi (ARC options): Document -G option.
14662 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14664 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
14666 * config/arc/arc.c (arc_print_operand): Output scalled address for
14667 sdata whenever is possible.
14668 (arc_in_small_data_p): Allow sdata for 64bit datum when double
14669 load/stores are available.
14670 (compact_sda_memory_operand): Check for the alignment required by
14671 code density instructions.
14672 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
14674 * config/arc/constraints.md (Usd): Update constraint.
14675 (Us0): New constraint.
14676 (Usc): Update constraint.
14678 2017-08-31 Richard Biener <rguenther@suse.de>
14680 PR middle-end/82054
14681 * dwarf2out.c (dwarf2out_early_global_decl): Process each
14682 function only once.
14684 2017-08-31 Tamar Christina <tamar.christina@arm.com>
14686 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
14687 Resize type_signature.
14689 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14690 Alan Hayward <alan.hayward@arm.com>
14691 David Sherwood <david.sherwood@arm.com>
14693 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
14694 subregs whose inner modes can be stored in GPRs.
14695 (aarch64_classify_index): Likewise.
14697 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14698 Alan Hayward <alan.hayward@arm.com>
14699 David Sherwood <david.sherwood@arm.com>
14701 * config/aarch64/iterators.md (V_cmp_result): Rename to...
14702 (V_INT_EQUIV): ...this.
14703 (v_cmp_result): Rename to...
14704 (v_int_equiv): ...this.
14705 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
14706 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
14707 (copysign<mode>3): Likewise.
14708 (aarch64_simd_bsl<mode>_internal): Likewise.
14709 (aarch64_simd_bsl<mode>): Likewise.
14710 (vec_cmp<mode><mode>): Likewise.
14711 (vcond<mode><mode>): Likewise.
14712 (vcond<v_cmp_mixed><mode>): Likewise.
14713 (vcondu<mode><v_cmp_mixed>): Likewise.
14714 (aarch64_cm<optab><mode>): Likewise.
14715 (aarch64_cmtst<mode>): Likewise.
14716 (aarch64_fac<optab><mode>): Likewise.
14717 (vec_perm_const<mode>): Likewise.
14718 (vcond_mask_<mode><v_cmp_result>): Rename to...
14719 (vcond_mask_<mode><v_int_equiv>): ...this.
14720 (vec_cmp<mode><v_cmp_result>): Rename to...
14721 (vec_cmp<mode><v_int_equiv>): ...this.
14723 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14724 Alan Hayward <alan.hayward@arm.com>
14725 David Sherwood <david.sherwood@arm.com>
14727 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
14729 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
14730 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
14731 (UNSPEC_LD4_DREG): New unspecs.
14732 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
14733 (aarch64_ld2<mode>_dreg_be): Replace with...
14734 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
14736 (aarch64_ld3<mode>_dreg_le)
14737 (aarch64_ld3<mode>_dreg_be): Replace with...
14738 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
14740 (aarch64_ld4<mode>_dreg_le)
14741 (aarch64_ld4<mode>_dreg_be): Replace with...
14742 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
14745 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14747 PR tree-optimization/81987
14748 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
14749 insert an initializer in a location not dominated by the stride
14752 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
14754 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
14755 on the entire header of the finally block in the fallthru case.
14757 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
14759 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
14761 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
14763 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
14764 rs6000_emit_move_from_cr and call renamed function.
14765 (rs6000_emit_prologue): Call renamed functions.
14766 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
14767 movesi_from_cr, remove volatile CRs.
14769 2017-08-30 Jon Beniston <jon@beniston.com>
14770 Richard Biener <rguenther@suse.de>
14772 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
14773 of VECTOR_MODE_P check.
14774 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
14775 element vector types.
14777 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14779 * df.h (df_read_modify_subreg_p): Remove in favor of...
14780 * rtl.h (read_modify_subreg_p): ...this new function. Take a
14781 const_rtx instead of an rtx.
14782 * cprop.c (local_cprop_find_used_regs): Update accordingly.
14783 * df-problems.c (df_word_lr_mark_ref): Likewise.
14784 * ira-lives.c (mark_pseudo_reg_live): Likewise.
14785 (mark_pseudo_reg_dead): Likewise.
14786 (mark_ref_dead): Likewise.
14787 * reginfo.c (init_subregs_of_mode): Likewise.
14788 * sched-deps.c (sched_analyze_1): Likewise.
14789 * df-scan.c (df_def_record_1): Likewise.
14790 (df_uses_record): Likewise.
14791 (df_read_modify_subreg_p): Remove in favor of...
14792 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
14793 const_rtx instead of an rtx.
14795 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14796 Alan Hayward <alan.hayward@arm.com>
14797 David Sherwood <david.sherwood@arm.com>
14799 * rtl.h (partial_subreg_p): New function.
14800 * caller-save.c (save_call_clobbered_regs): Use it.
14801 * calls.c (expand_call): Likewise.
14802 * combine.c (combinable_i3pat): Likewise.
14803 (simplify_set): Likewise.
14804 (make_extraction): Likewise.
14805 (make_compound_operation_int): Likewise.
14806 (gen_lowpart_or_truncate): Likewise.
14807 (force_to_mode): Likewise.
14808 (make_field_assignment): Likewise.
14809 (reg_truncated_to_mode): Likewise.
14810 (record_truncated_value): Likewise.
14811 (move_deaths): Likewise.
14812 * cse.c (record_jump_cond): Likewise.
14813 (cse_insn): Likewise.
14814 * cselib.c (cselib_lookup_1): Likewise.
14815 * expmed.c (extract_bit_field_using_extv): Likewise.
14816 * function.c (assign_parm_setup_reg): Likewise.
14817 * ifcvt.c (noce_convert_multiple_sets): Likewise.
14818 * ira-build.c (create_insn_allocnos): Likewise.
14819 * lra-coalesce.c (merge_pseudos): Likewise.
14820 * lra-constraints.c (match_reload): Likewise.
14821 (simplify_operand_subreg): Likewise.
14822 (curr_insn_transform): Likewise.
14823 * lra-lives.c (process_bb_lives): Likewise.
14824 * lra.c (new_insn_reg): Likewise.
14825 (lra_substitute_pseudo): Likewise.
14826 * regcprop.c (mode_change_ok): Likewise.
14827 (maybe_mode_change): Likewise.
14828 (copyprop_hardreg_forward_1): Likewise.
14829 * reload.c (push_reload): Likewise.
14830 (find_reloads): Likewise.
14831 (find_reloads_subreg_address): Likewise.
14832 * reload1.c (alter_reg): Likewise.
14833 (eliminate_regs_1): Likewise.
14834 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14836 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
14838 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
14839 back to if statements, including unpack.
14841 2017-08-30 Martin Liska <mliska@suse.cz>
14843 PR inline-asm/82001
14844 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
14846 (func_checker::compare_asm_inputs_outputs): ... this function.
14847 (func_checker::compare_gimple_asm): Use the function to compare
14848 also ASM constrains.
14849 * ipa-icf-gimple.h: Rename the function.
14851 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14852 Alan Hayward <alan.hayward@arm.com>
14853 David Sherwood <david.sherwood@arm.com>
14855 * coretypes.h (complex_mode): New type.
14856 * gdbhooks.py (build_pretty_printer): Handle it.
14857 * machmode.h (complex_mode): New class.
14858 (complex_mode::includes_p): New function.
14859 (is_complex_int_mode): Likewise.
14860 (is_complex_float_mode): Likewise.
14861 * genmodes.c (get_mode_class): Handle complex mode classes.
14862 * function.c (expand_function_end): Use is_complex_int_mode.
14864 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14865 Alan Hayward <alan.hayward@arm.com>
14866 David Sherwood <david.sherwood@arm.com>
14868 * coretypes.h (scalar_mode_pod): New typedef.
14869 * gdbhooks.py (build_pretty_printer): Handle it.
14870 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
14871 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
14872 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
14873 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
14874 as_a <scalar_mode>.
14876 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14877 Alan Hayward <alan.hayward@arm.com>
14878 David Sherwood <david.sherwood@arm.com>
14880 * machmode.h (mode_for_vector): Take a scalar_mode instead
14882 * stor-layout.c (mode_for_vector): Likewise.
14883 * explow.c (promote_mode): Use as_a <scalar_mode>.
14884 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
14886 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14887 Alan Hayward <alan.hayward@arm.com>
14888 David Sherwood <david.sherwood@arm.com>
14890 * target.def (preferred_simd_mode): Take a scalar_mode
14891 instead of a machine_mode.
14892 * targhooks.h (default_preferred_simd_mode): Likewise.
14893 * targhooks.c (default_preferred_simd_mode): Likewise.
14894 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
14895 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
14896 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
14897 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
14898 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
14899 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
14900 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
14901 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
14903 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
14904 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
14905 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
14906 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
14907 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
14908 * doc/tm.texi: Regenerate.
14909 * optabs-query.c (can_vec_mask_load_store_p): Return false for
14912 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14913 Alan Hayward <alan.hayward@arm.com>
14914 David Sherwood <david.sherwood@arm.com>
14916 * target.def (scalar_mode_supported_p): Take a scalar_mode
14917 instead of a machine_mode.
14918 * targhooks.h (default_scalar_mode_supported_p): Likewise.
14919 * targhooks.c (default_scalar_mode_supported_p): Likewise.
14920 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
14921 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
14922 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
14923 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
14924 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
14925 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
14926 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
14927 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
14928 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
14929 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
14930 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
14931 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
14933 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
14934 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
14935 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
14936 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
14937 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
14939 * doc/tm.texi: Regenerate.
14941 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14942 Alan Hayward <alan.hayward@arm.com>
14943 David Sherwood <david.sherwood@arm.com>
14945 * coretypes.h (opt_scalar_mode): New typedef.
14946 * gdbhooks.py (build_pretty_printers): Handle it.
14947 * machmode.h (mode_iterator::get_2xwider): Add overload for
14949 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
14951 * expr.c (convert_mode_scalar): Likewise.
14952 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14953 * optabs.c (expand_float): Likewise.
14954 (expand_fix): Likewise.
14955 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14957 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14958 Alan Hayward <alan.hayward@arm.com>
14959 David Sherwood <david.sherwood@arm.com>
14961 * optabs.c (expand_float): Explicitly check for scalars before
14962 using a branching expansion.
14963 (expand_fix): Likewise.
14965 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14966 Alan Hayward <alan.hayward@arm.com>
14967 David Sherwood <david.sherwood@arm.com>
14969 * expr.c (convert_mode): Split scalar handling out into...
14970 (convert_mode_scalar): ...this new function. Treat the modes
14973 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14974 Alan Hayward <alan.hayward@arm.com>
14975 David Sherwood <david.sherwood@arm.com>
14977 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
14979 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
14981 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14982 Alan Hayward <alan.hayward@arm.com>
14983 David Sherwood <david.sherwood@arm.com>
14985 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
14986 rather than a machine_mode.
14987 (fixed_from_string): Likewise.
14988 (fixed_convert): Likewise.
14989 (fixed_convert_from_int): Likewise.
14990 (fixed_convert_from_real): Likewise.
14991 (real_convert_from_fixed): Likewise.
14992 * fixed-value.c (fixed_from_double_int): Likewise.
14993 (fixed_from_string): Likewise.
14994 (fixed_convert): Likewise.
14995 (fixed_convert_from_int): Likewise.
14996 (fixed_convert_from_real): Likewise.
14997 (real_convert_from_fixed): Likewise.
14998 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
15000 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15001 Alan Hayward <alan.hayward@arm.com>
15002 David Sherwood <david.sherwood@arm.com>
15004 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
15005 of separate mode class checks. Do not allow vector modes here.
15006 (immed_wide_int_const): Use as_a <scalar_mode>.
15007 * explow.c (trunc_int_for_mode): Likewise.
15008 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
15009 (wi::shwi): Likewise.
15010 (wi::min_value): Likewise.
15011 (wi::max_value): Likewise.
15012 * dwarf2out.c (loc_descriptor): Likewise.
15013 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
15014 for CONST_WIDE_INT.
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 * tree.h (SCALAR_TYPE_MODE): New macro.
15021 * expr.c (expand_expr_addr_expr_1): Use it.
15022 (expand_expr_real_2): Likewise.
15023 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
15024 (fold_convert_const_fixed_from_int): Likewise.
15025 (fold_convert_const_fixed_from_real): Likewise.
15026 (native_encode_fixed): Likewise
15027 (native_encode_complex): Likewise
15028 (native_encode_vector): Likewise.
15029 (native_interpret_fixed): Likewise.
15030 (native_interpret_real): Likewise.
15031 (native_interpret_complex): Likewise.
15032 (native_interpret_vector): Likewise.
15033 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
15034 (simd_clone_adjust_argument_types): Likewise.
15035 (simd_clone_init_simd_arrays): Likewise.
15036 (simd_clone_adjust): Likewise.
15037 * stor-layout.c (layout_type): Likewise.
15038 * tree.c (build_minus_one_cst): Likewise.
15039 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15040 * tree-inline.c (estimate_move_cost): Likewise.
15041 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
15042 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
15043 (vectorizable_reduction): Likewise.
15044 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
15045 (vect_recog_mixed_size_cond_pattern): Likewise.
15046 (check_bool_pattern): Likewise.
15047 (adjust_bool_pattern): Likewise.
15048 (search_type_for_mask_1): Likewise.
15049 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
15050 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15051 (vectorizable_load): Likewise.
15052 (vectorizable_store): Likewise.
15053 * ubsan.c (ubsan_encode_value): Likewise.
15054 * varasm.c (output_constant): Likewise.
15056 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15057 Alan Hayward <alan.hayward@arm.com>
15058 David Sherwood <david.sherwood@arm.com>
15060 * coretypes.h (scalar_mode): New class.
15061 * machmode.h (scalar_mode): Likewise.
15062 (scalar_mode::includes_p): New function.
15063 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
15064 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
15065 * genmodes.c (get_mode_class): Handle remaining scalar modes.
15066 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
15067 * expmed.c (store_bit_field_1): Likewise.
15068 (extract_bit_field_1): Likewise.
15069 * expr.c (write_complex_part): Likewise.
15070 (read_complex_part): Likewise.
15071 (emit_move_complex_push): Likewise.
15072 (expand_expr_real_2): Likewise.
15073 * function.c (assign_parm_setup_reg): Likewise.
15074 (assign_parms_unsplit_complex): Likewise.
15075 * optabs.c (expand_binop): Likewise.
15076 * rtlanal.c (subreg_get_info): Likewise.
15077 * simplify-rtx.c (simplify_immed_subreg): Likewise.
15078 * varasm.c (output_constant_pool_2): Likewise.
15080 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15081 Alan Hayward <alan.hayward@arm.com>
15082 David Sherwood <david.sherwood@arm.com>
15084 * expmed.c (extract_high_half): Use scalar_int_mode and remove
15086 (expmed_mult_highpart_optab): Likewise.
15087 (expmed_mult_highpart): Likewise.
15089 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15090 Alan Hayward <alan.hayward@arm.com>
15091 David Sherwood <david.sherwood@arm.com>
15093 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
15094 instead of a machine_mode.
15095 (builtin_memset_read_str): Likewise.
15096 * builtins.c (c_readstr): Likewise.
15097 (builtin_memcpy_read_str): Likewise.
15098 (builtin_strncpy_read_str): Likewise.
15099 (builtin_memset_read_str): Likewise.
15100 (builtin_memset_gen_str): Likewise.
15101 (expand_builtin_signbit): Use scalar_int_mode for local variables.
15102 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
15103 instead of a machine_mode.
15104 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
15106 (make_extraction): Likewise.
15107 (try_widen_shift_mode): Take and return scalar_int_modes instead
15109 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
15110 a scalar_int_mode instead of a machine_mode.
15111 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
15112 (avr_addr_space_pointer_mode): Likewise.
15113 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
15114 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
15115 (msp430_unwind_word_mode): Likewise.
15116 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
15117 (spu_addr_space_pointer_mode): Likewise.
15118 (spu_addr_space_address_mode): Likewise.
15119 (spu_libgcc_cmp_return_mode): Likewise.
15120 (spu_libgcc_shift_count_mode): Likewise.
15121 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
15122 (rl78_addr_space_pointer_mode): Likewise.
15123 (fl78_unwind_word_mode): Likewise.
15124 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
15126 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
15127 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
15128 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
15129 (mips_valid_pointer_mode): Likewise.
15130 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
15131 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
15132 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
15134 (ft32_addr_space_address_mode): Likewise.
15135 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
15136 scalar_int_mode instead of a machine_mode.
15137 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
15139 (m32c_addr_space_address_mode): Likewise.
15140 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
15141 (rs6000_eh_return_filter_mode): Likewise.
15142 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
15143 (rs6000_eh_return_filter_mode): Likewise.
15144 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
15145 (s390_libgcc_shift_count_mode): Likewise.
15146 (s390_unwind_word_mode): Likewise.
15147 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
15149 * target.def (mode_rep_extended): Likewise.
15150 (valid_pointer_mode): Likewise.
15151 (addr_space.valid_pointer_mode): Likewise.
15152 (eh_return_filter_mode): Return a scalar_int_mode rather than
15154 (libgcc_cmp_return_mode): Likewise.
15155 (libgcc_shift_count_mode): Likewise.
15156 (unwind_word_mode): Likewise.
15157 (addr_space.pointer_mode): Likewise.
15158 (addr_space.address_mode): Likewise.
15159 * doc/tm.texi: Regenerate.
15160 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
15162 (do_jump): Use scalar_int_mode for local variables.
15163 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
15164 rather than a machine_mode.
15165 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
15166 (scompare_loc_descriptor_wide): Likewise.
15167 (scompare_loc_descriptor_narrow): Likewise.
15168 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
15170 * except.c (sjlj_emit_dispatch_table): Likewise.
15171 (expand_builtin_eh_copy_values): Likewise.
15172 * explow.c (convert_memory_address_addr_space_1): Likewise.
15173 Take a scalar_int_mode rather than a machine_mode.
15174 (convert_memory_address_addr_space): Take a scalar_int_mode rather
15175 than a machine_mode.
15176 (memory_address_addr_space): Use scalar_int_mode for local variables.
15177 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
15178 rather than a machine_mode.
15179 * expmed.c (mask_rtx): Likewise.
15180 (init_expmed_one_conv): Likewise.
15181 (expand_mult_highpart_adjust): Likewise.
15182 (extract_high_half): Likewise.
15183 (expmed_mult_highpart_optab): Likewise.
15184 (expmed_mult_highpart): Likewise.
15185 (expand_smod_pow2): Likewise.
15186 (expand_sdiv_pow2): Likewise.
15187 (emit_store_flag_int): Likewise.
15188 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
15190 (extract_low_bits): Likewise.
15191 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
15193 * expr.c (pieces_addr::adjust): Likewise.
15194 (can_store_by_pieces): Likewise.
15195 (store_by_pieces): Likewise.
15196 (clear_by_pieces_1): Likewise.
15197 (expand_expr_addr_expr_1): Likewise.
15198 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
15199 (expand_expr_real_1): Likewise.
15200 (try_casesi): Likewise.
15201 * final.c (shorten_branches): Likewise.
15202 * fold-const.c (fold_convert_const_int_from_fixed): Change the
15203 type of "mode" to machine_mode.
15204 * internal-fn.c (expand_arith_overflow_result_store): Take a
15205 scalar_int_mode rather than a machine_mode.
15206 (expand_mul_overflow): Use scalar_int_mode for local variables.
15207 * loop-doloop.c (doloop_modify): Likewise.
15208 (doloop_optimize): Likewise.
15209 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
15210 than a machine_mode.
15211 (expand_doubleword_shift_condmove): Likewise.
15212 (expand_doubleword_shift): Likewise.
15213 (expand_doubleword_clz): Likewise.
15214 (expand_doubleword_popcount): Likewise.
15215 (expand_doubleword_parity): Likewise.
15216 (expand_absneg_bit): Use scalar_int_mode for local variables.
15217 (prepare_float_lib_cmp): Likewise.
15218 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
15219 rather than a machine_mode.
15220 (convert_memory_address_addr_space): Likewise.
15221 (get_mode_bounds): Likewise.
15222 (get_address_mode): Return a scalar_int_mode rather than a
15224 * rtlanal.c (get_address_mode): Likewise.
15225 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
15226 than a machine_mode.
15227 * targhooks.c (default_mode_rep_extended): Likewise.
15228 (default_valid_pointer_mode): Likewise.
15229 (default_addr_space_valid_pointer_mode): Likewise.
15230 (default_eh_return_filter_mode): Return a scalar_int_mode rather
15231 than a machine_mode.
15232 (default_libgcc_cmp_return_mode): Likewise.
15233 (default_libgcc_shift_count_mode): Likewise.
15234 (default_unwind_word_mode): Likewise.
15235 (default_addr_space_pointer_mode): Likewise.
15236 (default_addr_space_address_mode): Likewise.
15237 * targhooks.h (default_eh_return_filter_mode): Likewise.
15238 (default_libgcc_cmp_return_mode): Likewise.
15239 (default_libgcc_shift_count_mode): Likewise.
15240 (default_unwind_word_mode): Likewise.
15241 (default_addr_space_pointer_mode): Likewise.
15242 (default_addr_space_address_mode): Likewise.
15243 (default_mode_rep_extended): Take a scalar_int_mode rather than
15245 (default_valid_pointer_mode): Likewise.
15246 (default_addr_space_valid_pointer_mode): Likewise.
15247 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
15249 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
15250 rather than a machine_mode.
15251 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
15252 for local variables.
15253 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15254 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
15255 than a machine_mode.
15257 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15258 Alan Hayward <alan.hayward@arm.com>
15259 David Sherwood <david.sherwood@arm.com>
15261 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
15262 the mode argument to scalar_int_mode.
15263 (do_jump_by_parts_zero_rtx): Likewise.
15264 (do_jump_by_parts_equality_rtx): Likewise.
15265 (do_jump_by_parts_greater): Take a mode argument.
15266 (do_jump_by_parts_equality): Likewise.
15267 (do_jump_1): Update calls accordingly.
15269 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15270 Alan Hayward <alan.hayward@arm.com>
15271 David Sherwood <david.sherwood@arm.com>
15273 * is-a.h (safe_dyn_cast): New function.
15274 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
15275 (jump_table_for_label): Likewise.
15276 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
15277 instead of an rtx_insn *.
15278 (shorten_branches): Use dyn_cast instead of LABEL_P and
15279 JUMP_TABLE_DATA_P. Use jump_table_for_label and
15280 rtx_jump_table_data::get_data_mode.
15281 (final_scan_insn): Likewise.
15283 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15284 Alan Hayward <alan.hayward@arm.com>
15285 David Sherwood <david.sherwood@arm.com>
15287 * combine.c (try_combine): Use is_a <scalar_int_mode> when
15288 trying to combine a full-register integer set with a subreg
15291 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15292 Alan Hayward <alan.hayward@arm.com>
15293 David Sherwood <david.sherwood@arm.com>
15295 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
15296 that is always either address_mode or pointer_mode.
15298 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15299 Alan Hayward <alan.hayward@arm.com>
15300 David Sherwood <david.sherwood@arm.com>
15302 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
15303 when the two are known to be equal.
15305 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15306 Alan Hayward <alan.hayward@arm.com>
15307 David Sherwood <david.sherwood@arm.com>
15309 * simplify-rtx.c (simplify_const_unary_operation): Use
15310 is_a <scalar_int_mode> instead of checking for a nonzero
15311 precision. Forcibly convert op_mode to a scalar_int_mode
15312 in that case. More clearly differentiate the operand and
15313 result modes and use the former when deciding what the value
15314 of a count-bits operation should be. Use is_int_mode instead
15315 of checking for a MODE_INT. Remove redundant check for whether
15316 this mode has a zero precision.
15318 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15319 Alan Hayward <alan.hayward@arm.com>
15320 David Sherwood <david.sherwood@arm.com>
15322 * optabs.c (widen_leading): Change the type of the mode argument
15323 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
15324 (widen_bswap): Likewise.
15325 (expand_parity): Likewise.
15326 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
15327 (expand_ffs): Likewise.
15328 (epand_unop): Check for scalar integer modes before calling the
15331 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15332 Alan Hayward <alan.hayward@arm.com>
15333 David Sherwood <david.sherwood@arm.com>
15335 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
15337 (expand_expr_real_1): Update call accordingly.
15339 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15340 Alan Hayward <alan.hayward@arm.com>
15341 David Sherwood <david.sherwood@arm.com>
15343 * expmed.c (store_bit_field_using_insv): Add op0_mode and
15344 value_mode arguments. Use scalar_int_mode internally.
15345 (store_bit_field_1): Rename the new integer mode from imode
15346 to op0_mode and use it instead of GET_MODE (op0). Update calls
15347 to store_split_bit_field, store_bit_field_using_insv and
15348 store_fixed_bit_field.
15349 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
15350 Use scalar_int_mode internally. Use a bit count rather than a mode
15351 when calculating the largest bit size for get_best_mode.
15352 Update calls to store_split_bit_field and store_fixed_bit_field_1.
15353 (store_fixed_bit_field_1): Add mode and value_mode arguments.
15354 Remove assertion that OP0 has a scalar integer mode.
15355 (store_split_bit_field): Add op0_mode and value_mode arguments.
15356 Update calls to extract_fixed_bit_field.
15357 (extract_bit_field_using_extv): Add an op0_mode argument.
15358 Use scalar_int_mode internally.
15359 (extract_bit_field_1): Rename the new integer mode from imode to
15360 op0_mode and use it instead of GET_MODE (op0). Update calls to
15361 extract_split_bit_field, extract_bit_field_using_extv and
15362 extract_fixed_bit_field.
15363 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
15364 to extract_split_bit_field and extract_fixed_bit_field_1.
15365 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
15366 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
15367 on the target mode.
15368 (extract_split_bit_field): Add an op0_mode argument. Update call
15369 to extract_fixed_bit_field.
15371 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15372 Alan Hayward <alan.hayward@arm.com>
15373 David Sherwood <david.sherwood@arm.com>
15375 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
15376 * explow.c (hard_function_value): Likewise.
15377 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
15378 convert_to_mode call outside the loop.
15379 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
15380 for the mode iterator. Require the mode specified by max_pieces
15382 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
15384 (copy_blkmode_to_reg): Likewise.
15385 (set_storage_via_setmem): Likewise.
15386 * optabs.c (prepare_cmp_insn): Likewise.
15387 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15388 * stor-layout.c (finish_bitfield_representative): Likewise.
15390 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15391 Alan Hayward <alan.hayward@arm.com>
15392 David Sherwood <david.sherwood@arm.com>
15394 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
15395 * expr.c (convert_move): Use them.
15396 (convert_modes): Likewise.
15397 (store_expr_with_bounds): Likewise.
15399 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15400 Alan Hayward <alan.hayward@arm.com>
15401 David Sherwood <david.sherwood@arm.com>
15403 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
15404 parameter for the mode of "x". Remove the "known_x", "known_mode"
15405 and "known_ret" arguments. Change the type of the mode argument
15406 to scalar_int_mode.
15407 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
15408 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
15409 (reg_num_sign_bit_copies_for_combine): Likewise.
15410 * rtlanal.c (nonzero_bits1): Likewise.
15411 (num_sign_bit_copies1): Likewise.
15412 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
15413 (reg_num_sign_bit_copies_general): Likewise.
15414 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
15415 (reg_nonzero_bits_general): Likewise.
15417 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15418 Alan Hayward <alan.hayward@arm.com>
15419 David Sherwood <david.sherwood@arm.com>
15421 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
15422 than in subroutines. Return 1 for non-integer modes.
15423 (cached_num_sign_bit_copies): Change the type of the mode parameter
15424 to scalar_int_mode.
15425 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
15426 classes. Handle CONST_INT_P first and then check whether X also
15427 has a scalar integer mode. Check the same thing for inner registers
15428 of a SUBREG and for values that are being extended or truncated.
15430 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15431 Alan Hayward <alan.hayward@arm.com>
15432 David Sherwood <david.sherwood@arm.com>
15434 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
15435 in subroutines. Return the mode mask for non-integer modes.
15436 (cached_nonzero_bits): Change the type of the mode parameter
15437 to scalar_int_mode.
15438 (nonzero_bits1): Likewise. Remove early exit for other mode
15439 classes. Handle CONST_INT_P first and then check whether X
15440 also has a scalar integer mode.
15442 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15443 Alan Hayward <alan.hayward@arm.com>
15444 David Sherwood <david.sherwood@arm.com>
15446 * expr.c (widest_int_mode_for_size): Make the comment match the code.
15447 Return a scalar_int_mode and assert that the size is greater than
15449 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
15450 (op_by_pieces_d::op_by_pieces_d): Likewise.
15451 (op_by_pieces_d::run): Likewise.
15452 (can_store_by_pieces): Likewise.
15454 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15455 Alan Hayward <alan.hayward@arm.com>
15456 David Sherwood <david.sherwood@arm.com>
15458 * combine.c (extract_left_shift): Add a mode argument and update
15460 (make_compound_operation_int): Change the type of the mode parameter
15461 to scalar_int_mode and update the call to extract_left_shift.
15463 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15464 Alan Hayward <alan.hayward@arm.com>
15465 David Sherwood <david.sherwood@arm.com>
15467 * combine.c (simplify_and_const_int): Change the type of the mode
15468 parameter to scalar_int_mode.
15469 (simplify_and_const_int_1): Likewise. Update recursive call.
15471 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15472 Alan Hayward <alan.hayward@arm.com>
15473 David Sherwood <david.sherwood@arm.com>
15475 * combine.c (simplify_compare_const): Check that the mode is a
15476 scalar_int_mode (rather than VOIDmode) before testing its
15478 (simplify_comparison): Move COMPARISON_P handling out of the
15479 loop and restrict the latter part of the loop to scalar_int_modes.
15480 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
15481 and when considering SUBREG_REGs. Use is_int_mode instead of
15482 checking GET_MODE_CLASS against MODE_INT.
15484 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15485 Alan Hayward <alan.hayward@arm.com>
15486 David Sherwood <david.sherwood@arm.com>
15488 * combine.c (try_widen_shift_mode): Move check for equal modes to...
15489 (simplify_shift_const_1): ...here. Use scalar_int_mode for
15490 shift_unit_mode and for modes involved in scalar shifts.
15492 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15493 Alan Hayward <alan.hayward@arm.com>
15494 David Sherwood <david.sherwood@arm.com>
15496 * combine.c (force_int_to_mode): New function, split out from...
15497 (force_to_mode): ...here. Keep xmode up-to-date and use it
15498 instead of GET_MODE (x).
15500 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15501 Alan Hayward <alan.hayward@arm.com>
15502 David Sherwood <david.sherwood@arm.com>
15504 * optabs-query.h (extraction_insn::struct_mode): Change type to
15505 opt_scalar_int_mode and update comment.
15506 (extraction_insn::field_mode): Change type to scalar_int_mode.
15507 (extraction_insn::pos_mode): Likewise.
15508 * combine.c (make_extraction): Update accordingly.
15509 * optabs-query.c (get_traditional_extraction_insn): Likewise.
15510 (get_optab_extraction_insn): Likewise.
15511 * recog.c (simplify_while_replacing): Likewise.
15512 * expmed.c (narrow_bit_field_mem): Change the type of the mode
15513 parameter to opt_scalar_int_mode.
15515 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15516 Alan Hayward <alan.hayward@arm.com>
15517 David Sherwood <david.sherwood@arm.com>
15519 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
15520 to a scalar_int_mode instead of a machine_mode.
15521 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
15522 (get_best_mode): Return a boolean and use a pointer argument to store
15523 the selected mode. Replace the limit mode parameter with a bit limit.
15524 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
15525 for the values returned by bit_field_mode_iterator::next_mode.
15526 (store_bit_field): Update call to get_best_mode.
15527 (store_fixed_bit_field): Likewise.
15528 (extract_fixed_bit_field): Likewise.
15529 * expr.c (optimize_bitfield_assignment_op): Likewise.
15530 * fold-const.c (optimize_bit_field_compare): Likewise.
15531 (fold_truth_andor_1): Likewise.
15532 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
15533 Update for new type of m_mode.
15534 (get_best_mode): As above.
15536 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15537 Alan Hayward <alan.hayward@arm.com>
15538 David Sherwood <david.sherwood@arm.com>
15540 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
15541 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
15542 (store_bit_field): Check is_a <scalar_int_mode> before calling
15543 strict_volatile_bitfield_p.
15544 (extract_bit_field): Likewise.
15546 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15547 Alan Hayward <alan.hayward@arm.com>
15548 David Sherwood <david.sherwood@arm.com>
15550 * target.def (cstore_mode): Return a scalar_int_mode.
15551 * doc/tm.texi: Regenerate.
15552 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
15553 * targhooks.h (default_cstore_mode): Likewise.
15554 * targhooks.c (default_cstore_mode): Likewise, using a forced
15556 * expmed.c (emit_cstore): Expect the target of the cstore to be
15559 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15560 Alan Hayward <alan.hayward@arm.com>
15561 David Sherwood <david.sherwood@arm.com>
15563 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
15565 (niter_desc): Likewise mode.
15566 (iv_analyze): Add a mode parameter.
15568 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
15569 and change its type to scalar_int_mode.
15570 * loop-iv.c: Update commentary at head of file.
15571 (iv_constant): Pass the mode paraeter before the rtx it describes
15572 and change its type to scalar_int_mode. Remove VOIDmode handling.
15573 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
15574 (iv_extend): Likewise.
15575 (shorten_into_mode): Likewise.
15576 (iv_add): Use scalar_int_mode.
15577 (iv_mult): Likewise.
15578 (iv_shift): Likewise.
15579 (canonicalize_iv_subregs): Likewise.
15580 (get_biv_step_1): Pass the outer_mode parameter before the rtx
15581 it describes and change its mode to scalar_int_mode. Also change
15582 the type of the returned inner_mode to scalar_int_mode.
15583 (get_biv_step): Likewise, turning outer_mode from a pointer
15584 into a direct parameter. Update call to get_biv_step_1.
15585 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
15586 iv_constant and get_biv_step.
15587 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
15588 and change its type to scalar_int_mode. Don't initialise iv->mode
15589 to VOIDmode and remove later checks for its still being VOIDmode.
15590 Update calls to iv_analyze_op and iv_analyze_expr. Check
15591 is_a <scalar_int_mode> when changing the mode under consideration.
15592 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
15593 Update call to iv_analyze_expr.
15594 (iv_analyze_op): Add a mode parameter. Reject subregs whose
15595 inner register is not also a scalar_int_mode. Update call to
15597 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
15598 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
15599 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
15600 separate mode class checks. Update calls to iv_analyze. Remove
15601 fix-up of VOIDmodes after iv_analyze_biv.
15602 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
15603 don't have a scalar_int_mode. Update call to biv_p.
15605 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15606 Alan Hayward <alan.hayward@arm.com>
15607 David Sherwood <david.sherwood@arm.com>
15609 * cfgexpand.c (convert_debug_memory_address): Use
15610 as_a <scalar_int_mode>.
15611 * combine.c (expand_compound_operation): Likewise.
15612 (make_extraction): Likewise.
15613 (change_zero_ext): Likewise.
15614 (simplify_comparison): Likewise.
15615 * cse.c (cse_insn): Likewise.
15616 * dwarf2out.c (minmax_loc_descriptor): Likewise.
15617 (mem_loc_descriptor): Likewise.
15618 (loc_descriptor): Likewise.
15619 * expmed.c (init_expmed_one_mode): Likewise.
15620 (synth_mult): Likewise.
15621 (emit_store_flag_1): Likewise.
15622 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
15623 of a comparison with size.
15624 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
15625 (reduce_to_bit_field_precision): Likewise.
15626 * function.c (expand_function_end): Likewise.
15627 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
15628 * loop-doloop.c (doloop_modify): Likewise.
15629 * optabs.c (expand_binop): Likewise.
15630 (expand_unop): Likewise.
15631 (expand_copysign_absneg): Likewise.
15632 (prepare_cmp_insn): Likewise.
15633 (maybe_legitimize_operand): Likewise.
15634 * recog.c (const_scalar_int_operand): Likewise.
15635 * rtlanal.c (get_address_mode): Likewise.
15636 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15637 (simplify_cond_clz_ctz): Likewise.
15638 * tree-nested.c (get_nl_goto_field): Likewise.
15639 * tree.c (build_vector_type_for_mode): Likewise.
15640 * var-tracking.c (use_narrower_mode): Likewise.
15642 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15643 Alan Hayward <alan.hayward@arm.com>
15644 David Sherwood <david.sherwood@arm.com>
15646 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
15647 * builtins.c (expand_builtin_signbit): Use it.
15648 * cfgexpand.c (expand_debug_expr): Likewise.
15649 * dojump.c (do_jump): Likewise.
15650 (do_compare_and_jump): Likewise.
15651 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
15652 * expmed.c (make_tree): Likewise.
15653 * expr.c (expand_expr_real_2): Likewise.
15654 (expand_expr_real_1): Likewise.
15655 (try_casesi): Likewise.
15656 * fold-const-call.c (fold_const_call_ss): Likewise.
15657 * fold-const.c (unextend): Likewise.
15658 (extract_muldiv_1): Likewise.
15659 (fold_single_bit_test): Likewise.
15660 (native_encode_int): Likewise.
15661 (native_encode_string): Likewise.
15662 (native_interpret_int): Likewise.
15663 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
15664 * internal-fn.c (expand_addsub_overflow): Likewise.
15665 (expand_neg_overflow): Likewise.
15666 (expand_mul_overflow): Likewise.
15667 (expand_arith_overflow): Likewise.
15668 * match.pd: Likewise.
15669 * stor-layout.c (layout_type): Likewise.
15670 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15671 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
15672 * tree-ssanames.c (get_range_info): Likewise.
15673 * tree-switch-conversion.c (array_value_type) Likewise.
15674 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
15675 (vect_recog_divmod_pattern): Likewise.
15676 (vect_recog_mixed_size_cond_pattern): Likewise.
15677 * tree-vrp.c (extract_range_basic): Likewise.
15678 (simplify_float_conversion_using_ranges): Likewise.
15679 * tree.c (int_fits_type_p): Likewise.
15680 * ubsan.c (instrument_bool_enum_load): Likewise.
15681 * varasm.c (mergeable_string_section): Likewise.
15682 (narrowing_initializer_constant_valid_p): Likewise.
15683 (output_constant): Likewise.
15685 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15686 Alan Hayward <alan.hayward@arm.com>
15687 David Sherwood <david.sherwood@arm.com>
15689 * machmode.h (NARROWEST_INT_MODE): New macro.
15690 * expr.c (alignment_for_piecewise_move): Use it instead of
15691 GET_CLASS_NARROWEST_MODE (MODE_INT).
15692 (push_block): Likewise.
15693 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
15695 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15697 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15698 Alan Hayward <alan.hayward@arm.com>
15699 David Sherwood <david.sherwood@arm.com>
15701 * postreload.c (move2add_valid_value_p): Change the type of the
15702 mode parameter to scalar_int_mode.
15703 (move2add_use_add2_insn): Add a mode parameter and use it instead
15705 (move2add_use_add3_insn): Likewise.
15706 (reload_cse_move2add): Update accordingly.
15708 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15709 Alan Hayward <alan.hayward@arm.com>
15710 David Sherwood <david.sherwood@arm.com>
15712 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
15714 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
15715 * optabs.c (expand_unop): Likewise.
15717 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15718 Alan Hayward <alan.hayward@arm.com>
15719 David Sherwood <david.sherwood@arm.com>
15721 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
15722 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
15723 (popcount_loc_descriptor): Likewise.
15724 (bswap_loc_descriptor): Likewise.
15725 (rotate_loc_descriptor): Likewise.
15726 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
15727 calling the functions above.
15729 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15730 Alan Hayward <alan.hayward@arm.com>
15731 David Sherwood <david.sherwood@arm.com>
15733 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
15735 (try_combine): Likewise.
15736 (simplify_if_then_else): Likewise.
15737 * cse.c (cse_insn): Likewise.
15738 * dwarf2out.c (mem_loc_descriptor): Likewise.
15739 * emit-rtl.c (gen_lowpart_common): Likewise.
15740 * simplify-rtx.c (simplify_truncation): Likewise.
15741 (simplify_binary_operation_1): Likewise.
15742 (simplify_const_relational_operation): Likewise.
15743 (simplify_ternary_operation): Likewise.
15744 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
15746 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15747 Alan Hayward <alan.hayward@arm.com>
15748 David Sherwood <david.sherwood@arm.com>
15750 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
15751 * reload.c (push_reload): Likewise.
15752 (find_reloads): Likewise.
15754 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15755 Alan Hayward <alan.hayward@arm.com>
15756 David Sherwood <david.sherwood@arm.com>
15758 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
15759 (make_compound_operation_int): Likewise.
15760 (change_zero_ext): Likewise.
15761 * expr.c (convert_move): Likewise.
15762 (convert_modes): Likewise.
15763 * fwprop.c (forward_propagate_subreg): Likewise.
15764 * loop-iv.c (get_biv_step_1): Likewise.
15765 * optabs.c (widen_operand): Likewise.
15766 * postreload.c (move2add_valid_value_p): Likewise.
15767 * recog.c (simplify_while_replacing): Likewise.
15768 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15769 (simplify_binary_operation_1): Likewise. Remove redundant
15770 mode equality check.
15772 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15773 Alan Hayward <alan.hayward@arm.com>
15774 David Sherwood <david.sherwood@arm.com>
15776 * combine.c (combine_simplify_rtx): Add checks for
15777 is_a <scalar_int_mode>.
15778 (simplify_if_then_else): Likewise.
15779 (make_field_assignment): Likewise.
15780 (simplify_comparison): Likewise.
15781 * ifcvt.c (noce_try_bitop): Likewise.
15782 * loop-invariant.c (canonicalize_address_mult): Likewise.
15783 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15785 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15786 Alan Hayward <alan.hayward@arm.com>
15787 David Sherwood <david.sherwood@arm.com>
15789 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
15790 is_a <scalar_int_mode> instead of != BLKmode.
15792 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15793 Alan Hayward <alan.hayward@arm.com>
15794 David Sherwood <david.sherwood@arm.com>
15796 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
15797 instead of != VOIDmode.
15798 * combine.c (if_then_else_cond): Likewise.
15799 (change_zero_ext): Likewise.
15800 * dwarf2out.c (mem_loc_descriptor): Likewise.
15801 (loc_descriptor): Likewise.
15802 * rtlanal.c (canonicalize_condition): Likewise.
15803 * simplify-rtx.c (simplify_relational_operation_1): 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 * simplify-rtx.c (simplify_binary_operation_1): Use
15810 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
15812 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15813 Alan Hayward <alan.hayward@arm.com>
15814 David Sherwood <david.sherwood@arm.com>
15816 * wide-int.h (int_traits<unsigned char>) New class.
15817 (int_traits<unsigned short>) Likewise.
15818 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
15819 Use GET_MODE_UNIT_PRECISION and remove redundant test for
15821 * combine.c (set_nonzero_bits_and_sign_copies): Use
15822 is_a <scalar_int_mode>.
15823 (find_split_point): Likewise.
15824 (combine_simplify_rtx): Likewise.
15825 (simplify_logical): Likewise.
15826 (expand_compound_operation): Likewise.
15827 (expand_field_assignment): Likewise.
15828 (make_compound_operation): Likewise.
15829 (extended_count): Likewise.
15830 (change_zero_ext): Likewise.
15831 (simplify_comparison): Likewise.
15832 * dwarf2out.c (scompare_loc_descriptor): Likewise.
15833 (ucompare_loc_descriptor): Likewise.
15834 (minmax_loc_descriptor): Likewise.
15835 (mem_loc_descriptor): Likewise.
15836 (loc_descriptor): Likewise.
15837 * expmed.c (init_expmed_one_mode): Likewise.
15838 * lra-constraints.c (lra_constraint_offset): Likewise.
15839 * optabs.c (prepare_libcall_arg): Likewise.
15840 * postreload.c (move2add_note_store): Likewise.
15841 * reload.c (operands_match_p): Likewise.
15842 * rtl.h (load_extend_op): Likewise.
15843 * rtlhooks.c (gen_lowpart_general): Likewise.
15844 * simplify-rtx.c (simplify_truncation): Likewise.
15845 (simplify_unary_operation_1): Likewise.
15846 (simplify_binary_operation_1): Likewise.
15847 (simplify_const_binary_operation): Likewise.
15848 (simplify_const_relational_operation): Likewise.
15849 (simplify_subreg): Likewise.
15850 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15851 * var-tracking.c (adjust_mems): Likewise.
15852 (prepare_call_arguments): Likewise.
15854 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15855 Alan Hayward <alan.hayward@arm.com>
15856 David Sherwood <david.sherwood@arm.com>
15858 * machmode.h (is_int_mode): New fuction.
15859 * combine.c (find_split_point): Use it.
15860 (combine_simplify_rtx): Likewise.
15861 (simplify_if_then_else): Likewise.
15862 (simplify_set): Likewise.
15863 (simplify_shift_const_1): Likewise.
15864 (simplify_comparison): Likewise.
15865 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
15866 * cse.c (notreg_cost): Likewise.
15867 (cse_insn): Likewise.
15868 * cselib.c (cselib_lookup_1): Likewise.
15869 * dojump.c (do_jump_1): Likewise.
15870 (do_compare_rtx_and_jump): Likewise.
15871 * dse.c (get_call_args): Likewise.
15872 * dwarf2out.c (rtl_for_decl_init): Likewise.
15873 (native_encode_initializer): Likewise.
15874 * expmed.c (emit_store_flag_1): Likewise.
15875 (emit_store_flag): Likewise.
15876 * expr.c (convert_modes): Likewise.
15877 (store_field): Likewise.
15878 (expand_expr_real_1): Likewise.
15879 * fold-const.c (fold_read_from_constant_string): Likewise.
15880 * gimple-ssa-sprintf.c (get_format_string): Likewise.
15881 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
15882 * optabs.c (expand_binop): Likewise.
15883 (expand_unop): Likewise.
15884 (expand_abs_nojump): Likewise.
15885 (expand_one_cmpl_abs_nojump): Likewise.
15886 * simplify-rtx.c (mode_signbit_p): Likewise.
15887 (val_signbit_p): Likewise.
15888 (val_signbit_known_set_p): Likewise.
15889 (val_signbit_known_clear_p): Likewise.
15890 (simplify_relational_operation_1): Likewise.
15891 * tree.c (vector_type_mode): Likewise.
15893 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15894 Alan Hayward <alan.hayward@arm.com>
15895 David Sherwood <david.sherwood@arm.com>
15897 * machmode.h (smallest_mode_for_size): Fix formatting.
15898 (smallest_int_mode_for_size): New function.
15899 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
15900 instead of smallest_mode_for_size.
15901 * combine.c (make_extraction): Likewise.
15902 * config/arc/arc.c (arc_expand_movmem): Likewise.
15903 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
15904 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15905 * config/s390/s390.c (s390_expand_insv): Likewise.
15906 * config/sparc/sparc.c (assign_int_registers): Likewise.
15907 * config/spu/spu.c (spu_function_value): Likewise.
15908 (spu_function_arg): Likewise.
15909 * coverage.c (get_gcov_type): Likewise.
15910 (get_gcov_unsigned_t): Likewise.
15911 * dse.c (find_shift_sequence): Likewise.
15912 * expmed.c (store_bit_field_1): Likewise.
15913 * expr.c (convert_move): Likewise.
15914 (store_field): Likewise.
15915 * internal-fn.c (expand_arith_overflow): Likewise.
15916 * optabs-query.c (get_best_extraction_insn): Likewise.
15917 * optabs.c (expand_twoval_binop_libfunc): Likewise.
15918 * stor-layout.c (layout_type): Likewise.
15919 (initialize_sizetypes): Likewise.
15920 * targhooks.c (default_get_mask_mode): Likewise.
15921 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
15923 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15924 Alan Hayward <alan.hayward@arm.com>
15925 David Sherwood <david.sherwood@arm.com>
15927 * machmode.h (opt_mode::else_blk): New function.
15928 (int_mode_for_mode): Declare.
15929 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
15930 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
15932 * cfgexpand.c (expand_debug_expr): Likewise.
15933 * combine.c (gen_lowpart_or_truncate): Likewise.
15934 (gen_lowpart_for_combine): Likewise.
15935 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
15936 * config/avr/avr.c (avr_to_int_mode): Likewise.
15937 (avr_out_plus_1): Likewise.
15938 (avr_out_plus): Likewise.
15939 (avr_out_round): Likewise.
15940 * config/i386/i386.c (ix86_split_to_parts): Likewise.
15941 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15942 (s390_expand_vcond): Likewise.
15943 * config/spu/spu.c (spu_split_immediate): Likewise.
15944 (spu_expand_mov): Likewise.
15945 * dse.c (get_stored_val): Likewise.
15946 * expmed.c (store_bit_field_1): Likewise.
15947 (convert_extracted_bit_field): Use int_mode_for_mode instead of
15949 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
15950 (extract_low_bits): Likewise.
15951 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
15952 handling rather than repeating the check.
15953 (emit_group_store): Likewise.
15954 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
15955 * optabs.c (expand_absneg_bit): Likewise.
15956 (expand_copysign_absneg): Likewise.
15957 (expand_copysign_bit): Likewise.
15958 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
15959 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
15960 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
15961 * var-tracking.c (prepare_call_arguments): Likewise.
15962 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
15963 int_mode_for_mode instead of mode_for_size.
15964 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15966 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15967 Alan Hayward <alan.hayward@arm.com>
15968 David Sherwood <david.sherwood@arm.com>
15970 * machmode.h (int_mode_for_size): New function.
15971 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
15972 instead of mode_for_size.
15973 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
15975 * combine.c (expand_field_assignment): Use int_mode_for_size
15976 instead of mode_for_size.
15977 (make_extraction): Likewise.
15978 (simplify_shift_const_1): Likewise.
15979 (simplify_comparison): Likewise.
15980 * dojump.c (do_jump): Likewise.
15981 * dwarf2out.c (mem_loc_descriptor): Likewise.
15982 * emit-rtl.c (init_derived_machine_modes): Likewise.
15983 * expmed.c (flip_storage_order): Likewise.
15984 (convert_extracted_bit_field): Likewise.
15985 * expr.c (copy_blkmode_from_reg): Likewise.
15986 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
15987 * internal-fn.c (expand_mul_overflow): Likewise.
15988 * lower-subreg.c (simple_move): Likewise.
15989 * optabs-libfuncs.c (init_optabs): Likewise.
15990 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15991 * tree.c (vector_type_mode): Likewise.
15992 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
15993 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
15994 * tree-vect-generic.c (expand_vector_parallel): Likewise.
15995 * tree-vect-stmts.c (vectorizable_load): Likewise.
15996 (vectorizable_store): Likewise.
15998 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15999 Alan Hayward <alan.hayward@arm.com>
16000 David Sherwood <david.sherwood@arm.com>
16002 * coretypes.h (pod_mode): New type.
16003 (scalar_int_mode_pod): New typedef.
16004 * machmode.h (pod_mode): New class.
16005 (int_n_data_t::m): Change type to scalar_int_mode_pod.
16006 * genmodes.c (emit_mode_int_n): Update accordingly.
16007 * lower-subreg.h (target_lower_subreg): Change type to
16008 scalar_int_mode_pod.
16009 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
16010 scalar_int_mode_pod.
16012 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16013 Alan Hayward <alan.hayward@arm.com>
16014 David Sherwood <david.sherwood@arm.com>
16016 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
16017 machine_mode to scalar_int_mode.
16018 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
16019 (rs6000_option_override_internal): Remove cast to int.
16020 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
16021 machine_mode to scalar_int_mode.
16022 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
16023 (rs6000_option_override_internal): Remove cast to int.
16024 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
16025 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
16027 * config/s390/s390.c (s390_expand_builtin): Likewise.
16028 * coretypes.h (scalar_int_mode): New type.
16029 (opt_scalar_int_mode): New typedef.
16030 * machmode.h (scalar_int_mode): New class.
16031 (scalar_int_mode::includes_p): New function.
16032 (byte_mode): Change type to scalar_int_mode.
16033 (word_mode): Likewise.
16034 (ptr_mode): Likewise.
16035 * emit-rtl.c (byte_mode): Likewise.
16036 (word_mode): Likewise.
16037 (ptr_mode): Likewise.
16038 (init_derived_machine_modes): Update accordingly.
16039 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
16040 and MODE_PARTIAL_INT.
16041 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
16042 opt_scalar_int_mode.
16044 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16045 Alan Hayward <alan.hayward@arm.com>
16046 David Sherwood <david.sherwood@arm.com>
16048 * target.def (libgcc_floating_mode_supported_p): Take a
16050 * doc/tm.texi: Regenerate.
16051 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
16053 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16054 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
16057 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16058 Alan Hayward <alan.hayward@arm.com>
16059 David Sherwood <david.sherwood@arm.com>
16061 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
16062 * doc/tm.texi: Regenerate.
16063 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
16064 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
16065 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
16066 * targhooks.h (default_floatn_mode): Likewise.
16067 * targhooks.c (default_floatn_mode): Likewise.
16068 * tree.c (build_common_tree_nodes): Update accordingly.
16070 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16071 Alan Hayward <alan.hayward@arm.com>
16072 David Sherwood <david.sherwood@arm.com>
16074 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
16075 (mode_iterator::iterate_p): Likewise.
16076 (mode_iterator::get_wider): Likewise.
16077 * expr.c (init_expr_target): Use opt_scalar_float_mode.
16079 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16080 Alan Hayward <alan.hayward@arm.com>
16081 David Sherwood <david.sherwood@arm.com>
16083 * coretypes.h (opt_scalar_float_mode): New typedef.
16084 * machmode.h (float_mode_for_size): New function.
16085 * emit-rtl.c (double_mode): Delete.
16086 (init_emit_once): Use float_mode_for_size.
16087 * stor-layout.c (layout_type): Likewise.
16088 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
16090 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16091 Alan Hayward <alan.hayward@arm.com>
16092 David Sherwood <david.sherwood@arm.com>
16094 * output.h (assemble_real): Take a scalar_float_mode.
16095 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
16096 * config/arm/arm.md (consttable_4): Likewise.
16097 (consttable_8): Likewise.
16098 (consttable_16): Likewise.
16099 * config/mips/mips.md (consttable_float): Likewise.
16100 * config/s390/s390.c (s390_output_pool_entry): Likewise.
16101 * varasm.c (assemble_real): Take a scalar_float_mode.
16102 (output_constant_pool_2): Update accordingly.
16103 (output_constant): Likewise.
16105 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16106 Alan Hayward <alan.hayward@arm.com>
16107 David Sherwood <david.sherwood@arm.com>
16109 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
16110 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
16111 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
16112 (native_encode_real): Likewise.
16113 (native_interpret_real): Likewise.
16114 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
16115 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16117 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16118 Alan Hayward <alan.hayward@arm.com>
16119 David Sherwood <david.sherwood@arm.com>
16121 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
16122 <scalar_float_mode>. Simplify.
16123 (gen_extend_conv_libfunc): Likewise.
16125 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16126 Alan Hayward <alan.hayward@arm.com>
16127 David Sherwood <david.sherwood@arm.com>
16129 * coretypes.h (scalar_float_mode): New type.
16130 * machmode.h (mode_traits::from_int): Use machine_mode if
16131 USE_ENUM_MODES is defined.
16132 (is_a): New function.
16134 (dyn_cast): Likewise.
16135 (scalar_float_mode): New class.
16136 (scalar_float_mode::includes_p): New function.
16137 (is_float_mode): Likewise.
16138 * gdbhooks.py (MachineModePrinter): New class.
16139 (build_pretty_printer): Use it for scalar_float_mode.
16140 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
16141 (format_helper::format_helper): Turn into a template.
16142 * genmodes.c (get_mode_class): New function.
16143 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
16144 or machine_mode if none.
16145 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
16146 as_a <scalar_float_mode>.
16147 * dwarf2out.c (mem_loc_descriptor): Likewise.
16148 (insert_float): Likewise.
16149 (add_const_value_attribute): Likewise.
16150 * simplify-rtx.c (simplify_immed_subreg): Likewise.
16151 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
16152 (expand_unop): Update accordingly.
16153 (expand_abs_nojump): Likewise.
16154 (expand_copysign_absneg): Take a scalar_float_mode.
16155 (expand_copysign_bit): Likewise.
16156 (expand_copysign): Update accordingly.
16158 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16159 Alan Hayward <alan.hayward@arm.com>
16160 David Sherwood <david.sherwood@arm.com>
16162 * coretypes.h (opt_mode): New class.
16163 * machmode.h (opt_mode): Likewise.
16164 (opt_mode::else_void): New function.
16165 (opt_mode::require): Likewise.
16166 (opt_mode::exists): Likewise.
16167 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
16168 (GET_MODE_2XWIDER_MODE): Likewise.
16169 (mode_iterator::get_wider): Update accordingly.
16170 (mode_iterator::get_2xwider): Likewise.
16171 (mode_iterator::get_known_wider): Likewise, turning into a template.
16172 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
16173 forcing a wider mode to exist.
16174 * config/cr16/cr16.h (LONG_REG_P): Likewise.
16175 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16176 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
16177 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
16178 * lower-subreg.c (init_lower_subreg): Likewise.
16179 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
16180 on the final iteration.
16181 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
16182 a wider mode exists before asking for a move pattern.
16183 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
16184 forcing a wider mode to exist.
16185 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
16186 returning false if no such mode exists.
16187 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
16188 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
16189 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
16190 Avoid checking for a MODE_INT if we already know the mode is not a
16192 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
16193 forcing a wider mode to exist.
16194 (expmed_mult_highpart_optab): Likewise.
16195 (expmed_mult_highpart): Likewise.
16196 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
16198 * lto-streamer-in.c (lto_input_mode_table): Likewise.
16199 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
16200 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16201 * internal-fn.c (expand_mul_overflow): Update use of
16202 GET_MODE_2XWIDER_MODE.
16203 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16204 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
16205 GET_MODE_WIDER_MODE.
16206 (convert_plusminus_to_widen): Likewise.
16207 * tree-switch-conversion.c (array_value_type): Likewise.
16208 * var-tracking.c (emit_note_insn_var_location): Likewise.
16209 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16210 Return false inside rather than outside the loop if no wider mode
16212 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
16213 and GET_MODE_2XWIDER_MODE
16214 (can_compare_p): Use else_void.
16215 * gdbhooks.py (OptMachineModePrinter): New class.
16216 (build_pretty_printer): Use it for opt_mode.
16218 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16219 Alan Hayward <alan.hayward@arm.com>
16220 David Sherwood <david.sherwood@arm.com>
16222 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
16223 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
16225 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16226 Alan Hayward <alan.hayward@arm.com>
16227 David Sherwood <david.sherwood@arm.com>
16229 * machmode.h (mode_traits): New structure.
16230 (get_narrowest_mode): New function.
16231 (mode_iterator::start): Likewise.
16232 (mode_iterator::iterate_p): Likewise.
16233 (mode_iterator::get_wider): Likewise.
16234 (mode_iterator::get_known_wider): Likewise.
16235 (mode_iterator::get_2xwider): Likewise.
16236 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
16237 (FOR_EACH_MODE): Likewise.
16238 (FOR_EACH_MODE_FROM): Likewise.
16239 (FOR_EACH_MODE_UNTIL): Likewise.
16240 (FOR_EACH_WIDER_MODE): Likewise.
16241 (FOR_EACH_2XWIDER_MODE): Likewise.
16242 * builtins.c (expand_builtin_strlen): Use new mode iterators.
16243 * combine.c (simplify_comparison): Likewise
16244 * config/i386/i386.c (type_natural_mode): Likewise.
16245 * cse.c (cse_insn): Likewise.
16246 * dse.c (find_shift_sequence): Likewise.
16247 * emit-rtl.c (init_derived_machine_modes): Likewise.
16248 (init_emit_once): Likewise.
16249 * explow.c (hard_function_value): Likewise.
16250 * expmed.c (extract_fixed_bit_field_1): Likewise.
16251 (extract_bit_field_1): Likewise.
16252 (expand_divmod): Likewise.
16253 (emit_store_flag_1): Likewise.
16254 * expr.c (init_expr_target): Likewise.
16255 (convert_move): Likewise.
16256 (alignment_for_piecewise_move): Likewise.
16257 (widest_int_mode_for_size): Likewise.
16258 (emit_block_move_via_movmem): Likewise.
16259 (copy_blkmode_to_reg): Likewise.
16260 (set_storage_via_setmem): Likewise.
16261 (compress_float_constant): Likewise.
16262 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16263 * optabs-query.c (get_best_extraction_insn): Likewise.
16264 * optabs.c (expand_binop): Likewise.
16265 (expand_twoval_unop): Likewise.
16266 (expand_twoval_binop): Likewise.
16267 (widen_leading): Likewise.
16268 (widen_bswap): Likewise.
16269 (expand_parity): Likewise.
16270 (expand_unop): Likewise.
16271 (prepare_cmp_insn): Likewise.
16272 (prepare_float_lib_cmp): Likewise.
16273 (expand_float): Likewise.
16274 (expand_fix): Likewise.
16275 (expand_sfix_optab): Likewise.
16276 * postreload.c (move2add_use_add2_insn): Likewise.
16277 * reg-stack.c (reg_to_stack): Likewise.
16278 * reginfo.c (choose_hard_reg_mode): Likewise.
16279 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16280 * stor-layout.c (mode_for_size): Likewise.
16281 (smallest_mode_for_size): Likewise.
16282 (mode_for_vector): Likewise.
16283 (finish_bitfield_representative): Likewise.
16284 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
16285 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
16286 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16287 * var-tracking.c (prepare_call_arguments): Likewise.
16289 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16290 Alan Hayward <alan.hayward@arm.com>
16291 David Sherwood <david.sherwood@arm.com>
16293 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
16294 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
16295 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
16296 * machmode.h (mode_size): Move earlier in file.
16297 (mode_precision): Likewise.
16298 (mode_inner): Likewise.
16299 (mode_nunits): Likewise.
16300 (mode_unit_size): Likewise.
16301 (unit_unit_precision): Likewise.
16302 (mode_wider): Likewise.
16303 (mode_2xwider): Likewise.
16304 (machine_mode): New class.
16305 (mode_to_bytes): New function.
16306 (mode_to_bits): Likewise.
16307 (mode_to_precision): Likewise.
16308 (mode_to_inner): Likewise.
16309 (mode_to_unit_size): Likewise.
16310 (mode_to_unit_precision): Likewise.
16311 (mode_to_nunits): Likewise.
16312 (GET_MODE_SIZE): Use mode_to_bytes.
16313 (GET_MODE_BITSIZE): Use mode_to_bits.
16314 (GET_MODE_PRECISION): Use mode_to_precision.
16315 (GET_MODE_INNER): Use mode_to_inner.
16316 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
16317 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
16318 (GET_MODE_NUNITS): Use mode_to_nunits.
16319 * system.h (ALWAYS_INLINE): New macro.
16320 * config/powerpcspe/powerpcspe-c.c
16321 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
16322 int for arg1_mode and arg2_mode.
16324 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16325 Alan Hayward <alan.hayward@arm.com>
16326 David Sherwood <david.sherwood@arm.com>
16328 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
16329 Prefix mode names with E_ in case statements.
16330 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16331 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
16332 (aarch64_split_simd_move): Likewise.
16333 (aarch64_gen_storewb_pair): Likewise.
16334 (aarch64_gen_loadwb_pair): Likewise.
16335 (aarch64_gen_store_pair): Likewise.
16336 (aarch64_gen_load_pair): Likewise.
16337 (aarch64_get_condition_code_1): Likewise.
16338 (aarch64_constant_pool_reload_icode): Likewise.
16339 (get_rsqrte_type): Likewise.
16340 (get_rsqrts_type): Likewise.
16341 (get_recpe_type): Likewise.
16342 (get_recps_type): Likewise.
16343 (aarch64_gimplify_va_arg_expr): Likewise.
16344 (aarch64_simd_container_mode): Likewise.
16345 (aarch64_emit_load_exclusive): Likewise.
16346 (aarch64_emit_store_exclusive): Likewise.
16347 (aarch64_expand_compare_and_swap): Likewise.
16348 (aarch64_gen_atomic_cas): Likewise.
16349 (aarch64_emit_bic): Likewise.
16350 (aarch64_emit_atomic_swap): Likewise.
16351 (aarch64_emit_atomic_load_op): Likewise.
16352 (aarch64_evpc_trn): Likewise.
16353 (aarch64_evpc_uzp): Likewise.
16354 (aarch64_evpc_zip): Likewise.
16355 (aarch64_evpc_ext): Likewise.
16356 (aarch64_evpc_rev): Likewise.
16357 (aarch64_evpc_dup): Likewise.
16358 (aarch64_gen_ccmp_first): Likewise.
16359 (aarch64_gen_ccmp_next): Likewise.
16360 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
16361 (alpha_emit_xfloating_libcall): Likewise.
16362 (emit_insxl): Likewise.
16363 (alpha_arg_type): Likewise.
16364 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
16365 (arc_preferred_simd_mode): Likewise.
16366 (arc_secondary_reload): Likewise.
16367 (get_arc_condition_code): Likewise.
16368 (arc_print_operand): Likewise.
16369 (arc_legitimate_constant_p): Likewise.
16370 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16371 * config/arc/arc.md (casesi_load): Likewise.
16372 (casesi_compact_jump): Likewise.
16373 * config/arc/predicates.md (proper_comparison_operator): Likewise.
16374 (cc_use_register): Likewise.
16375 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16376 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
16377 (arm_init_iwmmxt_builtins): Likewise.
16378 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
16379 (neon_expand_vector_init): Likewise.
16380 (arm_attr_length_move_neon): Likewise.
16381 (maybe_get_arm_condition_code): Likewise.
16382 (arm_emit_vector_const): Likewise.
16383 (arm_preferred_simd_mode): Likewise.
16384 (arm_output_iwmmxt_tinsr): Likewise.
16385 (thumb1_output_casesi): Likewise.
16386 (thumb2_output_casesi): Likewise.
16387 (arm_emit_load_exclusive): Likewise.
16388 (arm_emit_store_exclusive): Likewise.
16389 (arm_expand_compare_and_swap): Likewise.
16390 (arm_evpc_neon_vuzp): Likewise.
16391 (arm_evpc_neon_vzip): Likewise.
16392 (arm_evpc_neon_vrev): Likewise.
16393 (arm_evpc_neon_vtrn): Likewise.
16394 (arm_evpc_neon_vext): Likewise.
16395 (arm_validize_comparison): Likewise.
16396 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
16397 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
16398 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
16399 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
16400 (c6x_preferred_simd_mode): Likewise.
16401 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
16402 (epiphany_rtx_costs): Likewise.
16403 * config/epiphany/predicates.md (proper_comparison_operator):
16405 * config/frv/frv.c (condexec_memory_operand): Likewise.
16406 (frv_emit_move): Likewise.
16407 (output_move_single): Likewise.
16408 (output_condmove_single): Likewise.
16409 (frv_hard_regno_mode_ok): Likewise.
16410 (frv_matching_accg_mode): Likewise.
16411 * config/h8300/h8300.c (split_adds_subs): Likewise.
16412 (h8300_rtx_costs): Likewise.
16413 (h8300_print_operand): Likewise.
16414 (compute_mov_length): Likewise.
16415 (output_logical_op): Likewise.
16416 (compute_logical_op_length): Likewise.
16417 (compute_logical_op_cc): Likewise.
16418 (h8300_shift_needs_scratch_p): Likewise.
16419 (output_a_shift): Likewise.
16420 (compute_a_shift_length): Likewise.
16421 (compute_a_shift_cc): Likewise.
16422 (expand_a_rotate): Likewise.
16423 (output_a_rotate): Likewise.
16424 * config/i386/i386.c (classify_argument): Likewise.
16425 (function_arg_advance_32): Likewise.
16426 (function_arg_32): Likewise.
16427 (function_arg_64): Likewise.
16428 (function_value_64): Likewise.
16429 (ix86_gimplify_va_arg): Likewise.
16430 (ix86_legitimate_constant_p): Likewise.
16431 (put_condition_code): Likewise.
16432 (split_double_mode): Likewise.
16433 (ix86_avx256_split_vector_move_misalign): Likewise.
16434 (ix86_expand_vector_logical_operator): Likewise.
16435 (ix86_split_idivmod): Likewise.
16436 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
16437 (ix86_build_const_vector): Likewise.
16438 (ix86_build_signbit_mask): Likewise.
16439 (ix86_match_ccmode): Likewise.
16440 (ix86_cc_modes_compatible): Likewise.
16441 (ix86_expand_branch): Likewise.
16442 (ix86_expand_sse_cmp): Likewise.
16443 (ix86_expand_sse_movcc): Likewise.
16444 (ix86_expand_int_sse_cmp): Likewise.
16445 (ix86_expand_vec_perm_vpermi2): Likewise.
16446 (ix86_expand_vec_perm): Likewise.
16447 (ix86_expand_sse_unpack): Likewise.
16448 (ix86_expand_int_addcc): Likewise.
16449 (ix86_split_to_parts): Likewise.
16450 (ix86_vectorize_builtin_gather): Likewise.
16451 (ix86_vectorize_builtin_scatter): Likewise.
16452 (avx_vpermilp_parallel): Likewise.
16453 (inline_memory_move_cost): Likewise.
16454 (ix86_tieable_integer_mode_p): Likewise.
16455 (x86_maybe_negate_const_int): Likewise.
16456 (ix86_expand_vector_init_duplicate): Likewise.
16457 (ix86_expand_vector_init_one_nonzero): Likewise.
16458 (ix86_expand_vector_init_one_var): Likewise.
16459 (ix86_expand_vector_init_concat): Likewise.
16460 (ix86_expand_vector_init_interleave): Likewise.
16461 (ix86_expand_vector_init_general): Likewise.
16462 (ix86_expand_vector_set): Likewise.
16463 (ix86_expand_vector_extract): Likewise.
16464 (emit_reduc_half): Likewise.
16465 (ix86_emit_i387_round): Likewise.
16466 (ix86_mangle_type): Likewise.
16467 (ix86_expand_round_sse4): Likewise.
16468 (expand_vec_perm_blend): Likewise.
16469 (canonicalize_vector_int_perm): Likewise.
16470 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
16471 (expand_vec_perm_1): Likewise.
16472 (expand_vec_perm_interleave3): Likewise.
16473 (expand_vec_perm_even_odd_pack): Likewise.
16474 (expand_vec_perm_even_odd_1): Likewise.
16475 (expand_vec_perm_broadcast_1): Likewise.
16476 (ix86_vectorize_vec_perm_const_ok): Likewise.
16477 (ix86_expand_vecop_qihi): Likewise.
16478 (ix86_expand_mul_widen_hilo): Likewise.
16479 (ix86_expand_sse2_abs): Likewise.
16480 (ix86_expand_pextr): Likewise.
16481 (ix86_expand_pinsr): Likewise.
16482 (ix86_preferred_simd_mode): Likewise.
16483 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
16484 * config/i386/sse.md (*andnot<mode>3): Likewise.
16485 (<mask_codefor><code><mode>3<mask_name>): Likewise.
16486 (*<code><mode>3): Likewise.
16487 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
16488 (ia64_expand_atomic_op): Likewise.
16489 (ia64_arg_type): Likewise.
16490 (ia64_mode_to_int): Likewise.
16491 (ia64_scalar_mode_supported_p): Likewise.
16492 (ia64_vector_mode_supported_p): Likewise.
16493 (expand_vec_perm_broadcast): Likewise.
16494 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
16495 (iq2000_function_arg_advance): Likewise.
16496 (iq2000_function_arg): Likewise.
16497 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
16498 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
16499 (m68k_libcall_value): Likewise.
16500 (m68k_function_value): Likewise.
16501 (sched_attr_op_type): Likewise.
16502 * config/mcore/mcore.c (mcore_output_move): Likewise.
16503 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
16505 (microblaze_function_arg): Likewise.
16506 * config/mips/mips.c (mips16_build_call_stub): Likewise.
16507 (mips_print_operand): Likewise.
16508 (mips_mode_ok_for_mov_fmt_p): Likewise.
16509 (mips_vector_mode_supported_p): Likewise.
16510 (mips_preferred_simd_mode): Likewise.
16511 (mips_expand_vpc_loongson_even_odd): Likewise.
16512 (mips_expand_vec_unpack): Likewise.
16513 (mips_expand_vi_broadcast): Likewise.
16514 (mips_expand_vector_init): Likewise.
16515 (mips_expand_vec_reduc): Likewise.
16516 (mips_expand_msa_cmp): Likewise.
16517 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
16518 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
16519 (cc_flags_for_mode): Likewise.
16520 * config/msp430/msp430.c (msp430_print_operand): Likewise.
16521 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
16522 (nds32_output_casesi_pc_relative): Likewise.
16523 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16524 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
16525 (nvptx_gen_unpack): Likewise.
16526 (nvptx_gen_pack): Likewise.
16527 (nvptx_gen_shuffle): Likewise.
16528 (nvptx_gen_wcast): Likewise.
16529 (nvptx_preferred_simd_mode): Likewise.
16530 * config/pa/pa.c (pa_secondary_reload): Likewise.
16531 * config/pa/predicates.md (base14_operand): Likewise.
16532 * config/powerpcspe/powerpcspe-c.c
16533 (altivec_resolve_overloaded_builtin): Likewise.
16534 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
16536 (rs6000_preferred_simd_mode): Likewise.
16537 (output_vec_const_move): Likewise.
16538 (rs6000_expand_vector_extract): Likewise.
16539 (rs6000_split_vec_extract_var): Likewise.
16540 (reg_offset_addressing_ok_p): Likewise.
16541 (rs6000_legitimate_offset_address_p): Likewise.
16542 (rs6000_legitimize_address): Likewise.
16543 (rs6000_emit_set_const): Likewise.
16544 (rs6000_const_vec): Likewise.
16545 (rs6000_emit_move): Likewise.
16546 (spe_build_register_parallel): Likewise.
16547 (rs6000_darwin64_record_arg_recurse): Likewise.
16548 (swap_selector_for_mode): Likewise.
16549 (spe_init_builtins): Likewise.
16550 (paired_init_builtins): Likewise.
16551 (altivec_init_builtins): Likewise.
16552 (do_load_for_compare): Likewise.
16553 (rs6000_generate_compare): Likewise.
16554 (rs6000_expand_float128_convert): Likewise.
16555 (emit_load_locked): Likewise.
16556 (emit_store_conditional): Likewise.
16557 (rs6000_output_function_epilogue): Likewise.
16558 (rs6000_handle_altivec_attribute): Likewise.
16559 (rs6000_function_value): Likewise.
16560 (emit_fusion_gpr_load): Likewise.
16561 (emit_fusion_p9_load): Likewise.
16562 (emit_fusion_p9_store): Likewise.
16563 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
16564 (fusion_gpr_mem_load): Likewise.
16565 (fusion_addis_mem_combo_load): Likewise.
16566 (fusion_addis_mem_combo_store): Likewise.
16567 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
16568 (fusion_gpr_mem_load): Likewise.
16569 (fusion_addis_mem_combo_load): Likewise.
16570 (fusion_addis_mem_combo_store): Likewise.
16571 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16573 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
16574 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16575 (rs6000_preferred_simd_mode): Likewise.
16576 (output_vec_const_move): Likewise.
16577 (rs6000_expand_vector_extract): Likewise.
16578 (rs6000_split_vec_extract_var): Likewise.
16579 (reg_offset_addressing_ok_p): Likewise.
16580 (rs6000_legitimate_offset_address_p): Likewise.
16581 (rs6000_legitimize_address): Likewise.
16582 (rs6000_emit_set_const): Likewise.
16583 (rs6000_const_vec): Likewise.
16584 (rs6000_emit_move): Likewise.
16585 (rs6000_darwin64_record_arg_recurse): Likewise.
16586 (swap_selector_for_mode): Likewise.
16587 (paired_init_builtins): Likewise.
16588 (altivec_init_builtins): Likewise.
16589 (rs6000_expand_float128_convert): Likewise.
16590 (emit_load_locked): Likewise.
16591 (emit_store_conditional): Likewise.
16592 (rs6000_output_function_epilogue): Likewise.
16593 (rs6000_handle_altivec_attribute): Likewise.
16594 (rs6000_function_value): Likewise.
16595 (emit_fusion_gpr_load): Likewise.
16596 (emit_fusion_p9_load): Likewise.
16597 (emit_fusion_p9_store): Likewise.
16598 * config/rx/rx.c (rx_gen_move_template): Likewise.
16599 (flags_from_mode): Likewise.
16600 * config/s390/predicates.md (s390_alc_comparison): Likewise.
16601 (s390_slb_comparison): Likewise.
16602 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
16603 (s390_vector_mode_supported_p): Likewise.
16604 (s390_cc_modes_compatible): Likewise.
16605 (s390_match_ccmode_set): Likewise.
16606 (s390_canonicalize_comparison): Likewise.
16607 (s390_emit_compare_and_swap): Likewise.
16608 (s390_branch_condition_mask): Likewise.
16609 (s390_rtx_costs): Likewise.
16610 (s390_secondary_reload): Likewise.
16611 (__SECONDARY_RELOAD_CASE): Likewise.
16612 (s390_expand_cs): Likewise.
16613 (s390_preferred_simd_mode): Likewise.
16614 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
16615 * config/sh/sh.c (sh_print_operand): Likewise.
16616 (dump_table): Likewise.
16617 (sh_secondary_reload): Likewise.
16618 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16619 * config/sh/sh.md (casesi_worker_1): Likewise.
16620 (casesi_worker_2): Likewise.
16621 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
16622 (fcc_comparison_operator): Likewise.
16623 * config/sparc/sparc.c (sparc_expand_move): Likewise.
16624 (emit_soft_tfmode_cvt): Likewise.
16625 (sparc_preferred_simd_mode): Likewise.
16626 (output_cbranch): Likewise.
16627 (sparc_print_operand): Likewise.
16628 (sparc_expand_vec_perm_bmask): Likewise.
16629 (vector_init_bshuffle): Likewise.
16630 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
16631 (spu_vector_mode_supported_p): Likewise.
16632 (spu_expand_insv): Likewise.
16633 (spu_emit_branch_or_set): Likewise.
16634 (spu_handle_vector_attribute): Likewise.
16635 (spu_builtin_splats): Likewise.
16636 (spu_builtin_extract): Likewise.
16637 (spu_builtin_promote): Likewise.
16638 (spu_expand_sign_extend): Likewise.
16639 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
16640 (tilegx_simd_int): Likewise.
16641 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
16642 (tilepro_simd_int): Likewise.
16643 * config/v850/v850.c (const_double_split): Likewise.
16644 (v850_print_operand): Likewise.
16645 (ep_memory_offset): Likewise.
16646 * config/vax/vax.c (vax_rtx_costs): Likewise.
16647 (vax_output_int_move): Likewise.
16648 (vax_output_int_add): Likewise.
16649 (vax_output_int_subtract): Likewise.
16650 * config/visium/predicates.md (visium_branch_operator): Likewise.
16651 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
16652 (visium_print_operand_address): Likewise.
16653 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16654 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
16655 (xtensa_expand_conditional_branch): Likewise.
16656 (xtensa_copy_incoming_a7): Likewise.
16657 (xtensa_output_literal): Likewise.
16658 * dfp.c (decimal_real_maxval): Likewise.
16659 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16661 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16662 Alan Hayward <alan.hayward@arm.com>
16663 David Sherwood <david.sherwood@arm.com>
16665 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
16666 (mode_nunits_inline): Likewise.
16667 (mode_inner_inline): Likewise.
16668 (mode_unit_size_inline): Likewise.
16669 (mode_unit_precision_inline): Likewise.
16670 (emit_insn_modes_h): Likewise. Also emit a #define of the
16672 (emit_mode_wider): Add an E_ prefix to mode names.
16673 (emit_mode_complex): Likewise.
16674 (emit_mode_inner): Likewise.
16675 (emit_mode_adjustments): Likewise.
16676 (emit_mode_int_n): Likewise.
16677 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
16678 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
16679 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
16680 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
16681 (CRC32_BUILTIN, ENTRY): Likewise.
16682 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
16683 (aarch64_pop_regs): Likewise.
16684 (aarch64_process_components): Likewise.
16685 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
16686 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
16687 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
16688 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
16689 * config/arm/arm.c (arm_init_libfuncs): Likewise.
16690 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
16692 * config/i386/i386-builtin.def (pcmpestr): Likewise.
16693 (pcmpistr): Likewise.
16694 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
16695 * config/mmix/mmix.c (mmix_output_condition): Likewise.
16696 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
16698 * config/rl78/rl78.c (mduc_regs): Likewise.
16699 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16700 (htm_expand_builtin): Likewise.
16701 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
16702 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
16703 * config/xtensa/xtensa.c (print_operand): Likewise.
16704 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
16705 (NUM_MODE_VECTOR_INT): Likewise.
16706 * genoutput.c (null_operand): Likewise.
16707 (output_operand_data): Likewise.
16708 * genrecog.c (print_parameter_value): Likewise.
16709 * lra.c (debug_operand_data): Likewise.
16711 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16713 * dwarf2out.c (scompare_loc_descriptor_wide)
16714 (scompare_loc_descriptor_narrow): New functions, split out from...
16715 (scompare_loc_descriptor): ...here.
16716 * expmed.c (emit_store_flag_int): New function, split out from...
16717 (emit_store_flag): ...here.
16719 2017-08-30 Richard Biener <rguenther@suse.de>
16721 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
16722 (dwarf2out_early_finish): Move setting of AT_pubnames from
16723 early debug output to early finish.
16725 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16727 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
16728 and -mdata-region to the assembler.
16730 2017-08-30 Richard Biener <rguenther@suse.de>
16732 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
16734 (gen_subprogram_die): Add DW_AT_object_pointer only early.
16735 (dwarf2out_early_global_decl): Only generate a DIE for the
16736 abstract origin if it doesn't already exist or is a declaration DIE.
16737 (resolve_addr): Do not add the linkage name twice when
16738 generating a stub DIE for the DW_TAG_GNU_call_site target.
16740 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16742 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16743 Use machine_mode rather than int for arg1_mode.
16745 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
16748 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
16749 that the second argument of the built-in functions to unpack
16750 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
16751 switch statement instead a lot of if statements.
16752 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
16753 Allow 64-bit values to be in Altivec registers as well as
16754 traditional floating point registers.
16755 (pack<mode>, FMOVE128_VSX iterator): Likewise.
16757 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
16759 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
16760 MAX_REGS_PER_ADDRESS == 1.
16762 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
16764 * config/i386/i386.opt (flag_fentry): Do not init to -1.
16765 * config/i386/i386.c (ix86_option_override_internal): Simplify
16766 setting of opts->x_flag_entry.
16768 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16769 Jakub Jelinek <jakub@redhat.com>
16770 Richard Biener <rguenther@suse.de>
16772 PR tree-optimization/81503
16773 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
16774 folded constant fits in the target type; reorder tests for clarity.
16776 2017-08-29 Martin Liska <mliska@suse.cz>
16778 * passes.def: Include pass_lower_switch.
16779 * stmt.c (dump_case_nodes): Remove and move to
16780 tree-switch-conversion.
16781 (case_values_threshold): Likewise.
16782 (expand_switch_as_decision_tree_p): Likewise.
16783 (emit_case_decision_tree): Likewise.
16784 (expand_case): Likewise.
16785 (balance_case_nodes): Likewise.
16786 (node_has_low_bound): Likewise.
16787 (node_has_high_bound): Likewise.
16788 (node_is_bounded): Likewise.
16789 (emit_case_nodes): Likewise.
16790 (struct simple_case_node): New struct.
16791 (add_case_node): Remove.
16792 (emit_case_dispatch_table): Use vector instead of case_list.
16793 (reset_out_edges_aux): Remove.
16794 (compute_cases_per_edge): Likewise.
16795 (expand_case): Build list of simple_case_node.
16796 (expand_sjlj_dispatch_table): Use it.
16797 * tree-switch-conversion.c (struct case_node): Moved from
16798 stmt.c and adjusted.
16799 (emit_case_nodes): Likewise.
16800 (node_has_low_bound): Likewise.
16801 (node_has_high_bound): Likewise.
16802 (node_is_bounded): Likewise.
16803 (case_values_threshold): Likewise.
16804 (reset_out_edges_aux): Likewise.
16805 (compute_cases_per_edge): Likewise.
16806 (add_case_node): Likewise.
16807 (dump_case_nodes): Likewise.
16808 (balance_case_nodes): Likewise.
16809 (expand_switch_as_decision_tree_p): Likewise.
16810 (emit_jump): Likewise.
16811 (emit_case_decision_tree): Likewise.
16812 (try_switch_expansion): Likewise.
16813 (do_jump_if_equal): Likewise.
16814 (emit_cmp_and_jump_insns): Likewise.
16815 (fix_phi_operands_for_edge): New function.
16816 (record_phi_operand_mapping): Likewise.
16817 (class pass_lower_switch): New pass.
16818 (pass_lower_switch::execute): New function.
16819 (make_pass_lower_switch): Likewise.
16820 (conditional_probability):
16821 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
16822 * tree-pass.h: Add make_pass_lower_switch.
16824 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16827 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
16830 2017-08-29 Richard Biener <rguenther@suse.de>
16832 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
16833 we do not add a DW_AT_inline attribute twice.
16834 (gen_subprogram_die): Remove code setting DW_AT_inline on
16835 DECL_ABSTRACT_P nodes.
16837 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
16839 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
16840 calls to internal functions.
16841 (gimplify_modify_expr): Likewise.
16842 * tree-call-cdce.c (use_internal_fn): Likewise.
16843 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
16844 (convert_to_divmod): Set the nothrow flag.
16845 * tree-if-conv.c (predicate_mem_writes): Likewise.
16846 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16847 (vectorizable_call): Likewise.
16848 (vectorizable_store): Likewise.
16849 (vectorizable_load): Likewise.
16850 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
16851 (vect_recog_mask_conversion_pattern): Likewise.
16853 2017-08-29 Martin Liska <mliska@suse.cz>
16856 * gcc.c (driver_handle_option): Add new argument.
16857 * opts-common.c (handle_option): Pass
16858 target_option_override_hook.
16859 * opts-global.c (lang_handle_option): Add new option.
16860 (set_default_handlers): Add new argument.
16861 (decode_options): Likewise.
16862 * opts.c (target_handle_option): Likewise.
16863 (common_handle_option): Call target_option_override_hook.
16864 * opts.h (struct cl_option_handler_func): Add hook for
16865 target option override.
16866 (struct cl_option_handlers): Likewise.
16867 (set_default_handlers): Add new argument.
16868 (decode_options): Likewise.
16869 (common_handle_option): Likewise.
16870 (target_handle_option): Likewise.
16871 * toplev.c (toplev::main): Pass targetm.target_option.override
16874 2017-08-29 Richard Biener <rguenther@suse.de>
16875 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
16877 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
16878 life to the active subtree.
16880 2017-08-28 Jeff Law <law@redhat.com>
16882 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
16883 derive_equivalences.
16884 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
16885 Code moved into....
16886 (edge_info::derive_equivalences): New private member function
16888 * tree-ssa-dom.c (class edge_info): Changed from a struct
16889 to a class. Add ctor/dtor, methods and data members.
16890 (edge_info::edge_info): Renamed from allocate_edge_info.
16891 Initialize additional members.
16892 (edge_info::~edge_info): New.
16893 (free_dom_edge_info): Delete the edge info.
16894 (record_edge_info): Use new class & associated member functions.
16895 Tighten forms for testing for edge equivalences.
16896 (record_temporary_equivalences): Iterate over the simple
16897 equivalences rather than assuming there's only one per edge.
16898 (cprop_into_successor_phis): Iterate over the simple
16899 equivalences rather than assuming there's only one per edge.
16900 (optimize_stmt): Use operand_equal_p rather than pointer
16901 equality for mini-DSE code.
16903 2017-08-28 Nathan Sidwell <nathan@acm.org>
16905 * gcc.c (execute): Fold SIGPIPE handling into switch
16906 statement. Adjust internal error message.
16908 2017-08-28 Richard Biener <rguenther@suse.de>
16911 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
16912 Do nothing for removed DIEs.
16914 2017-08-28 Richard Biener <rguenther@suse.de>
16916 PR tree-optimization/81977
16917 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
16920 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
16923 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
16924 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
16925 using targetm.gen_mem_thread_fence.
16927 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
16930 * config/i386/i386.md (*<btsc><mode>): Change operand 2
16931 predicate to register_operand. Reorder operands.
16932 (*btr<mode>): Ditto.
16933 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
16934 (*btr<mode>_mask): Ditto.
16936 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
16938 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
16939 * config/rs6000/xmmintrin.h: New file.
16940 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
16942 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16945 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
16946 parameter and_insn and return it.
16947 (recombine_lvx_pattern): Insert a copy to ensure availability of
16948 the base register of the copied masking operation at the point of
16949 the instruction replacement.
16950 (recombine_stvx_pattern): Likewise.
16952 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
16954 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
16955 undocumented switches.
16956 (-mpower9-dform-vector): Likewise.
16957 (-mpower9-dform): Likewise.
16958 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
16959 comments to delete references to -mpower9-dform* switches.
16960 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
16961 Delete reference to -mpower9-dform* switches, test for
16962 -mpower9-vector instead.
16963 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
16964 (OTHER_P9_VECTOR_MASKS): Likewise.
16965 (POWERPC_MASKS): Likewise.
16966 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
16967 tests against -mpower9-dform* to -mpower9-vector. Delete code
16968 that checked for -mpower9-dform* consistancy with other options.
16969 Add test for -mpower9-misc to enable other power9 switches.
16970 (rs6000_init_hard_regno_mode_ok): Likewise.
16971 (rs6000_option_override_internal): Likewise.
16972 (rs6000_emit_prologue): Likewise.
16973 (rs6000_emit_epilogue): Likewise.
16974 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
16975 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
16976 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
16978 (emit_fusion_p9_store): Likewise.
16979 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
16980 resetting these macros if the assembler does not support ISA 3.0
16982 (TARGET_P9_DFORM_VECTOR): Likewise.
16983 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
16984 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
16986 2017-08-25 Alan Modra <amodra@gmail.com>
16988 PR rtl-optimization/81747
16989 * cse.c (cse_extended_basic_block): Don't attempt to record
16990 equivalences for degenerate conditional branches that branch
16991 to their fall-through.
16993 2017-08-24 Martin Sebor <msebor@redhat.com>
16995 PR middle-end/81908
16996 * gimple-fold.c (size_must_be_zero_p): New function.
16997 (gimple_fold_builtin_memory_op): Call it.
16999 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
17001 * config/rs6000/mm_malloc.h: New file.
17003 2017-08-24 Bin Cheng <bin.cheng@arm.com>
17005 PR tree-optimization/81913
17006 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
17007 analysis when either IVs in condition can wrap.
17009 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
17011 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
17012 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
17014 2017-08-24 Richard Biener <rguenther@suse.de>
17017 * targhooks.c (default_target_can_inline_p): Properly
17018 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
17019 is present and always compare.
17020 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
17021 infer -mfpmath=sse from TARGET_SSE_P.
17022 (ix86_can_inline_p): Properly use target_option_default_node when
17023 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
17025 2017-08-24 Richard Biener <rguenther@suse.de>
17028 * dwarf2out.c (output_die): Handle flag_generate_offload like
17030 (output_comp_unit): Likewise.
17031 (gen_array_type_die): Likewise.
17032 (dwarf2out_early_finish): Likewise.
17033 (note_variable_value_in_expr): Likewise.
17034 (dwarf2out_finish): Likewise. Adjust assert.
17035 * cgraphunit.c (symbol_table::compile): Move setting of
17036 flag_generate_offload earlier ...
17037 (symbol_table::finalize_compilation_unit): ... here, before
17038 early debug finalization.
17040 2017-08-24 Richard Biener <rguenther@suse.de>
17042 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
17043 and ipa-fnsummary.h.
17044 (ix86_can_inline_p): When ix86_fpmath flags do not match
17045 check whether the callee uses FP math at all.
17047 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
17049 PR middle-end/81931
17050 * tree-ssanames.c (get_nonzero_bits): Use element_precision
17051 instead of TYPE_PRECISION.
17053 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
17054 Alan Hayward <alan.hayward@arm.com>
17055 David Sherwood <david.sherwood@arm.com>
17057 * combine.c (make_extraction): Use subreg_offset_from_lsb.
17059 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
17061 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
17063 (ix86_frame::stack_realign_allocate): New field.
17064 (struct machine_frame_state): Modify comments.
17065 (machine_frame_state::sp_realigned_fp_end): New field.
17066 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
17067 layout calculation.
17068 (sp_valid_at): Add assertion to assure no attempt to access invalid
17069 offset of a realigned stack.
17070 (fp_valid_at): Likewise.
17071 (choose_baseaddr): Modify comments.
17072 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
17073 ix86_expand_prologue.
17074 (ix86_expand_prologue): Modify stack realignment and allocation.
17075 (ix86_expand_epilogue): Modify comments.
17076 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
17077 avx2_runtime, avx512f, and avx512f_runtime.
17079 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
17081 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
17082 (mstackrealign): Do not init to -1.
17083 * config/i386/i386.c (ix86_option_override_internal):
17084 Check opts_set, not opts when setting default value of
17085 opts->x_ix86_force_align_arg_pointer.
17087 2017-08-23 Richard Biener <rguenther@suse.de>
17089 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
17090 lang_hooks.decl_printable_name.
17091 * print-rtl-function.c (print_rtx_function): Likewise.
17092 * tree-pretty-print.c (dump_function_header): Likewise.
17094 2017-08-23 Richard Biener <rguenther@suse.de>
17097 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
17098 -g0 at compile-time.
17100 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17102 PR middle-end/19706
17103 * doc/sourcebuild.texi (Other hardware attributes):
17106 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17108 PR middle-end/19706
17109 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
17110 Fix single-use check.
17112 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
17114 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
17116 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17118 * doc/install.texi: Modify to add more details on running selected
17121 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17123 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
17124 is combined with -mabi=ms.
17125 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
17128 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17130 PR tree-optimization/81488
17131 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
17132 and cached_basis fields.
17133 (MAX_SPREAD): New constant.
17134 (alloc_cand_and_find_basis): Initialize new fields.
17135 (clear_visited): New function.
17136 (create_phi_basis_1): Rename from create_phi_basis, set visited
17137 and cached_basis fields.
17138 (create_phi_basis): New wrapper function.
17139 (phi_add_costs_1): Rename from phi_add_costs, add spread
17140 parameter, set visited field, short-circuit when limits reached.
17141 (phi_add_costs): New wrapper function.
17142 (record_phi_increments_1): Rename from record_phi_increments, set
17144 (record_phi_increments): New wrapper function.
17145 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
17146 (phi_incr_cost): New wrapper function.
17147 (all_phi_incrs_profitable_1): Rename from
17148 all_phi_incrs_profitable, set visited field.
17149 (all_phi_incrs_profitable): New wrapper function.
17151 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17152 Alan Hayward <alan.hayward@arm.com>
17153 David Sherwood <david.sherwood@arm.com>
17155 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
17156 that takes the outer and inner modes.
17157 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
17158 comparison as the canonical test for a paradoxical subreg.
17159 * combine.c (simplify_set): Use paradoxical_subreg_p.
17160 (make_extraction): Likewise.
17161 (force_to_mode): Likewise.
17162 (rtx_equal_for_field_assignment_p): Likewise.
17163 (gen_lowpart_for_combine): Likewise.
17164 (simplify_comparison): Likewise.
17165 * cse.c (equiv_constant): Likewise.
17166 * expmed.c (store_bit_field_1): Likewise.
17167 * final.c (alter_subreg): Likewise.
17168 * fwprop.c (propagate_rtx): Likewise.
17169 (forward_propagate_subreg): Likewise.
17170 * ira-conflicts.c (ira_build_conflicts): Likewise.
17171 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
17172 * lra-constraints.c (curr_insn_transform): Likewise.
17173 (split_reg): Likewise.
17174 * lra-eliminations.c (move_plus_up): Likewise.
17175 (lra_eliminate_regs_1): Likewise.
17176 * recog.c (general_operand): Likewise.
17177 * ree.c (combine_reaching_defs): Likewise.
17178 * reload.c (push_reload): Likewise.
17179 (find_reloads): Likewise.
17180 * reload1.c (elimination_effects): Likewise.
17181 (compute_reload_subreg_offset): Likewise.
17182 (choose_reload_regs): Likewise.
17183 * rtlanal.c (subreg_lsb_1): Likewise.
17184 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17185 (simplify_subreg): Likewise.
17186 * var-tracking.c (track_loc_p): Likewise.
17187 * emit-rtl.c (byte_lowpart_offset): Likewise.
17188 (paradoxical_subreg_p): Delete out-of-line definition.
17190 2017-08-22 Jeff Law <law@redhat.com>
17192 PR tree-optimization/81741
17193 PR tree-optimization/71947
17194 * tree-ssa-dom.c: Include tree-inline.h.
17195 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
17196 equivalences if one is more expensive to compute than the other.
17197 * tree-ssa-scopedtables.h (class const_or_copies): Make
17198 record_const_or_copy_raw method private.
17199 (class avail_exprs_stack): New method simplify_binary_operation.
17200 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
17201 avail_exprs_stack::simplify_binary_operation as needed.
17202 (avail_exprs_stack::simplify_binary_operation): New function.
17204 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
17206 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
17207 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
17208 (DOT_SYMBOLS): Likewise.
17209 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
17210 (RELOCATABLE_NEEDS_FIXUP): Likewise.
17211 (RS6000_ABI_NAME): Likewise.
17212 (TARGET_CMODEL): Likewise.
17213 (TOC_SECTION_ASM_OP): Likewise.
17214 (SET_CMODEL): New macro.
17215 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
17217 2017-08-22 Richard Biener <rguenther@suse.de>
17219 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
17220 to allow for free-lang-data replacements similar to verify_type_variant.
17222 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17223 Alan Hayward <alan.hayward@arm.com>
17224 David Sherwood <david.sherwood@arm.com>
17226 * config/aarch64/aarch64.md (casesi): Use DImode rather than
17227 VOIDmode for the LABEL_REF.
17229 2017-08-22 Richard Biener <rguenther@suse.de>
17231 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
17233 2017-08-22 Richard Biener <rguenther@suse.de>
17235 * common.opt (feliminate-dwarf2-dups): Ignore.
17236 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
17237 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
17238 same_die_p_wrap, compute_section_prefix,
17239 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
17240 (comdat_symbol_id, comdat_symbol_number): Likewise.
17241 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
17243 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
17244 (output_die): Mark unreachable path unreachable.
17245 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
17246 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
17247 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
17248 (dwarf2out_early_finish): Likewise.
17250 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
17252 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
17254 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
17257 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
17258 not VAR_P. Filter attribute warnings with OPT_Wattributes.
17259 (avr_attribute_table) <io, io_low, address>: Initialize
17260 .decl_required with true.
17262 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
17264 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
17265 undocumented debugging options.
17266 (-mvsx-scalar-double): Likewise.
17267 (-mallow-df-permute): Likewise.
17268 (-mvectorize-builtins): Likewise.
17269 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17270 (rs6000_builtin_vectorized_function): Likewise.
17271 (rs6000_builtin_md_vectorized_function): Likewise.
17272 (rs6000_opt_vars): Likewise.
17274 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
17277 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
17278 (*btrq_imm): Rename from *btrq.
17279 (*btcq_imm): Rename from *btcq.
17280 (btsc): New code attribute.
17281 (*<btsc><mode>): New insn pattern.
17282 (*btr<mode>): Ditto.
17283 (*<btsc><mode>_mask): New insn_and_split pattern.
17284 (*btr<mode>_mask): Ditto.
17286 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17287 Alan Hayward <alan.hayward@arm.com>
17288 David Sherwood <david.sherwood@arm.com>
17290 * function.c (pad_below): Simplify padding calculation.
17292 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17293 Alan Hayward <alan.hayward@arm.com>
17294 David Sherwood <david.sherwood@arm.com>
17296 * target.def (function_prologue): Remove frame size argument.
17297 (function_epilogue): Likewise.
17298 * doc/tm.texi: Regenerate.
17299 * final.c (final_start_function): Update call to function_prologue.
17300 (final_end_function): Update call to function_epilogue.
17301 (default_function_pro_epilogue): Remove frame size argument.
17302 * output.h (default_function_pro_epilogue): Likewise.
17303 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
17304 (arm_output_function_prologue): Likewise.
17305 * config/frv/frv.c (frv_function_prologue): Likewise.
17306 (frv_function_epilogue): Likewise.
17307 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
17308 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
17309 (ia64_output_function_epilogue): Likewise.
17310 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
17311 (m32r_output_function_epilogue): Likewise.
17312 * config/microblaze/microblaze.c (microblaze_function_prologue)
17313 (microblaze_function_epilogue): Likewise.
17314 * config/mips/mips.c (mips_output_function_prologue): Likewise.
17315 (mips_output_function_epilogue): Likewise.
17316 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
17317 (mmix_target_asm_function_epilogue): Likewise.
17318 * config/msp430/msp430.c (msp430_start_function): Likewise.
17319 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
17320 (nds32_asm_function_epilogue): Likewise.
17321 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
17322 * config/pa/pa.c (pa_output_function_prologue): Likewise.
17323 (pa_output_function_epilogue): Likewise.
17324 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
17325 (rs6000_output_function_epilogue): Likewise.
17326 * config/rl78/rl78.c (rl78_start_function): Likewise.
17327 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
17328 (rs6000_output_function_epilogue): Likewise.
17329 * config/rx/rx.c (rx_output_function_prologue): Likewise.
17330 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
17331 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
17332 (sparc_asm_function_epilogue): Likewise.
17334 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17336 * tree.h (type_has_mode_precision_p): New function.
17337 * convert.c (convert_to_integer_1): Use it.
17338 * expr.c (expand_expr_real_2): Likewise.
17339 (expand_expr_real_1): Likewise.
17340 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
17341 * match.pd: Likewise.
17342 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
17343 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
17344 * tree-tailcall.c (process_assignment): Likewise.
17345 * tree-vect-loop.c (vectorizable_reduction): Likewise.
17346 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
17347 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
17348 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17349 (vectorizable_assignment): Likewise.
17350 (vectorizable_shift): Likewise.
17351 (vectorizable_operation): Likewise.
17352 * tree-vrp.c (register_edge_assert_for_2): Likewise.
17354 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
17356 * match.pd: Add pow (C, x) simplification.
17358 2017-08-21 Richard Biener <rguenther@suse.de>
17360 PR tree-optimization/81900
17361 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
17362 for blocks with abnormal predecessors.
17363 (compute_antic): Do not set visited flag prematurely.
17365 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
17368 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
17370 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17372 * stor-layout.h (vector_type_mode): Move to...
17373 * tree.h (vector_type_mode): ...here.
17374 * stor-layout.c (vector_type_mode): Move to...
17375 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
17377 2017-08-21 Richard Biener <rguenther@suse.de>
17379 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
17380 register_external_die hooks.
17381 (debug_false_tree_charstarstar_uhwistar): Declare.
17382 (debug_nothing_tree_charstar_uhwi): Likewise.
17383 * debug.c (do_nothing_debug_hooks): Adjust.
17384 (debug_false_tree_charstarstar_uhwistar): New do nothing.
17385 (debug_nothing_tree_charstar_uhwi): Likewise.
17386 * dbxout.c (dbx_debug_hooks): Adjust.
17387 (xcoff_debug_hooks): Likewise.
17388 * sdbout.c (sdb_debug_hooks): Likewise.
17389 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
17390 * dwarf2out.c (macinfo_label_base): New global.
17391 (dwarf2out_register_external_die): New function for the
17392 register_external_die hook.
17393 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
17394 (dwarf2_debug_hooks): Use them.
17395 (dwarf2_lineno_debug_hooks): Adjust.
17396 (struct die_struct): Add with_offset flag.
17397 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
17398 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
17399 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
17400 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
17401 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
17402 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
17403 defining section names for the early LTO debug variants.
17404 (reset_indirect_string): New helper.
17405 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
17406 (print_dw_val): Add support for offsetted symbol references.
17407 (get_ultimate_context): Split out from is_cxx.
17408 (is_cxx): Use get_ultimate_context.
17409 (is_fortran): Add decl overload.
17410 (compute_comp_unit_symbol): Split out worker from
17411 compute_section_prefix.
17412 (compute_section_prefix): Call compute_comp_unit_symbol and
17413 set comdat_type_p here.
17414 (output_die): Skip DIE symbol output for the LTO added one.
17415 Handle DIE symbol references with offset.
17416 (output_comp_unit): Guard section name mangling properly.
17417 For LTO debug sections emit a symbol at the section beginning
17418 which we use to refer to its DIEs.
17419 (add_abstract_origin_attribute): For DIEs registered via
17420 dwarf2out_register_external_die directly refer to the early
17421 DIE rather than indirectly through the shadow one we created.
17422 Remove obsolete call to dwarf2out_abstract_function for
17423 non-function/block origins.
17424 (gen_array_type_die): When generating early LTO debug do
17425 not emit DW_AT_string_length.
17426 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
17427 late when in LTO. As suggested place a gcc_unreachable for
17428 the DECL_ABSTRACT_P case.
17429 (gen_subprogram_die): Avoid another specification DIE
17430 for early built declarations/definitions for the late LTO case.
17431 (gen_variable_die): Add type references for late duplicated VLA dies
17433 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
17434 we have the abstract instance already.
17435 (process_scope_var): Adjust decl DIE contexts in LTO which
17436 first puts them in limbo.
17437 (gen_decl_die): Do not generate type DIEs late apart from
17438 types for VLAs or for decls we do not yet have a DIE. Do not
17439 call dwarf2out_abstract_function late.
17440 (dwarf2out_early_global_decl): Make sure to create DIEs
17441 for abstract instances of a decl first.
17442 (dwarf2out_late_global_decl): Adjust comment.
17443 (output_macinfo_op): With multiple macro sections use
17444 macinfo_label_base to distinguish labels.
17445 (output_macinfo): Likewise. Update macinfo_label_base.
17446 Pass in the line info label.
17447 (note_variable_value_in_expr): When generating LTO resolve
17448 all variable values here by generating DIEs as needed.
17449 (init_sections_and_labels): Add early LTO debug flag parameter
17450 and generate different sections and names if set. Add generation
17451 counter for the labels so we can have multiple of them.
17452 (reset_dies): Helper to allow DIEs to be output multiple times.
17453 (dwarf2out_finish): When outputting DIEs to the fat part of an
17454 LTO object first reset DIEs.
17455 (dwarf2out_early_finish): Output early DIEs when generating LTO.
17456 (modified_type_die): Check for decl_ultimate_origin being self
17458 (gen_type_die_with_usage): Likewise.
17459 (gen_typedef_die): Allow decl_ultimate_origin being self.
17460 (set_decl_abstract_flags): Remove.
17461 (set_block_abstract_flags): Likewise.
17462 (dwarf2out_abstract_function): Treat the early generated DIEs
17463 as the abstract copy and only add DW_AT_inline and
17464 DW_AT_artificial here and call set_decl_origin_self.
17465 If the DIE has an abstract origin don't do anything.
17466 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
17467 if we have none yet (Go fails to build one, PR78628).
17468 (variably_modified_type_p): Prevent endless recursion for Ada
17469 cyclic pointer types.
17470 * lto-streamer-in.c: Include debug.h.
17471 (dref_queue): New global.
17472 (lto_read_tree_1): Stream in DIE references.
17473 (lto_input_tree): Register DIE references.
17474 (input_function): Stream DECL_DEBUG_ARGS.
17475 * lto-streamer-out.c: Include debug.h.
17476 (lto_write_tree_1): Output DIE references.
17477 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
17478 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
17479 (output_function): Stream DECL_DEBUG_ARGS.
17480 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
17481 Stream DECL_ABSTRACT_ORIGIN.
17482 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
17483 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
17484 DECL_CONTEXT for file-scope decls.
17485 * lto-streamer.h (struct dref_entry): Declare.
17486 (dref_queue): Likewise.
17487 * cfgexpand.c (pass_expand::execute): Do not call the
17488 outlining_inline_function hook here.
17489 * lto-wrapper.c (debug_obj): New global.
17490 (tool_cleanup): Unlink it if required.
17491 (debug_objcopy): New function.
17492 (run_gcc): Handle early debug sections in the IL files by
17493 extracting them to separate files, partially linkin them and
17494 feeding the result back as result to the linker.
17495 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
17496 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
17497 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
17498 sections into a separate segment.
17499 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
17501 (darwin_asm_dwarf_section): Likewise.
17502 (darwin_asm_output_dwarf_offset): Likewise.
17503 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
17505 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17506 Alan Hayward <alan.hayward@arm.com>
17507 David Sherwood <david.sherwood@arm.com>
17509 * read-md.h (md_reader::record_potential_iterator_use): Replace
17510 pointer argument with an rtx and an index.
17511 * read-rtl.c (iterator_group::apply_iterator): Likewise.
17512 (apply_mode_iterator): Likewise.
17513 (apply_code_iterator): Likewise.
17514 (apply_int_iterator): Likewise.
17515 (apply_subst_iterator): Likewise.
17516 (record_iterator_use): Likewise.
17517 (record_attribute_use): Likewise.
17518 (md_reader::record_potential_iterator_use): Likewise. Update calls
17519 to record_iterator_use and apply_iterator.
17520 (iterator_use): Replace ptr with x and index.
17521 (attribute_use): Likewise.
17522 (apply_attribute_uses): Update calls to apply_iterator.
17523 (apply_iterators): Likewise. Update initialization of iterator_use.
17524 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
17525 and record_potential_iterator_use.
17526 (rtx_reader::read_rtx_operand): Likewise.
17528 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17529 Alan Hayward <alan.hayward@arm.com>
17530 David Sherwood <david.sherwood@arm.com>
17532 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
17535 2017-08-21 Richard Biener <rguenther@suse.de>
17537 PR middle-end/81884
17538 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
17539 at struct end conservatively when comparing common bases.
17541 2017-08-21 Richard Biener <rguenther@suse.de>
17543 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
17544 (mem_ref_in_stmt): Remove.
17545 (determine_max_movement): Use ref index to get at the reference.
17546 (invariantness_dom_walker::before_dom_children): Deal with
17547 lim data already initialized.
17548 (gather_mem_refs_stmt): Initialize lim data and record ref index.
17550 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
17552 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
17553 (TARGET_ISA_ROUND): Ditto.
17554 (TARGET_ROUND): Ditto.
17555 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
17556 * config/i386/i386.md: Ditto.
17557 * config/i386/sse.md: Ditto.
17558 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
17559 with OPTION_MASK_ISA_SSE4_1.
17561 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
17564 * doc/extend.texi (x86 Built-in Functions): Correct the name of
17565 __builtin_ia32_lzcnt_u16.
17567 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
17570 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
17571 (rs6000_set_current_function): Rewrite function to use it.
17573 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
17576 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
17577 and TYPE_WARN_IF_NOT_ALIGN.
17578 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
17579 (handle_warn_if_not_align): New.
17580 (place_union_field): Call handle_warn_if_not_align.
17581 (place_field): Call handle_warn_if_not_align.
17582 Copy TYPE_WARN_IF_NOT_ALIGN.
17583 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
17584 (layout_type): Likewise.
17585 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
17587 (tree_decl_common): Add warn_if_not_align.
17588 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
17589 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
17590 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
17591 (DECL_WARN_IF_NOT_ALIGN): Likewise.
17592 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
17593 * doc/extend.texi: Document warn_if_not_aligned attribute.
17594 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
17596 2017-08-17 Martin Liska <mliska@suse.cz>
17599 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
17600 (get_data_dependence): Use it as pointer type.
17601 (distribute_loop): Likewise.
17603 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17605 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
17606 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
17607 (p8_vmrgow_<mode>_direct): New define_insn.
17608 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
17609 handle endianness for vmrgew and vmrgow permute patterns.
17611 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
17613 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
17614 * config/rs6000/rs6000-cpus.def: Remove comment.
17615 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
17616 (POWERPC_MASKS): Likewise.
17617 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
17618 use of TARGET_VSX_TIMODE.
17619 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
17620 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
17621 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
17622 (rs6000_option_override_internal): Remove dead code.
17623 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
17624 (rs6000_legitimize_reload_address): Likewise.
17625 (rs6000_legitimate_address_p): Likewise.
17626 (rs6000_opt_masks): Delete "vsx-timode".
17627 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
17628 from function comment.
17629 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
17630 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
17631 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
17633 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
17634 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
17635 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
17636 (VSX_TI): Likewise.
17638 (define_peephole2): Likewise.
17640 2017-08-17 Martin Sebor <msebor@redhat.com>
17643 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
17644 past the end of an array.
17645 (test_pp_format): Add test cases.
17647 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
17649 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
17650 missing ECF_NOTHROW flags.
17652 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
17655 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
17656 operands residing in integer registers.
17657 (*vsx_le_perm_load_<mode>): Likewise.
17658 (*vsx_le_perm_store_<mode>): Likewise.
17659 (define_peephole2): Add peepholes to optimize the above.
17661 2017-08-17 Marek Polacek <polacek@redhat.com>
17663 PR middle-end/81814
17664 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
17665 to mimic what shorten_compare did. Change the return type to bool.
17666 (fold_cond_expr_with_comparison): Update call to
17667 operand_equal_for_comparison_p.
17668 (fold_ternary_loc): Likewise.
17670 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
17672 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
17674 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
17676 2017-08-17 Richard Biener <rguenther@suse.de>
17678 * tree-ssa-structalias.c (solve_graph): When propagating
17679 to successors update the graphs succ edges and avoid duplicate work.
17681 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
17684 * config/i386/i386.c (ix86_option_override_internal): Save target
17685 specific options after ix86_stack_protector_guard_reg was changed.
17687 2017-08-17 Richard Biener <rguenther@suse.de>
17689 PR tree-optimization/81827
17690 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
17691 (new_var_info): Initialize it conservatively.
17692 (get_call_vi): Mark register vars.
17693 (new_scalar_tmp_constraint_exp): Likewise.
17694 (handle_rhs_call): Likewise.
17695 (handle_const_call): Likewise.
17696 (create_function_info_for): Likewise.
17697 (solve_constraints): Sort varinfos to separate register from
17698 non-register vars to pack points-to solution bitmaps during
17701 2017-08-17 Marek Polacek <polacek@redhat.com>
17703 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
17705 2017-08-17 Richard Biener <rguenther@suse.de>
17707 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
17708 to true when overflow is undefined and we saturated the result.
17710 2017-08-17 Alan Modra <amodra@gmail.com>
17713 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
17714 Don't use store multiple if only one reg needs saving.
17715 (interesting_frame_related_regno): New function.
17716 (rs6000_frame_related): Don't emit frame info for regs that
17718 (rs6000_emit_epilogue): Likewise.
17720 2017-08-16 Nathan Sidwell <nathan@acm.org>
17722 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
17723 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
17724 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
17725 * tree.c (free_lang_data_in_type): Use else-if chain. Always
17726 clear TYPE_LANG_1. Remove obsolete member-function stripping.
17727 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
17728 (verify_type): Adjust for TYPE_BINFO move.
17729 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
17730 process TYPE_BINFO directly.
17731 (hash_tree): Likewise.
17732 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17734 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17737 2017-08-16 David Malcolm <dmalcolm@redhat.com>
17739 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
17741 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
17744 * config/i386/i386.md (*anddi_1_btr): Change predicates of
17745 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
17746 Add ix86_binary_operator_ok to insn constraint.
17747 (*iordi_1_bts): Ditto.
17748 (*xordi_1_btc): Ditto.
17749 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
17750 Update corresponding peephole2 pattern.
17754 2017-08-16 Bin Cheng <bin.cheng@arm.com>
17756 PR tree-optimization/81832
17757 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
17758 copy loop header which has IFN_LOOP_DIST_ALIAS call.
17760 2017-08-16 Marek Polacek <polacek@redhat.com>
17763 * fold-const.c (fold_indirect_ref_1): Restore original behavior
17764 regarding size_zero_node.
17766 2017-08-16 Martin Liska <mliska@suse.cz>
17769 * config.gcc: Respect previously set extra_objs in case
17772 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
17774 PR tree-optimization/81835
17775 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
17776 the phi SSA_NAME. Check that the condition in a COND_EXPR does
17777 not depend on the phi.
17779 2017-08-16 Alan Modra <amodra@gmail.com>
17781 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
17784 2017-08-16 Alan Modra <amodra@gmail.com>
17786 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
17787 (save_reg_p): ..into this. Update all callers.
17788 (first_reg_to_save): Simplify.
17790 2017-08-16 Alan Modra <amodra@gmail.com>
17792 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
17795 2017-08-15 Joseph Myers <joseph@codesourcery.com>
17799 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
17800 constant count if that count is less than 32.
17802 2017-08-15 Nathan Sidwell <nathan@acm.org>
17804 * gcc.c (execute): Emit friendlier message if inferior is killed
17805 by an external cause.
17807 2017-08-15 Richard Biener <rguenther@suse.de>
17809 PR tree-optimization/81790
17810 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
17811 CONSTRUCTORs from simplifying and VN.
17813 2017-08-14 Martin Sebor <msebor@redhat.com>
17815 * builtin-attrs.def: Add comments.
17817 2017-08-14 Martin Sebor <msebor@redhat.com>
17820 * doc/extend.texi (attribute nonstring): Document new attribute.
17822 2017-08-14 Martin Sebor <msebor@redhat.com>
17825 * tree-diagnostic.c (default_tree_printer): Handle %G.
17826 * gimple-pretty-print.h (percent_G_format): Declare new function.
17827 * gimple-pretty-print.c (percent_G_format): Define.
17828 * tree-pretty-print.c (percent_K_format): Add argument.
17830 2017-08-14 Martin Sebor <msebor@redhat.com>
17832 PR translation/79998
17833 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
17834 Remove a stray space.
17836 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
17839 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
17840 (*iordi_1_bts): Ditto.
17841 (*xordi_1_btc): Ditto.
17843 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17846 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
17847 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17849 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
17850 quoted strings, and make more translator-friendly.
17851 (darwin_rs6000_override_options): Likewise.
17852 (rs6000_option_override_internal): Likewise.
17853 (rs6000_return_in_memory): Fix overlong line.
17854 (init_cmulative_args): Use quoted strings, and make more
17855 translator-friendly.
17856 (rs6000_pass_by_reference): Fix overlong line.
17857 (def_builtin): Use quoted strings.
17858 (altivec_expand_predicate_builtin): Use quoted strings, and make
17859 more translator-friendly.
17860 (htm_expand_builtin): Use quoted strings.
17861 (cpu_expand_builtin): Use quoted strings, and make more
17862 translator-friendly.
17863 (altivec_expand_builtin): Likewise.
17864 (paired_expand_predicate_builtin): Likewise.
17865 (rs6000_invalid_builtin): Likewise.
17866 (builtin_function_type): Use quoted strings.
17867 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
17868 more translator-friendly.
17869 (rs6000_trampoline_init): Likewise.
17870 (rs6000_handle_altivec_attribute): Likewise.
17871 (rs6000_inner_target_options): Use quoted strings.
17872 (rs6000_disable_incompatible_switches): Likewise.
17873 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
17874 strings, and make more translator-friendly.
17875 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
17877 2017-08-14 Bin Cheng <bin.cheng@arm.com>
17879 PR tree-optimization/81799
17880 * tree-loop-distribution.c (version_loop_by_alias_check): Force
17881 cond_expr to simple gimple operand.
17883 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
17885 PR middle-end/46932
17886 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
17888 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
17892 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
17893 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
17894 TARGET_GASISR_PROLOGUES.
17895 * config/avr/avr.c (avr_option_override): Same.
17896 (avr_pass_pre_proep::execute): Same.
17898 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
17901 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
17902 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
17904 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
17906 * config/i386/i386.md (*load_tp_<mode>): Redefine as
17907 define_insn_and_split. Split to a memory load from 0 in
17908 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
17909 using PTR mode iterator.
17910 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
17911 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
17912 (*add_tp_<mode>): Redefine as define_insn_and_split.
17913 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
17914 address space. Merge with *add_tp_x32 using PTR mode iterator.
17915 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
17916 Split to an add with a memory load from 0 in
17917 DEFAULT_TLS_SEG_REG address space.
17919 2017-08-12 Andrew Pinski <apinski@cavium.com>
17921 * config/aarch64/aarch64-option-extensions.def (rdma):
17922 Fix feature string to what Linux prints out in /proc/cpuinfo.
17924 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
17927 * dwarf2out.c (modified_type_die): For C typedef types that have
17928 an ultimate origin, process the ultimate origin instead of the
17930 (gen_typedef_die): Assert that input DECLs have no ultimate
17932 (gen_type_die_with_usage): For typedef variants that have an
17933 ultimate origin, just call gen_decl_die on the original DECL.
17934 (process_scope_var): Avoid creating DIEs for local typedefs and
17935 concrete static variables.
17937 2017-08-12 Alan Modra <amodra@gmail.com>
17941 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
17942 match gnu-user.h startfile.
17943 (ENDFILE_LINUX_SPEC): Similarly.
17945 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
17948 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
17950 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
17952 2017-08-11 Tamar Christina <tamar.christina@arm.com>
17953 * config/aarch64/aarch64.md (mov<mode>): Change.
17954 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
17955 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
17956 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
17958 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
17960 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
17961 for storage order barriers.
17963 2017-08-11 Martin Liska <mliska@suse.cz>
17966 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
17967 variables of void type.
17969 2017-08-11 Martin Liska <mliska@suse.cz>
17971 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
17972 TARGET_SUPPORTS_ALIASES.
17973 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
17974 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
17975 (optimize_weakref): Likewise.
17976 * symtab.c (symtab_node::noninterposable_alias): Likewise.
17977 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
17978 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
17980 2017-08-11 Martin Liska <mliska@suse.cz>
17983 * config/i386/i386.c (make_resolver_func): Do complete
17984 refactoring of the function.
17986 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17989 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
17990 * config/i386/i386.c (ix86_stack_protect_guard): Use
17991 ix86_stack_protect_guard_symbol_str to generate varible declaration.
17992 * doc/invoke.texi (x86 Options): Document
17993 -mstack-protector-guard-symbol= option.
17995 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17997 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
17998 * config/i386/i386.c (ix86_split_stack_guard): New function.
17999 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
18000 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
18001 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
18002 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
18003 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
18004 (split_stack_space_check): Call ix86_split_stack_guard.
18006 2017-08-10 Martin Sebor <msebor@redhat.com>
18008 * print-tree.c (print_node): Print location using the established
18010 Replace spaces with colons.
18011 (debug_raw, debug): Ditto.
18013 2017-08-10 Martin Sebor <msebor@redhat.com>
18016 * pretty-print.c (pp_format): Correct the handling of %s precision.
18018 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
18021 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18023 (ix86_finalize_stack_frame_flags): This. Also clear
18024 frame_pointer_needed if -fno-omit-frame-pointer is used without
18026 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18027 with ix86_finalize_stack_frame_flags.
18028 (ix86_expand_epilogue): Likewise.
18029 (ix86_expand_split_stack_prologue): Likewise.
18030 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
18032 2017-08-10 Martin Liska <mliska@suse.cz>
18035 * c-attribs.c (handle_target_attribute):
18036 Report warning for an empty string argument of target attribute.
18038 2017-08-09 Jakub Jelinek <jakub@redhat.com>
18041 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
18043 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
18044 or DECL_NONLOCAL labels.
18045 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
18046 or DECL_NONLOCAL labels here.
18048 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
18050 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
18051 to indicate when early gimple folding has been disabled.
18052 (rs6000_gimple_fold_builtin): Add debug content.
18053 (rs6000_invalid_builtin): Fix whitespace.
18054 (rs6000_expand_builtin): Fix whitespace.
18055 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
18057 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
18060 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
18061 SAVE_MULTIPLE if not all the registers that saves, should be saved.
18063 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
18065 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
18066 (qdf24xx): Likewise.
18067 * config/aarch64/aarch64.md: Include falkor.md.
18068 * config/aarch64/falkor.md: New.
18070 2017-08-09 Marek Polacek <polacek@redhat.com>
18073 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
18074 * diagnostic.c (emit_diagnostic): Add a comment.
18075 (emit_diagnostic_valist): New function.
18077 2017-08-09 Marek Polacek <polacek@redhat.com>
18080 * input.c (make_location): New overload.
18081 * input.h (make_location): Declare.
18083 2017-08-08 Alan Modra <amodra@gmail.com>
18084 H.J. Lu <hongjiu.lu@intel.com>
18087 * gcc.c (NO_PIE_SPEC): Delete.
18088 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
18090 (LINK_PIE_SPEC): ..to here.
18091 (LINK_COMMAND_SPEC): Support -no-pie.
18092 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
18093 chain of crtbegin*.o selection, update for PIE_SPEC changes and
18095 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
18096 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
18097 (ENDFILE_CRTEND_SPEC): Similarly.
18099 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
18102 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
18103 (mstack-protector-guard-offset=): Ditto.
18104 * config/i386/i386.c (ix86_option_override): Handle
18105 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
18107 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
18108 ix86_stack_protect_guard_offset variables.
18109 (TARGET_STACK_PROTECT_GUARD): Always define.
18110 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
18111 and -mstack-protector-guard-offset= options.
18113 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18115 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
18116 boundary case for the last candidate.
18118 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18120 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
18121 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
18123 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18125 PR middle-end/19706
18126 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
18127 * config/aarch64/aarch64-builtins.c
18128 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
18129 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
18130 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
18132 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18133 Andrew Pinski <pinskia@gmail.com>
18135 PR middle-end/19706
18136 * internal-fn.def (XORSIGN): New.
18137 * optabs.def (xorsign_optab): New.
18138 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
18139 (convert_expand_mult_copysign): New.
18140 (pass_optimize_widening_mul::execute): Call
18141 convert_expand_mult_copysign.
18143 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18145 PR tree-optimization/81354
18146 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
18147 Insert on edges rather than explicitly creating landing pads.
18148 (analyze_candidates_and_replace): Commit edge inserts.
18150 2017-08-08 Richard Biener <rguenther@suse.de>
18152 PR middle-end/81719
18153 * tree-ssa-loop-niter.c: Include tree-dfa.h.
18154 (expand_simple_operations): Also look through ADDR_EXPRs with
18155 MEM_REF bases treating them as POINTER_PLUS_EXPR.
18157 2017-08-08 Richard Biener <rguenther@suse.de>
18159 PR tree-optimization/81723
18160 * tree-vect-slp.c (struct bst_traits): New hash traits.
18161 (bst_fail): New global.
18162 (vect_build_slp_tree_2): New worker, split out from ...
18163 (vect_build_slp_tree): ... this now wrapping it with using
18164 bst_fail set to cache SLP tree build fails. Properly handle
18166 (vect_analyze_slp_instance): Allocate and free bst_fail.
18168 2017-08-08 Martin Liska <mliska@suse.cz>
18171 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
18172 LABEL_DECLs that can be from a different function.
18174 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18176 PR tree-optimization/81744
18177 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
18178 loop's number of iterations.
18180 2017-08-08 Martin Liska <mliska@suse.cz>
18182 * asan.c: Include header files.
18183 * attribs.c (build_decl_attribute_variant): New function moved
18185 (build_type_attribute_qual_variant): Likewise.
18186 (cmp_attrib_identifiers): Likewise.
18187 (simple_cst_list_equal): Likewise.
18188 (omp_declare_simd_clauses_equal): Likewise.
18189 (attribute_value_equal): Likewise.
18190 (comp_type_attributes): Likewise.
18191 (build_type_attribute_variant): Likewise.
18192 (lookup_ident_attribute): Likewise.
18193 (remove_attribute): Likewise.
18194 (merge_attributes): Likewise.
18195 (merge_type_attributes): Likewise.
18196 (merge_decl_attributes): Likewise.
18197 (merge_dllimport_decl_attributes): Likewise.
18198 (handle_dll_attribute): Likewise.
18199 (attribute_list_equal): Likewise.
18200 (attribute_list_contained): Likewise.
18201 * attribs.h (lookup_attribute): New function moved from tree.[ch].
18202 (lookup_attribute_by_prefix): Likewise.
18203 * bb-reorder.c: Include header files.
18204 * builtins.c: Likewise.
18205 * calls.c: Likewise.
18206 * cfgexpand.c: Likewise.
18207 * cgraph.c: Likewise.
18208 * cgraphunit.c: Likewise.
18209 * convert.c: Likewise.
18210 * dwarf2out.c: Likewise.
18211 * final.c: Likewise.
18212 * fold-const.c: Likewise.
18213 * function.c: Likewise.
18214 * gimple-expr.c: Likewise.
18215 * gimple-fold.c: Likewise.
18216 * gimple-pretty-print.c: Likewise.
18217 * gimple.c: Likewise.
18218 * gimplify.c: Likewise.
18219 * hsa-common.c: Likewise.
18220 * hsa-gen.c: Likewise.
18221 * internal-fn.c: Likewise.
18222 * ipa-chkp.c: Likewise.
18223 * ipa-cp.c: Likewise.
18224 * ipa-devirt.c: Likewise.
18225 * ipa-fnsummary.c: Likewise.
18226 * ipa-inline.c: Likewise.
18227 * ipa-visibility.c: Likewise.
18229 * lto-cgraph.c: Likewise.
18230 * omp-expand.c: Likewise.
18231 * omp-general.c: Likewise.
18232 * omp-low.c: Likewise.
18233 * omp-offload.c: Likewise.
18234 * omp-simd-clone.c: Likewise.
18235 * opts-global.c: Likewise.
18236 * passes.c: Likewise.
18237 * predict.c: Likewise.
18238 * sancov.c: Likewise.
18239 * sanopt.c: Likewise.
18240 * symtab.c: Likewise.
18241 * toplev.c: Likewise.
18242 * trans-mem.c: Likewise.
18243 * tree-chkp.c: Likewise.
18244 * tree-eh.c: Likewise.
18245 * tree-into-ssa.c: Likewise.
18246 * tree-object-size.c: Likewise.
18247 * tree-parloops.c: Likewise.
18248 * tree-profile.c: Likewise.
18249 * tree-ssa-ccp.c: Likewise.
18250 * tree-ssa-live.c: Likewise.
18251 * tree-ssa-loop.c: Likewise.
18252 * tree-ssa-sccvn.c: Likewise.
18253 * tree-ssa-structalias.c: Likewise.
18254 * tree-ssa.c: Likewise.
18255 * tree-streamer-in.c: Likewise.
18256 * tree-vectorizer.c: Likewise.
18257 * tree-vrp.c: Likewise.
18258 * tsan.c: Likewise.
18259 * ubsan.c: Likewise.
18260 * varasm.c: Likewise.
18261 * varpool.c: Likewise.
18262 * tree.c: Remove functions moved to attribs.[ch].
18263 * tree.h: Likewise.
18264 * config/aarch64/aarch64.c: Add attrs.h header file.
18265 * config/alpha/alpha.c: Likewise.
18266 * config/arc/arc.c: Likewise.
18267 * config/arm/arm.c: Likewise.
18268 * config/avr/avr.c: Likewise.
18269 * config/bfin/bfin.c: Likewise.
18270 * config/c6x/c6x.c: Likewise.
18271 * config/cr16/cr16.c: Likewise.
18272 * config/cris/cris.c: Likewise.
18273 * config/darwin.c: Likewise.
18274 * config/epiphany/epiphany.c: Likewise.
18275 * config/fr30/fr30.c: Likewise.
18276 * config/frv/frv.c: Likewise.
18277 * config/ft32/ft32.c: Likewise.
18278 * config/h8300/h8300.c: Likewise.
18279 * config/i386/winnt.c: Likewise.
18280 * config/ia64/ia64.c: Likewise.
18281 * config/iq2000/iq2000.c: Likewise.
18282 * config/lm32/lm32.c: Likewise.
18283 * config/m32c/m32c.c: Likewise.
18284 * config/m32r/m32r.c: Likewise.
18285 * config/m68k/m68k.c: Likewise.
18286 * config/mcore/mcore.c: Likewise.
18287 * config/microblaze/microblaze.c: Likewise.
18288 * config/mips/mips.c: Likewise.
18289 * config/mmix/mmix.c: Likewise.
18290 * config/mn10300/mn10300.c: Likewise.
18291 * config/moxie/moxie.c: Likewise.
18292 * config/msp430/msp430.c: Likewise.
18293 * config/nds32/nds32-isr.c: Likewise.
18294 * config/nds32/nds32.c: Likewise.
18295 * config/nios2/nios2.c: Likewise.
18296 * config/nvptx/nvptx.c: Likewise.
18297 * config/pa/pa.c: Likewise.
18298 * config/pdp11/pdp11.c: Likewise.
18299 * config/powerpcspe/powerpcspe.c: Likewise.
18300 * config/riscv/riscv.c: Likewise.
18301 * config/rl78/rl78.c: Likewise.
18302 * config/rx/rx.c: Likewise.
18303 * config/s390/s390.c: Likewise.
18304 * config/sh/sh.c: Likewise.
18305 * config/sol2.c: Likewise.
18306 * config/sparc/sparc.c: Likewise.
18307 * config/spu/spu.c: Likewise.
18308 * config/stormy16/stormy16.c: Likewise.
18309 * config/tilegx/tilegx.c: Likewise.
18310 * config/tilepro/tilepro.c: Likewise.
18311 * config/v850/v850.c: Likewise.
18312 * config/vax/vax.c: Likewise.
18313 * config/visium/visium.c: Likewise.
18314 * config/xtensa/xtensa.c: Likewise.
18316 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18319 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
18320 constraints since the -mupper-regs-* switches have been
18322 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
18323 into a vector from a double word element that was extracted from
18324 another vector, and eliminate extra XXPERMDI instructions.
18325 (vsx_concat_<mode>_2): Likewise.
18326 (vsx_concat_<mode>_3): Likewise.
18327 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
18328 concat to allow optimizing inserts from previous extracts.
18330 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
18332 * config/i386/i386.c (ix86_stack_protect_guard): Generate
18333 memory reference to a SSP offset in TLS address space.
18334 (ix86_print_operand) <case '@'>: Remove.
18335 (ix86_print_operand_punct_valid_p): Remove '@' code.
18336 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
18337 UNSPEC_SP_TLS_TEST.
18338 (stack_tls_protect_set_<mode>): Remove.
18339 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
18340 (stack_tls_protect_test_<mode>): Remove.
18341 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
18343 2017-08-07 Olivier Hainque <hainque@adacore.com>
18346 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
18348 2017-08-07 Douglas Rupp <rupp@adacore.com>
18350 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
18351 variable was referenced as multidir in command.
18353 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18356 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
18359 2017-08-07 Martin Liska <mliska@suse.cz>
18361 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
18362 * config/rl78/rl78.c: Add include of attribs.h.
18363 * config/sh/sh.c: Likewise.
18364 * config/v850/v850.c: Likewise.
18366 2017-08-07 Tom de Vries <tom@codesourcery.com>
18368 PR middle-end/78266
18369 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
18371 2017-08-07 Martin Liska <mliska@suse.cz>
18373 * config/mips/mips.c: Include attribs.h.
18375 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
18378 * doc/invoke.texi: Document change in behvaior for -Ofast for
18381 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
18383 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
18385 (aarch64_pop_regs): Likewise.
18386 (aarch64_gen_load_pair): Likewise.
18387 (aarch64_save_callee_saves): Likewise.
18388 (aarch64_restore_callee_saves): Likewise.
18390 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18392 * config/i386/i386.c: Revert the last change.
18394 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18397 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18399 (ix86_finalize_stack_frame_flags): This. Also clear
18400 frame_pointer_needed if -fno-omit-frame-pointer is used without
18402 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18403 with ix86_finalize_stack_frame_flags.
18404 (ix86_expand_epilogue): Likewise.
18405 (ix86_expand_split_stack_prologue): Likewise.
18407 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18410 * config/i386/i386.c (get_builtin_code_for_version): Set priority
18411 to P_AES for Westmere.
18413 2017-08-07 Jonathan Yong <10walls@gmail.com>
18415 * config/i386/mingw.opt (fset-stack-executable): Removed.
18416 * config/i386/cygming.opt (fset-stack-executable): Moved
18418 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
18420 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
18422 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
18424 2017-08-07 Marek Polacek <polacek@redhat.com>
18426 PR middle-end/81737
18427 * fold-const.c (fold_indirect_ref_1): Check type_domain.
18429 2017-08-07 Martin Liska <mliska@suse.cz>
18431 * attribs.h (canonicalize_attr_name): New function.
18432 (cmp_attribs): Move from c-format.c and adjusted.
18433 (is_attribute_p): Moved from tree.h.
18434 * tree-inline.c: Add new includes.
18435 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
18436 (private_is_attribute_p): Remove.
18437 (private_lookup_attribute): Likewise.
18438 (private_lookup_attribute_by_prefix): Simplify.
18439 (remove_attribute): Use is_attribute_p.
18440 * tree.h: Remove removed declarations.
18442 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18444 PR middle-end/81698
18445 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
18446 instead of computing it in the function. Formatting fix.
18447 (expand_case): Don't rely on default_edge being the first edge,
18448 clear it if removing it, pass default_edge to
18449 emit_case_dispatch_table.
18450 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
18453 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
18455 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
18456 insn in the function, emit NOP after the insn.
18458 2017-08-06 Tom de Vries <tom@codesourcery.com>
18460 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
18463 2017-08-06 Tom de Vries <tom@codesourcery.com>
18465 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
18468 2017-08-04 Yury Gribov <tetra2005@gmail.com>
18470 PR tree-optimization/57371
18471 * match.pd: New pattern.
18473 2017-08-04 Marek Polacek <polacek@redhat.com>
18475 PR middle-end/81695
18476 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
18477 perform the computation in offset_int.
18479 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18481 PR tree-optimization/81136
18482 * tree-vectorizer.h: Include tree-hash-traits.h.
18483 (vec_base_alignments): New typedef.
18484 (vec_info): Add a base_alignments field.
18485 (vect_record_base_alignments): Declare.
18486 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
18488 (DR_IS_CONDITIONAL_IN_STMT): New macro.
18489 (create_data_ref): Add an is_conditional_in_stmt argument.
18490 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
18491 the is_conditional_in_stmt field.
18492 (data_ref_loc): Add an is_conditional_in_stmt field.
18493 (get_references_in_stmt): Set the is_conditional_in_stmt field.
18494 (find_data_references_in_stmt): Update call to create_data_ref.
18495 (graphite_find_data_references_in_stmt): Likewise.
18496 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
18497 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
18498 (vect_record_base_alignment): New function.
18499 (vect_record_base_alignments): Likewise.
18500 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
18501 for nested statements even if we fail to compute a misalignment.
18502 Use pooled base alignments for unconditional references.
18503 (vect_find_same_alignment_drs): Compare base addresses instead
18505 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
18506 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
18508 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18510 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
18511 Add an explicit name for the enum. Use auto_vec for slp_instances
18512 and grouped_stores.
18513 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
18515 (_bb_vec_info): Add a constructor and destructor.
18516 (vinfo_for_stmt): Return NULL for uids of -1 as well.
18517 (destroy_loop_vec_info): Delete.
18518 (vect_destroy_datarefs): Likewise.
18519 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
18520 (vec_info::vec_info): New function.
18521 (vec_info::~vec_info): Likewise.
18522 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
18523 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
18524 destroy_loop_vec_info.
18525 * tree-vect-loop.c (new_loop_vec_info): Replace with...
18526 (_loop_vec_info::_loop_vec_info): ...this.
18527 (destroy_loop_vec_info): Replace with...
18528 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
18529 the stmt_vec_infos. Leave handling of vec_info information to its
18530 destructor. Remove explicit vector releases.
18531 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
18532 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
18533 * tree-vect-slp.c (new_bb_vec_info): Replace with...
18534 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
18535 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
18536 (destroy_bb_vec_info): Replace with...
18537 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
18538 information to its destructor.
18539 (vect_slp_analyze_bb_1): Use new and delete instead of
18540 new_bb_vec_info and destroy_bb_vec_info.
18541 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
18544 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18546 * tree-data-ref.h (subscript): Add access_fn field.
18547 (data_dependence_relation): Add could_be_independent_p.
18548 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
18549 (same_access_functions): Move to tree-data-ref.c.
18550 * tree-data-ref.c (ref_contains_union_access_p): New function.
18551 (access_fn_component_p): Likewise.
18552 (access_fn_components_comparable_p): Likewise.
18553 (dr_analyze_indices): Add a reference to access_fn_component_p.
18554 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
18556 (constant_access_functions): Likewise.
18557 (add_other_self_distances): Likewise.
18558 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
18559 (initialize_data_dependence_relation): Use XCNEW and remove
18560 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
18561 of access functions that have the same type. Allow the
18562 subsequence to end with different bases in some circumstances.
18563 Record the chosen access functions in SUB_ACCESS_FN.
18564 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
18565 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
18566 (subscript_dependence_tester_1): Likewise dra and drb.
18567 (build_classic_dist_vector): Update calls accordingly.
18568 (subscript_dependence_tester): Likewise.
18569 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
18570 DDR_COULD_BE_INDEPENDENT_P.
18571 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
18572 comp_alias_ddrs instead of may_alias_ddrs.
18573 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
18575 (vect_analyze_data_ref_dependence): Use it if
18576 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
18577 distance vectors if that fails.
18578 (dependence_distance_ge_vf): New function.
18579 (vect_prune_runtime_alias_test_list): Use it. Don't clear
18580 LOOP_VINFO_MAY_ALIAS_DDRS.
18582 2017-08-04 Richard Biener <rguenther@suse.de>
18584 PR middle-end/81705
18585 * fold-const.c (fold_binary_loc): Properly restrict
18586 minus_var0 && minus_var1 case when associating undefined overflow
18589 2017-08-04 Tom de Vries <tom@codesourcery.com>
18591 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
18593 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18595 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18596 Don't start diagnostic messages with a capital letter.
18597 * config/rs6000/rs6000.c (rs6000_option_override_internal):
18599 (rs6000_invalid_builtin): Likewise.
18600 (rs6000_trampoline_init): Likewise.
18602 2017-08-03 Jakub Jelinek <jakub@redhat.com>
18605 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
18606 after setting changeable df flags.
18608 2017-08-03 Richard Biener <rguenther@suse.de>
18610 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
18611 up if the use is in USE - X.
18613 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
18615 * toplev.c (dumpfile.h): New include.
18616 (internal_error_reentered): New static function. Use it...
18617 (internal_error_function): ...here to handle reentered internal_error.
18619 2017-08-03 Richard Biener <rguenther@suse.de>
18621 PR middle-end/81148
18622 * fold-const.c (split_tree): Add minus_var and minus_con
18623 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
18624 here but always use minus_*.
18625 (associate_trees): Assert we never associate with MINUS_EXPR
18626 and NULL first operand. Do not recurse for PLUS_EXPR operands
18627 when associating as MINUS_EXPR either.
18628 (fold_binary_loc): Track minus_var and minus_con.
18630 2017-08-03 Tom de Vries <tom@codesourcery.com>
18633 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
18634 ACCEL_COMPILER, apply finish_options on
18635 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
18637 2017-08-03 Tom de Vries <tom@codesourcery.com>
18640 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
18641 function_entry_patch_area_size > 0.
18643 2017-08-03 Jakub Jelinek <jakub@redhat.com>
18646 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
18647 instead of 10??LU, perform unit multiplication in wide_int,
18648 don't change alloc_object_size_limit if the limit is larger
18651 PR tree-optimization/81655
18652 PR tree-optimization/81588
18653 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
18654 the case when ranges[i].low and high are 1 for unsigned type with
18657 PR middle-end/81052
18658 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
18659 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
18661 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18663 * tree-vrp.h: Add include guard.
18665 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
18668 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
18669 (ud2): New insn pattern.
18670 * config/i386/i386.c (ix86_expand_epilogue):
18671 For naked functions, generate ud2 instead of trap insn.
18673 2017-08-02 Marek Polacek <polacek@redhat.com>
18676 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
18678 2017-08-02 Tom de Vries <tom@codesourcery.com>
18679 Cesar Philippidis <cesar@codesourcery.com>
18681 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
18682 Add missing edge probabilities.
18684 2017-08-02 Tamar Christina <tamar.christina@arm.com>
18686 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
18687 Correct endianness.
18689 2017-08-02 Jakub Jelinek <jakub@redhat.com>
18691 PR middle-end/79499
18692 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
18693 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
18694 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
18695 prologue_seq sequences - if any.
18697 2017-08-02 Richard Biener <rguenther@suse.de>
18699 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
18700 via vectors if supported, integer extracts via punning if supported
18701 or otherwise vector extracts.
18703 2017-08-02 Richard Biener <rguenther@suse.de>
18705 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
18707 (bitmap_insert_into_set): ... this.
18709 2017-08-02 Richard Biener <rguenther@suse.de>
18711 PR tree-optimization/81633
18713 2015-08-17 Alan Hayward <alan.hayward@arm.com>
18715 PR tree-optimization/71752
18716 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
18718 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
18720 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
18721 (machine_function::call_ms2sysv_pad_out): Remove field.
18722 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
18723 (ix86_compute_frame_layout): Likewise.
18725 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
18728 * config/i386/i386.c (ix86_set_func_type): Disallow naked
18729 attribute with interrupt attribute.
18731 2017-08-01 Andrew Pinski <apinski@cavium.com>
18733 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
18734 BIT_INSERT_EXPR's operand 1
18735 to see if the types precision matches.
18737 2017-08-01 Martin Liska <mliska@suse.cz>
18739 PR middle-end/70140
18740 * builtins.c (expand_builtin_memcpy_args): Remove.
18741 (expand_builtin_memcpy): Call newly added function
18742 expand_builtin_memory_copy_args.
18743 (expand_builtin_memcpy_with_bounds): Likewise.
18744 (expand_builtin_mempcpy): Remove last argument.
18745 (expand_builtin_mempcpy_with_bounds): Likewise.
18746 (expand_builtin_memory_copy_args): New function created from
18747 expand_builtin_mempcpy_args with small modifications.
18748 (expand_builtin_mempcpy_args): Remove.
18749 (expand_builtin_stpcpy): Remove unused argument.
18750 (expand_builtin): Likewise.
18751 (expand_builtin_with_bounds): Likewise.
18753 2017-08-01 Martin Liska <mliska@suse.cz>
18756 Make mempcpy more optimal (PR middle-end/70140).
18758 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18761 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
18762 __builtin_vec_cmpne verify both arguments are compatible vectors
18763 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
18764 verify arg1_type is a pointer or array type. For __builtin_vec_st,
18765 move computation of aligned to after checking the argument types.
18769 * config/rs6000/vsx.md (vextract_fp_from_shorth,
18770 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
18773 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
18774 Doug Rupp <rupp@adacore.com>
18775 Olivier Hainque <hainque@adacore.com>
18777 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
18778 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
18780 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
18781 for TARGET_OS_CPP_BUILTIN.
18782 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
18783 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
18785 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
18786 passing required abi options to the assembler for EABI configurations.
18787 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
18788 of .text.hot and .text.unlikely sections for kernel modules when
18789 using ARM style exceptions.
18790 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
18791 options. Add EXTRA_CC1_SPEC.
18792 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
18794 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
18796 (ARM_TARGET2_DWARF_FORMAT): Define.
18797 * config/arm/t-vxworks: Adjust multilib control to removal of the
18798 Diab command line options.
18800 2017-08-01 Martin Liska <mliska@suse.cz>
18802 PR gcov-profile/81561
18803 * gcov.c (unblock): Make unblocking safe as we need to preserve
18804 index correspondence of blocks and block_lists.
18806 2017-08-01 Richard Biener <rguenther@suse.de>
18808 PR tree-optimization/81181
18809 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
18810 (compute_antic): ... end of iteration here.
18812 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
18814 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
18815 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
18816 (ftree-slp-vectorize): Likewise.
18817 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
18818 can no longer be set independent of flag_tree_loop_vectorize.
18819 * omp-general.c (emp_max_vf): Likewise.
18820 * opts.c (enable_fdo_optimizations): Remove references to
18821 flag_tree_vectorize, these are now implicit.
18822 (common_handle_option): Remove handling for OPT_ftree_vectorize,
18823 and leave it for the options machinery.
18825 2017-08-01 Martin Liska <mliska@suse.cz>
18827 PR middle-end/70140
18828 * builtins.c (expand_builtin_memcpy_args): Remove.
18829 (expand_builtin_memcpy): Call newly added function
18830 expand_builtin_memory_copy_args.
18831 (expand_builtin_memcpy_with_bounds): Likewise.
18832 (expand_builtin_mempcpy): Remove last argument.
18833 (expand_builtin_mempcpy_with_bounds): Likewise.
18834 (expand_builtin_memory_copy_args): New function created from
18835 expand_builtin_mempcpy_args with small modifications.
18836 (expand_builtin_mempcpy_args): Remove.
18837 (expand_builtin_stpcpy): Remove unused argument.
18838 (expand_builtin): Likewise.
18839 (expand_builtin_with_bounds): Likewise.
18841 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18844 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
18845 print "ds:" only for immediates in generic address space.
18847 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18850 * config/i386/i386.c (ix86_funciton_naked): New prototype.
18851 (ix86_function_ok_for_sibcall): Return false for naked functions.
18853 2017-08-01 Richard Biener <rguenther@suse.de>
18855 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
18856 (compute_antic): Seed worklist with exit block predecessors.
18857 * cfganal.c (dfs_find_deadend): For a cycle return the source
18858 of the edge closing it.
18860 2017-08-01 Tamar Christina <tamar.christina@arm.com>
18862 * config/aarch64/aarch64.c
18863 (aarch64_can_const_movi_rtx_p): Move 0 check.
18865 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18867 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
18868 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
18870 * match.pd: Ditto in address comparison pattern.
18872 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18874 PR tree-optimization/81627
18875 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
18876 closed ssa form for store-store chain.
18878 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18880 PR tree-optimization/81620
18881 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
18882 for store-store chain.
18884 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18886 PR tree-optimization/81588
18887 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
18888 ranges[i].in_p, invert comparison code ccode. For >/>=,
18889 swap rhs1 and rhs2 and comparison code unconditionally,
18890 for </<= don't do that. Don't swap rhs1/rhs2 again if
18891 ranges[i].in_p, instead invert comparison code ccode if
18892 opcode or oe->rank is BIT_IOR_EXPR.
18895 * optabs.def (vec_extract_optab, vec_init_optab): Change from
18896 a direct optab to conversion optab.
18897 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
18898 with GET_MODE_INNER as last argument instead of optab_handler.
18899 * expmed.c (extract_bit_field_1): Likewise. Use vector from
18900 vector extraction if possible and optab is available.
18901 * expr.c (store_constructor): Use convert_optab_handler instead
18902 of optab_handler. Use vector initialization from smaller
18903 vectors if possible and optab is available.
18904 * tree-vect-stmts.c (vectorizable_load): Likewise.
18905 * doc/md.texi (vec_extract, vec_init): Document that the optabs
18906 now have two modes.
18907 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
18908 of vec_init from half-sized vectors with the same element mode.
18909 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
18910 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
18911 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
18912 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
18913 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
18914 after mode in gen_vec_extract* calls.
18915 (vec_extract<mode>): Renamed to ...
18916 (vec_extract<mode><ssescalarmodelower>): ... this.
18917 (vec_extract<mode><ssehalfvecmodelower>): New expander.
18918 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
18919 element mode after mode in gen_vec_init* calls.
18920 (VEC_INIT_HALF_MODE): New mode iterator.
18921 (vec_init<mode>): Renamed to ...
18922 (vec_init<mode><ssescalarmodelower>): ... this.
18923 (vec_init<mode><ssehalfvecmodelower>): New expander.
18924 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
18925 (vec_extractv2sfsf): ... this.
18926 (vec_initv2sf): Renamed to ...
18927 (vec_initv2sfsf): ... this.
18928 (vec_extractv2si): Renamed to ...
18929 (vec_extractv2sisi): ... this.
18930 (vec_initv2si): Renamed to ...
18931 (vec_initv2sisi): ... this.
18932 (vec_extractv4hi): Renamed to ...
18933 (vec_extractv4hihi): ... this.
18934 (vec_initv4hi): Renamed to ...
18935 (vec_initv4hihi): ... this.
18936 (vec_extractv8qi): Renamed to ...
18937 (vec_extractv8qiqi): ... this.
18938 (vec_initv8qi): Renamed to ...
18939 (vec_initv8qiqi): ... this.
18940 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
18941 (vec_init<mode>): Renamed to ...
18942 (vec_init<mode><VEC_base_l>): ... this.
18943 (vec_extract<mode>): Renamed to ...
18944 (vec_extract<mode><VEC_base_l>): ... this.
18945 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
18946 (vec_initv2sfsf): ... this.
18947 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
18948 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
18949 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
18950 element mode after mode in gen_vec_init* calls.
18951 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
18952 (vec_init<mode><Vel>): ... this.
18953 (vec_extract<mode>): Renamed to ...
18954 (vec_extract<mode><Vel>): ... this.
18955 * config/aarch64/iterators.md (Vel): New mode attribute.
18956 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
18957 Add element mode after mode in gen_vec_extract* calls.
18958 * config/s390/vector.md (non_vec_l): New mode attribute.
18959 (vec_extract<mode>): Renamed to ...
18960 (vec_extract<mode><non_vec_l>): ... this.
18961 (vec_init<mode>): Renamed to ...
18962 (vec_init<mode><non_vec_l>): ... this.
18963 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
18964 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
18966 * config/arm/iterators.md (V_elem_l): New mode attribute.
18967 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
18968 (vec_extract<mode><V_elem_l>): ... this.
18969 (vec_extractv2di): Renamed to ...
18970 (vec_extractv2didi): ... this.
18971 (vec_init<mode>): Renamed to ...
18972 (vec_init<mode><V_elem_l>): ... this.
18973 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
18974 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
18975 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
18976 Add element mode after gen_vec_extract* calls.
18977 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
18978 (vec_init<mode><unitmode>): ... this.
18979 (vec_extract<mode>): Renamed to ...
18980 (vec_extract<mode><unitmode>): ... this.
18981 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
18982 (vec_init<mode><unitmode>): ... this.
18983 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
18984 (vec_initv2sfsf): ... this.
18985 (vec_extractv2sf): Renamed to ...
18986 (vec_extractv2sfsf): ... this.
18987 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
18988 Add element mode after gen_vec_extract* calls.
18989 * config/mips/mips.md (unitmode): New mode iterator.
18990 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
18991 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
18992 * config/spu/spu.md (inner_l): New mode attribute.
18993 (vec_init<mode>): Renamed to ...
18994 (vec_init<mode><inner_l>): ... this.
18995 (vec_extract<mode>): Renamed to ...
18996 (vec_extract<mode><inner_l>): ... this.
18997 * config/sparc/sparc.md (veltmode): New mode iterator.
18998 (vec_init<VMALL:mode>): Renamed to ...
18999 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
19000 * config/ia64/vect.md (vec_initv2si): Renamed to ...
19001 (vec_initv2sisi): ... this.
19002 (vec_initv2sf): Renamed to ...
19003 (vec_initv2sfsf): ... this.
19004 (vec_extractv2sf): Renamed to ...
19005 (vec_extractv2sfsf): ... this.
19006 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
19007 (vec_init<mode>): Renamed to ...
19008 (vec_init<mode><VEC_base_l>): ... this.
19009 (vec_extract<mode>): Renamed to ...
19010 (vec_extract<mode><VEC_base_l>): ... this.
19011 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
19012 (vec_initv2sfsf): ... this.
19013 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
19014 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
19015 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
19016 gen_vec_init* calls.
19018 2017-08-01 Richard Biener <rguenther@suse.de>
19020 PR tree-optimization/81297
19021 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
19022 TREE_OVERFLOW from INTEGER_CSTs.
19024 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
19026 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
19028 2017-07-31 Carl Love <cel@us.ibm.com>
19030 * config/rs6000/rs6000-c: Add support for built-in functions
19031 vector signed char vec_xl_be (signed long long, signed char *);
19032 vector unsigned char vec_xl_be (signed long long, unsigned char *);
19033 vector signed int vec_xl_be (signed long long, signed int *);
19034 vector unsigned int vec_xl_be (signed long long, unsigned int *);
19035 vector signed long long vec_xl_be (signed long long, signed long long *);
19036 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
19037 vector signed short vec_xl_be (signed long long, signed short *);
19038 vector unsigned short vec_xl_be (signed long long, unsigned short *);
19039 vector double vec_xl_be (signed long long, double *);
19040 vector float vec_xl_be (signed long long, float *);
19041 * config/rs6000/altivec.h (vec_xl_be): Add #define.
19042 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
19043 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
19045 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
19046 (altivec_expand_builtin): Add switch statement to call
19047 altivec_expand_xl_be for each builtin.
19048 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
19049 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
19050 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
19051 __builtin_vsx_le_be_v16qi.
19052 * doc/extend.texi: Update the built-in documentation file for the
19053 new built-in functions.
19055 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19058 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
19060 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
19062 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19064 * config.gcc: Add z14.
19065 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
19066 CPU model numbers for z13s and z14.
19067 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
19069 * config/s390/s390-opts.h (enum processor_type): Rename
19070 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19071 * config/s390/s390.c (processor_table): Add field for CPU name to
19072 be passed to Binutils.
19073 (s390_asm_output_machine_for_arch): Use the new field in
19074 processor_table for Binutils.
19075 (s390_expand_builtin): Replace arch12 with z14.
19076 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19077 (s390_get_sched_attrmask): Likewise.
19078 (s390_get_unit_mask): Likewise.
19079 * config/s390/s390.opt: Add z14 to processor_type enum.
19081 2017-07-31 Martin Jambor <mjambor@suse.cz>
19084 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
19085 regardless of optimization level.
19087 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
19088 Martin Liska <mliska@suse.cz>
19090 * predict.def: Remove old comment and adjust probability.
19091 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
19092 PREDICT statements.
19094 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19097 * config/i386/i386.c (ix86_function_naked): New function.
19098 (ix86_can_use_return_insn_p): Return false for naked functions.
19099 (ix86_expand_prologue): Skip prologue for naked functions.
19100 (ix86_expand_epilogue): Skip epilogue for naked functions
19101 and emit trap instruction.
19102 (ix86_warn_func_return): New function.
19103 (ix86_attribute_table): Add "naked" attribute specification.
19104 (TARGET_WARN_FUNC_RETURN): Define.
19105 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
19107 2017-07-31 Martin Liska <mliska@suse.cz>
19109 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
19110 (dump_gimple_bb_header): Always dump BB info.
19111 (pp_cfg_jump): Do not append info about BB when dumping a jump.
19113 2017-07-31 Martin Liska <mliska@suse.cz>
19116 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
19117 also with current_function_decl non-null equality.
19119 2017-07-31 Jakub Jelinek <jakub@redhat.com>
19122 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
19123 change type to the element type, instead add eltype variable and
19124 use it where we are interested in the element type.
19126 PR tree-optimization/81603
19127 * ipa-polymorphic-call.c
19128 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
19129 offset arithmetic in offset_int, bail out if the resulting bit offset
19130 doesn't fit into shwi.
19132 2017-07-31 Martin Liska <mliska@suse.cz>
19134 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
19135 (gimplify_save_expr): Fix comment.
19137 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
19140 * config/i386/i386.c (ix86_function_arg): Update arguments for
19142 (ix86_compute_frame_layout): Set the initial stack offset to
19143 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
19144 INCOMING_FRAME_SP_OFFSET.
19145 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
19146 stack before exception handler returns.
19147 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
19148 the 'ERROR_CODE' for exception handler.
19150 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
19152 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
19153 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
19154 (ASM_OUTPUT_REG_POP): Ditto.
19155 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
19156 instead of asm_fprintf to output pure string.
19158 2017-07-29 Jakub Jelinek <jakub@redhat.com>
19160 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
19161 to imported_module_or_decl hook.
19162 (debug_nothing_tree_tree_tree_bool): Remove.
19163 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
19164 * debug.c (do_nothing_debug_hooks): Use
19165 debug_nothing_tree_tree_tree_bool_bool instead of
19166 debug_nothing_tree_tree_tree_bool.
19167 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19168 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
19169 * sdbout.c (sdb_debug_hooks): Likewise.
19170 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
19171 (gen_namespace_die): Add DW_AT_export_symbols attribute if
19173 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
19174 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
19175 attribute, don't add anything.
19177 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19179 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
19180 (fold_build2_stat_loc): Likewise.
19181 (fold_build3_stat_loc): Likewise.
19182 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
19183 (fold_build1_loc): Remove macro.
19184 (fold_build2_loc): Likewise.
19185 (fold_build3_loc): Likewise.
19187 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19189 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
19190 (gimple_build_debug_bind_source_stat): Likewise.
19191 * gimple.h (gimple_build_debug_bind): Remove macro.
19192 (gimple_build_debug_bind_source): Likewise.
19194 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19196 * bitmap.c (bitmap_alloc): Adjust.
19197 (bitmap_gc_alloc): Likewise.
19198 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
19200 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19202 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
19203 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
19204 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
19205 (bitmap_gc_alloc_stat): Likewise.
19206 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
19208 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19210 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
19211 * rtl.h (shallow_copy_rtx): Remove macro.
19213 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19215 * emit-rtl.c (gen_raw_REG): Adjust.
19216 * gengenrtl.c (gendef): Likewise.
19217 * rtl.c (rtx_alloc_stat): Remove _stat from name.
19218 * rtl.h (rtx_alloc): Remove macro.
19220 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19222 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
19223 (build_tree_list_stat): Likewise.
19224 * tree.h (build_tree_list): Remove macro.
19225 (build_tree_list_vec): Likewise.
19227 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19229 * tree.c (make_vector_stat): Remove _stat from name.
19230 (build_vector_stat): Likewise.
19231 * tree.h (make_vector_stat): Remove macro.
19232 (build_vector_stat): Likewise.
19234 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19236 * tree.h (build_var_debug_value): Remove prototype.
19238 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19240 * tree.c (tree_cons_stat): Remove _stat from name.
19241 * tree.h (tree_cons): Remove macro.
19243 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19245 * tree.c (build_vl_exp_stat): Remove _stat from name.
19246 * tree.h (build_vl_exp): Remove macro.
19248 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19250 * tree.c (build_decl_stat): Remove _stat from name.
19251 * tree.h (build_decl): Remove macro.
19253 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19255 * gimple.c (gimple_build_with_ops_stat): Adjust.
19256 (gimple_alloc_stat): Remove _stat from name.
19257 * gimple.h (gimple_alloc): Remove macro.
19259 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19261 * tree.c (make_tree_vec_stat): Remove _stat from name.
19262 (grow_tree_vec_stat): Likewise.
19263 * tree.h (make_tree_vec_stat): Adjust prototype.
19264 (grow_tree_vec_stat): Likewise.
19265 (make_tree_vec): Remove macro.
19266 (grow_tree_vec): Likewise.
19268 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19270 * fold-const.c (fold_build1_stat_loc): Adjust.
19271 (fold_build2_stat_loc): Likewise.
19272 (fold_build3_stat_loc): Likewise.
19273 * tree.c (build0_stat): Remove _stat from name.
19274 (build1_stat): Likewise.
19275 (build2_stat): Likewise.
19276 (build3_stat): Likewise.
19277 (build4_stat): Likewise.
19278 (build5_stat): Likewise.
19279 * tree.h (build1_loc): Remove macro, and rename _stat function
19281 (build2_loc): Likewise.
19282 (build3_loc): Likewise.
19283 (build4_loc): Likewise.
19284 (build5_loc): Likewise.
19286 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19288 * tree.c (make_int_cst_stat): Remove _stat from name.
19289 * tree.h (make_int_cst_stat): Adjust prototype.
19290 (make_int_cst): Remove macro.
19292 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19294 * tree.c (make_tre_binfo_stat): Remove _stat from name.
19295 * tree.h (make_tree_binfo_stat): Adjust prototype.
19296 (make_tree_binfo): Remove.
19298 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19300 * tree.c (copy_node_stat): Rename to copy_node.
19301 (build_distinct_type_copy): Adjust.
19302 * tree.h (copy_node_stat): Adjust prototype.
19303 (copy_node): Remove macro.
19305 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19307 * tree.c (make_node_stat): rename to make_node.
19308 (build_tree_list_stat): Adjust.
19309 (build0_stat): Likewise.
19310 (build2_stat): Likewise.
19311 (build3_stat): Likewise.
19312 (build4_stat): Likewise.
19313 (build5_stat): Likewise.
19314 (build_decl_stat): Likewise.
19315 * tree.h (make_node_stat): Adjust prototype.
19316 (make_node): remove macro.
19318 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
19320 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
19321 (PPC_FEATURE2_SCV): Likewise.
19322 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
19324 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19326 * config/aarch64/aarch64.c
19327 (aarch64_internal_mov_immediate): Add new special pattern.
19328 * config/aarch64/aarch64.md (*movdi_aarch64):
19329 Add reg/32bit const mov case.
19331 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19332 Richard Sandiford <richard.sandiford@linaro.org>
19334 * config/aarch64/aarch64.md (mov<mode>): Generalize.
19335 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
19336 Add integer and movi cases.
19337 (movi-split-hf-df-sf split, fp16): New.
19338 (enabled): Added TARGET_FP_F16INST.
19339 * config/aarch64/iterators.md (GPF_HF): New.
19340 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
19342 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19344 * config/aarch64/aarch64.c
19345 (aarch64_simd_container_mode): Add prototype.
19346 (aarch64_expand_mov_immediate): Add HI support.
19347 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
19348 (aarch64_can_const_movi_rtx_p): New.
19349 (aarch64_preferred_reload_class):
19350 Remove restrictions of using FP registers for certain SIMD operations.
19351 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
19352 (aarch64_valid_floating_const): Add integer move validation.
19353 (aarch64_simd_imm_scalar_p): Remove.
19354 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
19355 (aarch64_legitimate_constant_p): Expand list of supported cases.
19356 * config/aarch64/aarch64-protos.h
19357 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
19358 (aarch64_reinterpret_float_as_int): New.
19359 (aarch64_simd_imm_scalar_p): Remove.
19360 * config/aarch64/constraints.md (Uvi): New.
19361 (Dd): Split into Ds and new Dd.
19362 * config/aarch64/aarch64.md (*movsi_aarch64):
19364 (*movdi_aarch64): Add SIMD mov case.
19366 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19368 * tree-predcom.c: (struct chain): Handle store-store chain in which
19369 stores for elimination only store loop invariant values.
19370 (execute_pred_commoning_chain): Ditto.
19371 (prepare_initializers_chain_store_elim): Ditto.
19372 (prepare_finalizers): Ditto.
19373 (is_inv_store_elimination_chain): New function.
19374 (initialize_root_vars_store_elim_1): New function.
19376 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19378 * tree-predcom.c: Revise general description of the pass.
19379 (enum chain_type): New enum type for store elimination.
19380 (struct chain): New field supporting store elimination.
19381 (struct component): Ditto.
19382 (dump_chain): Dump store-stores chain.
19383 (release_chain): Release resources.
19384 (split_data_refs_to_components): Compute and create component
19385 contains only stores for elimination.
19386 (get_chain_last_ref_at): New function.
19387 (make_invariant_chain): Initialization.
19388 (make_rooted_chain): Specify chain type in parameter and record it.
19389 (add_looparound_copies): Skip for store-stores chain.
19390 (determine_roots_comp): Compute type of chain and pass it to
19392 (initialize_root_vars_store_elim_2): New function.
19393 (finalize_eliminated_stores): New function.
19394 (remove_stmt): Handle store for elimination.
19395 (execute_pred_commoning_chain): Execute predictive commoning on
19396 store-store chains.
19397 (determine_unroll_factor): Skip unroll for store-stores chain.
19398 (prepare_initializers_chain_store_elim): New function.
19399 (prepare_initializers_chain): Hanlde store-store chain.
19400 (prepare_finalizers_chain, prepare_finalizers): New function.
19401 (tree_predictive_commoning_loop): Return integer value indicating
19402 if loop is unrolled or lcssa form is corrupted.
19403 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
19405 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19407 * tree-predcom.c (initialize_root): Delete.
19408 (execute_pred_commoning_chain): Initialize root vars and replace
19409 reference of non-combined chain directly, rather than call above
19412 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19414 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
19415 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
19417 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19419 * tree-predcom.c (struct chain): New field init_seq.
19420 (release_chain): Release init_seq.
19421 (prepare_initializers_chain): Record intialization stmts in above
19423 (insert_init_seqs): New function.
19424 (tree_predictive_commoning_loop): Call insert_init_seqs.
19426 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19428 * tree-predcom.c (determine_roots_comp): Skip trivial components.
19430 2017-07-28 Richard Biener <rguenther@suse.de>
19432 * match.pd: Remove superfluous :c.
19433 * genmatch.c (simplify::id): Add member.
19434 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
19436 (current_id): New global.
19437 (dt_node::parent): Move from ...
19438 (dt_operand::parent): ... here. Add for_id member.
19439 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
19440 (decision_tree::find_node): Relax order requirement when
19441 merging DT_TRUE nodes to ones inbetween the current simplify
19442 and the one we try to merge with. Add diagnostic whenever
19443 we need to enforce pattern order by not merging.
19444 (decision_tree::insert): Set current_id.
19445 (decision_tree::print_node): Dump parent node and for_id.
19446 (parser::last_id): Add member.
19447 (parser::push_simplify): Assign unique id.
19448 (parser::parser): Initialize last_id.
19450 2017-07-28 Martin Liska <mliska@suse.cz>
19453 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
19454 gimple_build_debug_bind.
19456 2017-07-28 Richard Biener <rguenther@suse.de>
19458 PR tree-optimization/81502
19459 * match.pd: Add pattern combining BIT_INSERT_EXPR with
19461 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
19463 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
19464 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
19465 for BIT_FIELD_REF args.
19466 * fold-const.c (make_bit_field_ref): Likewise.
19467 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
19469 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19472 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
19473 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
19474 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
19475 Or it into SANITIZER_UNDEFINED.
19476 * ubsan.c: Include gimple-fold.h and varasm.h.
19477 (ubsan_expand_ptr_ifn): New function.
19478 (instrument_pointer_overflow): New function.
19479 (maybe_instrument_pointer_overflow): New function.
19480 (instrument_object_size): Formatting fix.
19481 (pass_ubsan::execute): Call instrument_pointer_overflow
19482 and maybe_instrument_pointer_overflow.
19483 * internal-fn.c (expand_UBSAN_PTR): New function.
19484 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
19485 * sanitizer.def (__ubsan_handle_pointer_overflow,
19486 __ubsan_handle_pointer_overflow_abort): New builtins.
19487 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
19488 * internal-fn.def (UBSAN_PTR): New internal function.
19489 * opts.c (sanitizer_opts): Add pointer-overflow.
19490 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
19491 * fold-const.c (build_range_check): Compute pointer range check in
19492 integral type if pointer arithmetics would be needed. Formatting
19495 2017-07-28 Martin Liska <mliska@suse.cz>
19498 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
19499 parameters that are of a variable-length.
19501 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
19503 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
19505 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
19506 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
19507 (CRT_CALL_STATIC_FUNCTION): Likewise.
19508 (ASM_DEFAULT_SPEC): New define.
19509 (ASM_SPEC32): Likewise.
19510 (ASM_SPEC64): Likewise.
19511 (ASM_SPEC_COMMON): Likewise.
19512 (ASM_SPEC): Likewise.
19513 (INVALID_64BIT): Likewise.
19514 (LINK_OS_DEFAULT_SPEC): Likewise.
19515 (LINK_OS_SPEC32): Likewise.
19516 (LINK_OS_SPEC64): Likewise.
19517 (POWERPC_LINUX): Likewise.
19518 (PTRDIFF_TYPE): Likewise.
19519 (RESTORE_FP_PREFIX): Likewise.
19520 (RESTORE_FP_SUFFIX): Likewise.
19521 (SAVE_FP_PREFIX): Likewise.
19522 (SAVE_FP_SUFFIX): Likewise.
19523 (SIZE_TYPE): Likewise.
19524 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
19525 (TARGET_64BIT): Likewise.
19526 (TARGET_64BIT): Likewise.
19527 (TARGET_AIX): Likewise.
19528 (WCHAR_TYPE_SIZE): Likewise.
19529 (WCHAR_TYPE): Undefine.
19530 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
19531 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
19532 (CPP_OS_RTEMS_SPEC): Delete.
19533 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
19534 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
19536 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
19538 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19540 PR tree-optimization/81578
19541 * tree-parloops.c (build_new_reduction): Bail out if
19542 reduction_code isn't one of the standard OpenMP reductions.
19543 Move the details printing after that decision.
19545 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19547 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
19548 related to reload_in_progress.
19549 (splat_input_operand): Likewise.
19550 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
19552 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
19554 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
19555 (TARGET_INSTANTIATE_DECLS): Likewise.
19556 (legitimate_indexed_address_p): Delete reload_in_progress code.
19557 (rs6000_debug_legitimate_address_p): Likewise.
19558 (rs6000_eliminate_indexed_memrefs): Likewise.
19559 (rs6000_emit_le_vsx_store): Likewise.
19560 (rs6000_emit_move_si_sf_subreg): Likewise.
19561 (rs6000_emit_move): Likewise.
19562 (register_to_reg_type): Likewise.
19563 (rs6000_pre_atomic_barrier): Likewise.
19564 (rs6000_machopic_legitimize_pic_address): Likewise.
19565 (rs6000_allocate_stack_temp): Likewise.
19566 (rs6000_address_for_fpconvert): Likewise.
19567 (rs6000_address_for_altivec): Likewise.
19568 (rs6000_secondary_memory_needed_rtx): Delete function.
19569 (rs6000_check_sdmode): Likewise.
19570 (rs6000_alloc_sdmode_stack_slot): Likewise.
19571 (rs6000_instantiate_decls): Likewise.
19572 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
19573 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
19574 Delete reload_in_progress.
19575 (*vec_reload_and_plus_<mptrsize>): Likewise.
19576 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
19577 (vsx_div_v2di): Likewise.
19578 (vsx_udiv_v2di): Likewise.
19580 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19582 * config/rs6000/rs6000.opt (mlra): Replace with stub.
19583 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
19584 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
19585 (rs6000_debug_reg_global): Delete print of LRA status.
19586 (rs6000_option_override_internal): Delete dead LRA related code.
19587 (rs6000_lra_p): Delete function.
19588 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
19590 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
19592 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
19593 * config/riscv/rtems.h: New file.
19595 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19596 Sudakshina Das <sudi.das@arm.com>
19598 * config/aarch64/aarch64.md
19599 (define_split for and<mode>3nr_compare): Move
19600 non aarch64_logical_operand to a register.
19601 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
19602 register immediate operand to a register.
19603 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
19605 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19607 PR middle-end/81564
19608 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
19610 2017-07-27 Richard Biener <rguenther@suse.de>
19612 PR tree-optimization/81573
19613 PR tree-optimization/81494
19614 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
19615 multi defuse cycle case.
19617 2017-07-27 Richard Biener <rguenther@suse.de>
19619 PR tree-optimization/81571
19620 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
19623 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
19625 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
19626 earlier and only if MASK_FPU is set. Adjust formatting.
19628 2017-07-27 Martin Liska <mliska@suse.cz>
19630 * opt-functions.awk: Add validation of value of Init.
19631 * optc-gen.awk: Pass new argument.
19633 2017-07-27 Martin Liska <mliska@suse.cz>
19635 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
19636 Fix wrong condition.
19638 2017-07-27 Martin Liska <mliska@suse.cz>
19640 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
19641 BBs and edges seen by autoFDO.
19643 2017-07-27 Richard Biener <rguenther@suse.de>
19645 PR tree-optimization/81502
19646 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
19647 with incompatible but same sized type.
19648 (execute_update_addresses_taken): Likewise.
19650 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
19652 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
19653 flag_tree_loop_vectorize rather than flag_tree_vectorize.
19655 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19658 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
19659 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
19660 Change s_operand to memory_operand.
19662 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
19664 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
19665 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
19666 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
19667 Emit instructions rather than returning an expression. Handle TFmode
19668 and KFmode by casting to TImode.
19669 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
19670 (rs6000_emit_le_vsx_store): Likewise.
19671 * config/rs6000/vsx.md (VSX_TI): New iterator.
19672 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
19673 (*vsx_le_undo_permute_<mode>): Likewise.
19674 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
19675 emit the split sequence.
19676 (*vsx_le_perm_store_<mode>): Likewise.
19678 2017-07-27 Jakub Jelinek <jakub@redhat.com>
19680 PR tree-optimization/81555
19681 PR tree-optimization/81556
19682 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
19683 if true, force CHANGED for the recursive invocation.
19684 (reassociate_bb): Remember original length of ops array, pass
19685 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
19687 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
19688 attributes for noipa attribute. For naked attribute use
19689 lookup_attribute first before lookup_attribute_spec.
19690 * final.c (rest_of_handle_final): Disable IPA RA for functions with
19692 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
19693 for functions with noipa attribute.
19694 (cgraph_externally_visible_p): Return true for functions with noipa
19696 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
19697 for functions with noipa attribute.
19698 * doc/extend.texi: Document noipa function attribute.
19699 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
19700 also for functions with noipa attribute.
19701 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
19703 2017-07-26 Andrew Pinski <apinski@cavium.com>
19705 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
19706 vec_unalign_load_cost and vec_unalign_store_cost.
19708 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
19710 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
19711 -mvsx-small-integer option.
19712 (ISA_3_0_MASKS_IEEE): Likewise.
19713 (OTHER_VSX_VECTOR_MASKS): Likewise.
19714 (POWERPC_MASKS): Likewise.
19715 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
19716 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
19717 code, only testing for DImode being allowed in non-VSX floating
19719 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
19720 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
19722 (rs6000_option_override_internal): Delete -mvsx-small-integer.
19723 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
19724 TARGET_P8_VECTOR test.
19725 (rs6000_secondary_reload_simple_move): Likewise.
19726 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
19727 since TARGET_P9_VECTOR was already tested.
19728 (rs6000_opt_masks): Remove -mvsx-small-integer.
19729 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
19730 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19732 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
19733 test for TARGET_VEXTRACTUB was used, and that uses
19735 (p9 extract splitter): Likewise.
19736 (vsx_extract_<mode>_di_p9): Likewise.
19737 (vsx_extract_<mode>_store_p9): Likewise.
19738 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
19739 for TARGET_P9_VECTOR was used. Delete code that is now dead with
19740 the elimination of TARGET_VSX_SMALL_INTEGER.
19741 (vsx_extract_<mode>_p8): Likewise.
19742 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
19743 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
19744 (vsx_set_<mode>_p9): Likewise.
19745 (vsx_set_v4sf_p9): Likewise.
19746 (vsx_set_v4sf_p9_zero): Likewise.
19747 (vsx_insert_extract_v4sf_p9): Likewise.
19748 (vsx_insert_extract_v4sf_p9_2): Likewise.
19749 * config/rs6000/rs6000.md (sign extend splitter): Change
19750 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
19751 (floatsi<mode>2_lfiwax_mem): Likewise.
19752 (floatunssi<mode>2_lfiwzx_mem): Likewise.
19753 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
19754 since a test for TARGET_P9_VECTOR was used.
19755 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19756 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
19757 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19758 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
19759 TARGET_P8_VECTOR test.
19760 (fix_trunc<mode>si2_stfiwx): Likewise.
19761 (fix_trunc<mode>si2_internal): Likewise.
19762 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
19763 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19765 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19766 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
19767 TARGET_P8_VECTOR test.
19768 (fixuns_trunc<mode>si2_stfiwx): Likewise.
19769 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
19770 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19772 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19773 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
19774 since a test for TARGET_P9_VECTOR was used.
19775 (splitter for loading small constants): Likewise.
19777 2017-07-26 Andrew Pinski <apinski@cavium.com>
19779 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
19782 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
19785 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
19786 (fp_valid_at): Likewise.
19788 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
19790 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
19791 (qdf24xx_addrcost_table): Likewise.
19792 (cortexa57_tunings): Update to use generic_branch_cost.
19793 (cortexa72_tunings): Likewise.
19794 (cortexa73_tunings): Likewise.
19795 (qdf24xx_tunings): Likewise.
19797 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
19799 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
19800 (thunderx2t99_branch_cost): Likewise.
19801 (cortexa35_tunings): Update to use generic_branch_cost.
19802 (cortexa53_tunings): Likewise.
19803 (cortexa57_tunings): Likewise.
19804 (cortexa72_tunings): Likewise.
19805 (cortexa73_tunings): Likewise.
19806 (thunderx2t99_tunings): Likewise.
19808 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19810 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
19811 (sparc_option_override): Honour MASK_FSMULD.
19812 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
19813 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
19814 * config/sparc/sparc.opt (mfsmuld): New option.
19815 * doc/invoke.texi (mfsmuld): Document option.
19817 2017-07-26 Marek Polacek <polacek@redhat.com>
19819 PR middle-end/70992
19820 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
19822 2017-07-26 Richard Biener <rguenther@suse.de>
19824 * gimple-match-head.c (do_valueize): Return OP if valueize
19826 (get_def): New helper to get at the def stmt of a SSA name
19827 if valueize allows.
19828 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
19829 do_valueize to get at the def stmt.
19830 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
19832 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
19834 PR middle-end/46932
19835 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
19837 2017-07-26 Martin Liska <mliska@suse.cz>
19840 * function.c (expand_function_start): Make expansion of
19841 nonlocal_goto_save_area after parm_birth_insn.
19843 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19845 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
19846 from all CPU target flags enable members.
19848 2017-07-26 Richard Biener <rguenther@suse.de>
19850 * genmatch.c (dt_simplify::gen): Make iterator vars const.
19851 (decision_tree::gen): Make 'type' const.
19852 (write_predicate): Likewise.
19854 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
19856 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
19857 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
19858 (rs6000_option_override_internal): Likewise.
19859 (rs6000_expand_vector_set): Likewise.
19860 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
19861 (TARGET_UPPER_REGS_SF): Likewise.
19862 (TARGET_UPPER_REGS_DI): Likewise.
19863 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
19864 (TARGET_DIRECT_MOVE_64BIT): Likewise.
19865 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
19866 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19867 (Splitters for DI constants in Altivec registers): Likewise.
19868 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
19869 (vsx_set_v4sf_p9): Likewise.
19870 (vsx_set_v4sf_p9_zero): Likewise.
19871 (vsx_insert_extract_v4sf_p9): Likewise.
19872 (vsx_insert_extract_v4sf_p9_2): Likewise.
19874 2017-07-25 Carl Love <cel@us.ibm.com>
19876 * doc/extend.texi: Update the built-in documentation file for the
19877 existing built-in functions
19878 vector signed char vec_cnttz (vector signed char);
19879 vector unsigned char vec_cnttz (vector unsigned char);
19880 vector signed short vec_cnttz (vector signed short);
19881 vector unsigned short vec_cnttz (vector unsigned short);
19882 vector signed int vec_cnttz (vector signed int);
19883 vector unsigned int vec_cnttz (vector unsigned int);
19884 vector signed long long vec_cnttz (vector signed long long);
19885 vector unsigned long long vec_cnttz (vector unsigned long long);
19887 2017-07-25 Andrew Pinski <apinski@cavium.com>
19889 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
19890 accesses where the use is for the first operand of a BIT_INSERT.
19892 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19895 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
19896 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
19898 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19900 * config/i386/gstabs.h: Delete.
19901 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
19903 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
19905 * config/i386/i386.c (ix86_decompose_address): Do not check for
19906 register RTX when looking at index_reg or base_reg.
19907 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
19909 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
19911 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
19912 to update EH info here.
19914 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19916 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
19918 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19920 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
19922 2017-07-25 Torsten Duwe <duwe@suse.de>
19924 * common.opt: Introduce -fpatchable-function-entry
19925 command line option, and its variables function_entry_patch_area_size
19926 and function_entry_patch_area_start.
19927 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
19928 including a two-value parser.
19929 * target.def (print_patchable_function_entry): New target hook.
19930 * targhooks.h (default_print_patchable_function_entry): New function.
19931 * targhooks.c (default_print_patchable_function_entry): Likewise.
19932 * toplev.c (process_options): Switch off IPA-RA if
19933 patchable function entries are being generated.
19934 * varasm.c (assemble_start_function): Look at the
19935 patchable-function-entry command line switch and current
19936 function attributes and maybe generate NOP instructions by
19937 calling the print_patchable_function_entry hook.
19938 * doc/extend.texi: Document patchable_function_entry attribute.
19939 * doc/invoke.texi: Document -fpatchable_function_entry
19940 command line option.
19941 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
19943 * doc/tm.texi: Re-generate.
19945 2017-07-25 Jakub Jelinek <jakub@redhat.com>
19948 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
19949 TARGET_AVX512DQ rather than TARGET_AVX512BW.
19951 2017-07-25 Tamar Christina <tamar.christina@arm.com>
19953 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
19955 2017-07-25 Richard Biener <rguenther@suse.de>
19957 PR tree-optimization/81455
19958 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
19959 not walk in cycles when looking for guards.
19961 2017-07-25 Richard Biener <rguenther@suse.de>
19963 PR tree-optimization/81529
19964 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
19965 when optimizing backedge uses.
19967 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
19969 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
19971 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
19972 to dl_section_ref. On AIX, append an expression to subtract
19973 the size of the section length to dl_section_ref.
19975 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
19977 * configure.ac: If any of the config.* scripts fail, exit 1.
19978 * configure: Regenerate.
19980 2017-07-25 Richard Biener <rguenther@suse.de>
19982 PR middle-end/81546
19983 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
19984 of immediate uses, be more verbose on errors.
19986 2017-07-25 Richard Biener <rguenther@suse.de>
19988 PR tree-optimization/81510
19989 * tree-vect-loop.c (vect_is_simple_reduction): When the
19990 reduction stmt is not inside the loop bail out.
19992 2017-07-25 Richard Biener <rguenther@suse.de>
19994 PR tree-optimization/81303
19995 * tree-vect-loop-manip.c (vect_loop_versioning): Build
19996 profitability check against LOOP_VINFO_NITERSM1.
19998 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20000 * domwalk.c (cmp_bb_postorder): Simplify.
20001 (sort_bbs_postorder): New function. Use it...
20002 (dom_walker::walk): ...here to optimize common cases.
20004 2017-07-25 Martin Liska <mliska@suse.cz>
20007 * ipa-visibility.c (function_and_variable_visibility): Make the
20008 redirection just on target that supports aliasing.
20009 Fix GNU coding style.
20011 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
20014 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
20016 * config/arm/rtems.h: Likewise.
20017 * config/bfin/rtems.h: Likewise.
20018 * config/i386/rtemself.h: Likewise.
20019 * config/lm32/rtems.h: Likewise.
20020 * config/m32c/rtems.h: Likewise.
20021 * config/m68k/rtemself.h: Likewise.
20022 * config/microblaze/rtems.h: Likewise.
20023 * config/mips/rtems.h: Likewise.
20024 * config/moxie/rtems.h: Likewise.
20025 * config/nios2/rtems.h: Likewise.
20026 * config/powerpcspe/rtems.h: Likewise.
20027 * config/rs6000/rtems.h: Likewise.
20028 * config/rtems.h: Likewise.
20029 * config/sh/rtems.h: Likewise.
20030 * config/sh/rtemself.h: Likewise.
20031 * config/sparc/rtemself.h: Likewise.
20033 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20036 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
20037 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
20038 * tree-ssa-structalias.c (alias_get_name): Same.
20040 2017-07-25 Bin Cheng <bin.cheng@arm.com>
20043 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
20044 instructions if no du chain is found.
20046 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20048 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
20050 2017-07-25 Richard Biener <rguenther@suse.de>
20052 PR middle-end/81505
20053 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
20056 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
20058 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
20059 upper-regs options.
20060 (ISA_2_7_MASKS_SERVER): Likewise.
20061 (ISA_3_0_MASKS_IEEE): Likewise.
20062 (OTHER_P8_VECTOR_MASKS): Likewise.
20063 (OTHER_VSX_VECTOR_MASKS): Likewise.
20064 (POWERPC_MASKS): Likewise.
20065 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
20066 duplicate list of options.
20067 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
20068 explicit -mupper-regs options.
20069 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
20070 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
20071 alias for -mupper-regs-df.
20072 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20073 (rs6000_init_hard_regno_mode_ok): Likewise.
20074 (rs6000_option_override_internal): Likewise.
20075 (rs6000_opt_masks): Likewise.
20076 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
20077 options in terms of whether -mvsx or -mpower8-vector was used.
20078 (TARGET_UPPER_REGS_DI): Likewise.
20079 (TARGET_UPPER_REGS_SF): Likewise.
20080 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
20081 -mupper-regs-* options.
20083 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
20085 * passes.c (emergency_dump_function): Print some empty lines and a
20086 header before the RTL dump.
20088 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
20090 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
20092 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
20095 * config/aarch64/aarch64.c (aarch64_classify_symbol):
20096 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
20098 2017-07-24 Carl Love <cel@us.ibm.com>
20100 * config/rs6000/rs6000-c.c: Add support for built-in functions
20101 vector float vec_extract_fp32_from_shorth (vector unsigned short);
20102 vector float vec_extract_fp32_from_shortl (vector unsigned short);
20103 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
20104 vec_extract_fp_from_shortl): Add defines for the two builtins.
20105 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
20106 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
20108 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
20109 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
20110 * doc/extend.texi: Update the built-in documentation file for the
20111 new built-in function.
20113 2017-07-24 Jakub Jelinek <jakub@redhat.com>
20116 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
20118 * doc/generic.texi: Likewise.
20119 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
20120 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
20122 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
20124 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
20125 (aarch64_mls_elt_merge<mode>): Likewise.
20127 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
20129 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
20130 having __cxa_atexit.
20132 2017-07-23 Michael Collison <michael.collison@arm.com>
20134 * config/arm/arm.c (arm_option_override): Deprecate
20135 use of -mstructure-size-boundary.
20136 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
20137 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
20139 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20142 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
20143 Reduce cost estimate for direct moves.
20145 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
20148 * config/i386/i386.c (ix86_option_override_internal): Disable
20149 BMI, BMI2 and TBM instructions for -m16.
20151 2017-07-21 Carl Love <cel@us.ibm.com>
20153 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20154 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20155 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20156 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20157 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20158 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20159 VMULOSW): New enum "unspec" values.
20160 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20161 altivec_vmulosw): New patterns.
20162 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20163 VMULOSW): Add definitions.
20165 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
20167 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
20168 (qdf24xx): Likewise.
20169 * config/aarch64/aarch64-options-extensions.def (rdma); New.
20170 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
20171 (AARCH64_FL_V8_1): Renumber.
20172 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
20173 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
20174 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
20175 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
20176 rdma to feature modifiers list.
20178 2017-07-21 Yury Gribov <tetra2005@gmail.com>
20180 PR middle-end/56727
20181 * ipa-visibility (function_and_variable_visibility): Convert
20182 recursive PLT call to direct call if appropriate.
20184 2017-07-21 Andrew Pinski <apinski@cavium.com>
20186 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
20187 operand 1 to see if the types precision matches.
20188 * fold-const.c (operand_equal_p): Likewise.
20190 2017-07-21 Richard Biener <rguenther@suse.de>
20192 PR tree-optimization/81303
20193 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
20194 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
20195 (vect_peeling_hash_get_lowest_cost): Adjust.
20196 (vect_enhance_data_refs_alignment): Likewise. Use
20197 vect_get_peeling_costs_all_drs to compute the penalty for no
20198 peeling to match up costs.
20200 2017-07-21 Richard Biener <rguenther@suse.de>
20202 PR tree-optimization/81500
20203 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
20204 we didn't identify a reduction path.
20206 2017-07-21 Tom de Vries <tom@codesourcery.com>
20207 Cesar Philippidis <cesar@codesourcery.com>
20209 PR gcov-profile/81442
20210 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
20213 2017-07-21 Tom de Vries <tom@codesourcery.com>
20216 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
20218 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
20219 nvptx_override_options_after_change.
20221 2017-07-21 Ulrich Drepper <drepper@redhat.com>
20223 * dwarf2out.c (output_file_names): Avoid double testing for
20224 dwarf_version >= 5.
20226 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
20228 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
20230 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20232 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
20234 (try_crossjump_to_edge): Do not punt on partitioned functions.
20236 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20238 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
20239 Put all BBs reachable only via paths crossing cold region to cold
20241 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
20243 2016-07-21 Richard Biener <rguenther@suse.de>
20245 PR tree-optimization/81303
20246 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
20247 into account prologue and epilogue iterations when raising
20248 min_profitable_iters to sth at least covering one vector iteration.
20250 2017-07-21 Tamar Christina <tamar.christina@arm.com>
20252 * config/arm/arm.c (arm_test_cpu_arch_dat):
20255 2017-07-20 Nathan Sidwell <nathan@acm.org>
20257 Remove TYPE_METHODS.
20258 * tree.h (TYPE_METHODS): Delete.
20259 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
20260 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
20261 (dbxout_type_methods): Scan TYPE_FIELDS.
20262 (dbxout_type): Don't check TYPE_METHODS here.
20263 * function.c (use_register_for_decl): Always ignore register for
20264 class types when not optimizing.
20265 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
20266 * tree.c (free_lang_data_in_type): Stitch out member functions and
20267 templates from TYPE_FIELDS.
20268 (build_distinct_type_copy, verify_type_variant,
20269 verify_type): Member fns are on TYPE_FIELDS.
20270 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
20271 * tree-pretty-print.c (dump_generic_node): Likewise.
20273 2017-07-20 Jakub Jelinek <jakub@redhat.com>
20276 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
20277 V2TImode and V4TImode.
20278 (ix86_expand_vector_extract): Likewise.
20279 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
20280 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
20281 (ssescalarmode): Handle V4TImode and V2TImode.
20282 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
20283 (*vec_extractv2ti, *vec_extractv4ti): New insns.
20284 (VEXTRACTI128_MODE): New mode iterator.
20285 (splitter for *vec_extractv?ti first element): New.
20286 (VEC_INIT_MODE): New mode iterator.
20287 (vec_init<mode>): Consolidate 3 expanders into one using
20288 VEC_INIT_MODE mode iterator.
20290 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20292 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
20293 non_spilled_static_chain_regno_p.
20295 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20297 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
20299 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
20301 * bb-reorder.c (connect_traces): Allow copying of blocks within
20304 2017-07-20 Richard Biener <rguenther@suse.de>
20306 * gimple.h (gimple_phi_result): Add gphi * overload.
20307 (gimple_phi_result_ptr): Likewise.
20308 (gimple_phi_arg): Likewise. Adjust index assert to only
20309 allow actual argument accesses rather than all slots available
20311 (gimple_phi_arg_def): Add gphi * overload.
20312 * tree-phinodes.c (make_phi_node): Initialize only actual
20314 (resize_phi_node): Clear memory not covered by old node,
20315 do not initialize excess argument slots.
20316 (reserve_phi_args_for_new_edge): Initialize new argument slot
20319 2017-07-20 Bin Cheng <bin.cheng@arm.com>
20321 PR tree-optimization/81388
20323 2016-07-21 Bin Cheng <bin.cheng@arm.com>
20325 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
20326 by removing computation of may_be_zero.
20328 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20329 Tom de Vries <tom@codesourcery.com>
20331 PR middle-end/81030
20332 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
20333 when gimple level profile disagrees with what RTL expander did.
20335 2017-07-20 Richard Biener <rguenther@suse.de>
20337 PR tree-optimization/61171
20338 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
20339 (vect_analyze_stmt): Add slp instance parameter.
20340 (vectorizable_reduction): Likewise.
20341 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
20342 (vect_is_simple_reduction): Deal with chains not detected
20343 as SLP reduction chain, specifically not properly associated
20344 chains containing a mix of plus/minus.
20345 (get_reduction_op): Remove.
20346 (get_initial_defs_for_reduction): Simplify, pass in whether
20347 this is a reduction chain, pass in the SLP node for the PHIs.
20348 (vect_create_epilog_for_reduction): Get the SLP instance as
20350 (vectorizable_reduction): Get the SLP instance as arg.
20351 During analysis remember the SLP node with the PHIs in the
20352 instance. Simplify getting at the vectorized reduction PHIs.
20353 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
20354 through SLP instance.
20355 (vect_slp_analyze_operations): Likewise.
20356 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
20357 (vect_transform_stmt): Likewise.
20359 2017-07-20 Tom de Vries <tom@codesourcery.com>
20361 PR tree-optimization/81489
20362 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
20363 read of phi arg location to before loop that modifies phi.
20365 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20367 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
20370 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20372 PR middle-end/81331
20373 * except.c (execute): Fix ordering issue.
20375 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20377 PR rtl-optimization/81423
20378 * combine.c (make_compound_operation_int): Don't try to optimize
20379 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
20381 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20383 PR rtl-optimization/81423
20384 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
20385 with a constant that is -1 in the truncated to mode.
20387 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20389 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
20390 (determine_unlikely_bbs): ... here.
20391 * predict.h (propagate_unlikely_bbs_forward): Declare.
20392 * cfgexpand.c (pass_expand::execute): Use it.
20393 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
20395 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
20396 propagate_unlikely_bbs_forward.
20398 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20400 PR middle-end/81331
20401 * except.c (maybe_add_nop_after_section_switch): New function.
20404 2017-07-19 Tom de Vries <tom@codesourcery.com>
20406 * gimple.h (gimple_phi_set_arg): Make assert more strict.
20408 2017-07-19 Tom de Vries <tom@codesourcery.com>
20410 * gimple.h (gimple_phi_arg): Make assert more strict.
20412 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
20414 * config.gcc (powerpc*-*-*): Add mmintrin.h.
20415 * config/rs6000/mmintrin.h: New file.
20416 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
20418 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20420 PR tree-optimization/81346
20421 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
20423 2017-07-19 Tom de Vries <tom@codesourcery.com>
20425 * config/nvptx/nvptx.md (VECIM): Add V2DI.
20427 2017-07-19 Tom de Vries <tom@codesourcery.com>
20429 * config/nvptx/nvptx-modes.def: Add V2DImode.
20430 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
20431 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
20432 (nvptx_output_mov_insn): Handle lack of mov.b128.
20433 (nvptx_print_operand): Handle 'H' and 'L' codes.
20434 (nvptx_vector_mode_supported): Allow V2DImode.
20435 (nvptx_preferred_simd_mode): New function.
20436 (nvptx_data_alignment): New function.
20437 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
20438 nvptx_preferred_simd_mode.
20439 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
20441 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
20443 2017-07-19 Tom de Vries <tom@codesourcery.com>
20445 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
20446 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
20447 (nvptx_vector_mode_supported): New function. Allow V2SImode.
20448 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
20449 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
20450 (mov<VECIM>_insn): New define_insn.
20451 (define_expand "mov<VECIM>): New define_expand.
20453 2017-07-19 Tom de Vries <tom@codesourcery.com>
20455 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
20457 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20459 PR tree-optimization/81346
20460 * fold-const.h (fold_div_compare, range_check_type): Declare.
20461 * fold-const.c (range_check_type): New function.
20462 (build_range_check): Use range_check_type.
20463 (fold_div_compare): No longer static, rewritten into
20464 a match.pd helper function.
20465 (fold_comparison): Don't call fold_div_compare here.
20466 * match.pd (X / C1 op C2): New optimization using fold_div_compare
20467 as helper function.
20469 2017-07-19 Nathan Sidwell <nathan@acm.org>
20471 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
20472 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
20473 * tree.c (find_decls_types_r, verify_type): Use
20474 TYPE_{MIN,MAX}_VALUE_RAW.
20475 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
20476 (hash_tree): Likewise.
20477 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20479 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20482 2017-07-18 Tom de Vries <tom@codesourcery.com>
20484 PR middle-end/81464
20485 * omp-expand.c (expand_omp_for_static_chunk): Handle
20486 equal-argument loop exit phi.
20488 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
20491 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
20492 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
20493 operand 2 predicate.
20494 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
20495 operand 2 predicate.
20496 (ror,rol -> rorx splitters): Use const_int_operand as
20497 operand 2 predicate.
20499 2017-06-18 Richard Biener <rguenther@suse.de>
20501 PR tree-optimization/81410
20502 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
20503 the gap in the ! slp_perm SLP case after each group.
20505 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20507 PR middle-end/81463
20508 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
20511 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20513 PR middle-end/81462
20514 * predict.c (set_even_probabilities): Cleanup; do not affect
20515 probabilities that are already known.
20516 (combine_predictions_for_bb): Call even when count is set.
20518 2017-07-18 Nathan Sidwell <nathan@acm.org>
20520 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
20523 2017-07-18 Bin Cheng <bin.cheng@arm.com>
20526 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
20527 optimization for loop niter analysis.
20529 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
20532 * config/avr/avr.c (avr_optimize_casesi): Don't use
20533 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
20535 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
20537 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
20538 body_cost_vec from _vect_peel_extended_info.
20539 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
20540 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
20543 2017-07-18 Bin Cheng <bin.cheng@arm.com>
20545 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
20547 2017-07-18 Richard Biener <rguenther@suse.de>
20549 PR tree-optimization/80620
20550 PR tree-optimization/81403
20551 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
20552 info when re-using a VN table entry.
20554 2017-07-18 Richard Biener <rguenther@suse.de>
20556 PR tree-optimization/81418
20557 * tree-vect-loop.c (vectorizable_reduction): Properly compute
20558 vectype_in. Verify that with lane-reducing reduction operations
20559 we have a single def-use cycle.
20561 2017-07-17 Carl Love <cel@us.ibm.com>
20563 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
20565 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20566 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20567 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20568 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20569 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20570 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20571 VMULOSW): New enum "unspec" values.
20572 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20573 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20574 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20575 altivec_vmulosw): New patterns.
20576 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20577 VMULOSW): Add definitions.
20579 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
20581 * config/alpha/alpha.c: Include predict.h.
20583 2017-07-17 Yury Gribov <tetra2005@gmail.com>
20585 * tree-vrp.c (compare_assert_loc): Fix comparison function
20586 to return predictable results.
20588 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20590 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
20592 (subdi3): Likewise.
20593 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
20594 * doc/invoke.texi (mexpand-adddi): Update text.
20596 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20598 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
20599 that also clobbers the CC register. The old expand code is moved
20601 (*arc_clzsi2): ... here.
20602 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
20603 the CC register. The old expand code is moved to ...
20604 (arc_ctzsi2): ... here.
20606 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20608 * config/arc/arc.opt (mindexed-loads): Use initial value
20609 TARGET_INDEXED_LOADS_DEFAULT.
20610 (mauto-modify-reg): Use initial value
20611 TARGET_AUTO_MODIFY_REG_DEFAULT.
20612 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
20613 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
20614 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
20615 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
20617 2017-07-17 Martin Liska <mliska@suse.cz>
20620 * opts.c (finish_options): Do not allow -fgnu-tm
20621 w/ -fsanitize={kernel-,}address. Say sorry.
20623 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20626 * tree-loop-distribution.c (classify_partition): Only assert on
20627 numer of iterations.
20628 (merge_dep_scc_partitions): Delete prameter. Update function call.
20629 (distribute_loop): Remove code handling loop with unknown niters.
20630 (pass_loop_distribution::execute): Skip loop with unknown niters.
20632 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20635 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
20636 function sort_partitions_by_post_order.
20638 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20640 PR tree-optimization/81374
20641 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
20642 the max index of basic blocks, rather than number of basic blocks.
20644 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20646 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
20648 (arc_legitimate_pic_operand_p): Likewise.
20649 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
20651 (arc_needs_pcl_p): Likewise.
20652 (arc_legitimate_pc_offset_p): Likewise.
20653 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
20654 function is also used in constrains.md.
20655 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
20656 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
20657 PLUS. Only return true/false in known cases, otherwise assert.
20658 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
20659 is already called in arc_legitimate_constant_p.
20660 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
20662 (LEGITIMATE_PIC_OPERAND_P): Use
20663 arc_raw_symbolic_reference_mentioned_p function.
20664 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
20669 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20670 Andrew Burgess <andrew.burgess@embecosm.com>
20672 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
20673 (arc_return_address_register): New function.
20674 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
20675 (arc_handle_fndecl_attribute): Add naked attribute.
20676 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
20677 (TARGET_WARN_FUNC_RETURN): Likewise.
20678 (arc_allocate_stack_slots_for_args): New function.
20679 (arc_warn_func_return): Likewise.
20680 (machine_function): Change type fn_type.
20681 (arc_compute_function_type): Consider new naked function type,
20682 change function return type.
20683 (arc_must_save_register): Adapt to handle new
20684 arc_compute_function_type's return type.
20685 (arc_expand_prologue): Likewise.
20686 (arc_expand_epilogue): Likewise.
20687 (arc_return_address_regs): Delete.
20688 (arc_return_address_register): New function.
20689 (arc_epilogue_uses): Use above function.
20690 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
20691 (arc_function_type): Change encoding, add naked type.
20692 (ARC_INTERRUPT_P): Change to handle the new encoding.
20693 (ARC_FAST_INTERRUPT_P): Likewise.
20694 (ARC_NORMAL_P): Define.
20695 (ARC_NAKED_P): Likewise.
20696 (arc_compute_function_type): Delete prototype.
20697 * config/arc/arc.md (in_ret_delay_slot): Use
20698 arc_return_address_register function.
20699 (simple_return): Likewise.
20700 (p_return_i): Likewise.
20702 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20704 PR tree-optimization/81428
20705 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
20706 can't be built for those types.
20708 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
20710 Remove stuff dead since r239246.
20712 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
20713 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
20714 (avr_inform_devices): Remove dead stuff.
20716 2017-07-17 Tamar Christina <tamar.christina@arm.com>
20718 * config/arm/arm_neon.h: Fix softp typo.
20720 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20722 PR tree-optimization/81365
20723 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
20724 aggregate moves onto bb predecessor edges, make sure there are no
20725 loads that could alias the lhs in between the start of bb and the
20728 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
20731 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
20732 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
20733 [LSHIFTRT, outer_code = TRUNCATE]: Same.
20735 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20737 PR tree-optimization/81396
20738 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
20739 (init_symbolic_number): Initialize it to 1.
20740 (perform_symbolic_merge): Add n_ops from both operands into the new
20742 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
20743 without base_addr as useless if they need more than one operation.
20744 (bswap_replace): Handle !bswap case for NULL base_addr.
20746 2017-07-17 Tom de Vries <tom@codesourcery.com>
20749 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
20752 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
20754 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
20755 conditional builtin define __FIX_LEON3FT_B2BST.
20757 2017-07-17 Daniel Cederman <cederman@gaisler.com>
20759 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
20760 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
20763 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20765 PR rtl-optimization/81424
20766 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
20767 to remove potential trapping from operands if -fnon-call-exceptions.
20769 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20771 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
20772 profile_proability for scalling.
20773 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
20775 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20777 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
20779 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20781 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
20782 fixpoint arithmetics.
20784 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20786 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
20787 fixpoint arithmetics.
20789 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20791 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
20792 fixpoint arithmetics.
20794 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20796 * profile-count.h (profile_probability::from_reg_br_prob_note,
20797 profile_probability::to_reg_br_prob_note): New functions.
20798 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
20799 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
20800 * predict.c (probability_reliable_p): Update.
20801 (edge_probability_reliable_p): Update.
20802 (br_prob_note_reliable_p): Update.
20803 (invert_br_probabilities): Update.
20804 (add_reg_br_prob_note): New function.
20805 (combine_predictions_for_insn): Update.
20806 * asan.c (asan_clear_shadow): Update.
20807 * cfgbuild.c (compute_outgoing_frequencies): Update.
20808 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
20809 (update_br_prob_note): Update.
20810 (rtl_verify_edges): Update.
20811 (purge_dead_edges): Update.
20812 (fixup_reorder_chain): Update.
20813 * emit-rtl.c (try_split): Update.
20814 * ifcvt.c (cond_exec_process_insns): Update.
20815 (cond_exec_process_if_block): Update.
20816 (dead_or_predicable): Update.
20817 * internal-fn.c (expand_addsub_overflow): Update.
20818 (expand_neg_overflow): Update.
20819 (expand_mul_overflow): Update.
20820 * loop-doloop.c (doloop_modify): Update.
20821 * loop-unroll.c (compare_and_jump_seq): Update.
20822 * optabs.c (emit_cmp_and_jump_insn_1): Update.
20823 * predict.h: Update.
20824 * reorg.c (mostly_true_jump): Update.
20826 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
20827 * config/alpha/alpha.c (emit_unlikely_jump): Update.
20828 * config/arc/arc.c: (emit_unlikely_jump): Update.
20829 * config/arm/arm.c: (emit_unlikely_jump): Update.
20830 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
20831 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
20832 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
20833 (ix86_print_operand): Update.
20834 (ix86_split_fp_branch): Update.
20835 (predict_jump): Update.
20836 * config/ia64/ia64.c (ia64_print_operand): Update.
20837 * config/mmix/mmix.c (mmix_print_operand): Update.
20838 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
20839 (rs6000_expand_split_stack_prologue): Update.
20840 * config/rs6000/rs6000.c: Update.
20841 * config/s390/s390.c (s390_expand_vec_strlen): Update.
20842 (s390_expand_vec_movstr): Update.
20843 (s390_expand_cs_tdsi): Update.
20844 (s390_expand_split_stack_prologue): Update.
20845 * config/sh/sh.c (sh_print_operand): Update.
20846 (expand_cbranchsi4): Update.
20847 (expand_cbranchdi4): Update.
20848 * config/sparc/sparc.c (output_v9branch): Update.
20849 * config/spu/spu.c (get_branch_target): Update.
20850 (ea_load_store_inline): Update.
20851 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
20852 * config/tilepro/tilepro.c: Update.
20854 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20856 * gimplify.c (mostly_copy_tree_r): Revert latest change.
20857 (gimplify_save_expr): Likewise.
20859 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20861 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
20863 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20865 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
20867 * timevar.def (TV_IPA_FNSUMMARY): Define.
20869 2017-07-16 Daniel Cederman <cederman@gaisler.com>
20871 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
20872 to back store errata sensitive sequence from being generated.
20873 (sqrtdf2_fix): Likewise.
20875 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20877 * tree-ssa-threadupdate.c (compute_path_counts,
20878 update_joiner_offpath_counts): Use profile_probability.
20880 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
20883 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20885 * config/arm/arm-c.c (arm_cpu_builtins): Define
20886 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20888 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
20890 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20891 array entries to represent __ieee128 versions of the
20892 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
20893 scalar_extract_sig, and scalar_insert_exp built-in functions.
20894 (altivec_resolve_overloaded_builtin): Add special case handling
20895 for the __builtin_scalar_insert_exp function, as represented by
20896 the P9V_BUILTIN_VEC_VSIEDP constant.
20897 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
20898 exponent support for __ieee128 argument.
20899 (VSESQP): Add scalar extract signature support for __ieee128
20901 (VSTDCNQP): Add scalar test negative support for __ieee128
20903 (VSIEQP): Add scalar insert exponent support for __int128 argument
20904 with __ieee128 result.
20905 (VSIEQPF): Add scalar insert exponent support for __ieee128
20906 argument with __ieee128 result.
20907 (VSTDCQP): Add scalar test data class support for __ieee128
20909 (VSTDCNQP): Add overload support for scalar test negative with
20910 __ieee128 argument.
20911 (VSTDCQP): Add overload support for scalar test data class
20912 __ieee128 argument.
20913 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
20914 UNSPEC_VSX_SXSIGDP.
20915 (UNSPEC_VSX_SIEXPQP): New constant.
20916 (xsxexpqp): New insn for VSX scalar extract exponent quad
20918 (xsxsigqp): New insn for VSX scalar extract significand quad
20920 (xsiexpqpf): New insn for VSX scalar insert exponent quad
20921 precision with floating point argument.
20922 (xststdcqp): New expand for VSX scalar test data class quad
20924 (xststdcnegqp): New expand for VSX scalar test negative quad
20926 (xststdcqp): New insn to match expansions for VSX scalar test data
20927 class quad precision and VSX scalar test negative quad precision.
20928 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
20929 special case operand checking to enforce that second operand of
20930 VSX scalar test data class with quad precision argument is a 7-bit
20932 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20933 prototypes and descriptions of __ieee128 versions of
20934 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
20935 scalar_test_data_class, and scalar_test_neg built-in functions.
20937 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20939 PR tree-optimization/81162
20940 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
20941 replace a negate with an add.
20943 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
20945 * doc/invoke.texi (arm/-mcpu): Document +crypto.
20947 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20949 * config/arm/arm-c.c (arm_cpu_builtins): Define
20950 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20952 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20954 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
20955 (armv8-r): Set ARM Cortex-R52 as default CPU.
20956 * config/arm/arm-tables.opt: Regenerate.
20957 * config/arm/arm-tune.md: Regenerate.
20958 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
20960 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
20961 extension for -mcpu=cortex-r52.
20963 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20965 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
20966 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
20967 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
20968 (fp-armv8): Define it as FP_ARMv8 only.
20969 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
20970 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
20972 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
20973 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
20974 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
20975 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
20976 than TARGET_FPU_ARMV8.
20977 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
20978 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
20979 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
20981 * config/arm/neon.md (neon_vrint): Likewise.
20982 (neon_vcvt): Likewise.
20983 (neon_<fmaxmin_op><mode>): Likewise.
20984 (<fmaxmin><mode>3): Likewise.
20985 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
20986 * config/arm/predicates.md (arm_cond_move_operator): Check against
20987 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
20989 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
20991 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
20992 to top of function.
20994 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20996 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
20997 loop in comment with memset.
20999 2017-07-14 Martin Liska <mliska@suse.cz>
21001 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
21002 * dwarf2out.c (is_java): Remove the function.
21003 (output_pubname): Remove usage of the function.
21004 (lower_bound_default): Remove usage of DW_LANG_Java.
21005 (gen_compile_unit_die): Likewise.
21006 * gcc.c: Remove compiler defaults for .java and .zip files.
21007 * gimple-expr.c (remove_suffix): Change as there's no longer
21008 extension than 4-letter one.
21009 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
21010 (gimplify_save_expr): Likewise.
21011 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
21012 as it's possible even for other languages than Java.
21013 * langhooks.h (struct lang_hooks): Remove Java from a comment.
21014 * lto-opts.c (lto_write_options): Remove reference to Java.
21015 * opts.c (strip_off_ending): Update file extension handling.
21016 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
21017 * tree-eh.c (lower_resx): Likewise.
21018 * tree.c (free_lang_data_in_type): Remove dead code.
21019 (find_decls_types_r): Likewise.
21020 (build_common_builtin_nodes): Remove Java from a comment.
21021 (verify_type): Remove dead code.
21022 * varasm.c (assemble_external): Remove Java from a comment.
21024 2017-07-14 Martin Liska <mliska@suse.cz>
21026 * opts.c (finish_options): Add quotes.
21027 (common_handle_option): Likewise.
21029 2017-07-14 Martin Liska <mliska@suse.cz>
21031 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
21032 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
21033 Remove N_SO_PASCAL.
21034 * dwarf2out.c (lower_bound_default): Do not handle
21036 (gen_compile_unit_die): Likewise.
21037 * gcc.c: Remove default extension binding for GNU Pascal.
21038 * stmt.c: Remove Pascal language from a comment.
21039 * xcoffout.c: Likewise.
21041 2017-07-13 David Malcolm <dmalcolm@redhat.com>
21044 * diagnostic-show-locus.c (fixit_cmp): New function.
21045 (layout::layout): Sort m_fixit_hints.
21046 (column_range::column_range): Assert that the values are valid.
21047 (struct char_span): New struct.
21048 (correction::overwrite): New method.
21049 (struct source_line): New struct.
21050 (line_corrections::add_hint): Add assertions. Reimplement memcpy
21051 calls in terms of classes source_line and char_span, and
21052 correction::overwrite.
21053 (selftest::test_overlapped_fixit_printing_2): New function.
21054 (selftest::diagnostic_show_locus_c_tests): Call it.
21056 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
21058 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
21059 early if there is no lhs.
21061 2017-07-13 Martin Liska <mliska@suse.cz>
21063 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
21064 (gen_reference_type_die): Likewise.
21065 * stor-layout.c: Remove Pascal-related comment.
21067 2017-07-13 Martin Liska <mliska@suse.cz>
21069 * opts.c (finish_options): Add quotes to error messages.
21070 (parse_sanitizer_options): Likewise.
21072 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
21074 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
21076 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
21078 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
21080 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
21082 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
21084 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
21086 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
21089 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
21090 provides the hardware capability bits, define the macro
21091 __BUILTIN_CPU_SUPPORTS__.
21092 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
21093 if GLIBC does not provide the hardware capability bits. Add a
21094 gcc_unreachable call if the built-in cpu function is neither
21095 __builtin_cpu_is nor __builtin_cpu_supports.
21096 (rs6000_get_function_versions_dispatcher): Change the warning
21097 that an old GLIBC is used which does not export the capability
21098 bits to be an error.
21099 * doc/extend.texi (target_clones attribute): Document the
21100 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
21101 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
21102 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
21103 the macros defined by GCC if the newer GLIBC is available.
21105 2017-07-12 Jeff Law <law@redhat.com>
21107 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
21108 remaining includes slightly.
21109 * config/riscv/riscv-builtins.c: Include profile-count.h.
21111 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21114 * config/avr/avr.c (avr_set_current_function): In diagnostic
21115 messages: Quote keywords and (parts of) identifiers.
21116 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
21119 2017-07-12 Carl Love <cel@us.ibm.com>
21121 * config/rs6000/rs6000-c.c: Add support for built-in functions
21122 vector bool char vec_revb (vector bool char);
21123 vector bool short vec_revb (vector short char);
21124 vector bool int vec_revb (vector bool int);
21125 vector bool long long vec_revb (vector bool long long);
21126 * doc/extend.texi: Update the built-in documentation file for the
21127 new built-in functions.
21129 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21131 * config/s390/s390.md: Remove movcc splitter.
21133 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21135 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
21136 load/store on condition.
21138 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21141 * config/avr/avr.c (avr_encode_section_info)
21142 [progmem && !TREE_READONLY]: Error if progmem object needs
21145 2017-07-11 Michael Collison <michael.collison@arm.com>
21147 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
21150 2017-07-11 Carl Love <cel@us.ibm.com>
21152 * config/rs6000/rs6000-c.c: Add support for builtins
21153 vector unsigned int vec_parity_lsbb (vector signed int);
21154 vector unsigned int vec_parity_lsbb (vector unsigned int);
21155 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
21156 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
21157 vector unsigned long long vec_parity_lsbb (vector signed long long);
21158 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
21159 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
21160 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
21161 * doc/extend.texi: Update the built-in documentation file for the
21162 new built-in functions.
21164 2017-07-11 David Malcolm <dmalcolm@redhat.com>
21166 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
21167 (layout::m_primary_loc): New field.
21168 (layout::layout): Initialize new field. Move location filtering
21169 logic from here to...
21170 (layout::maybe_add_location_range): ...this new method. Add
21171 support for filtering to just the lines already specified by other
21173 (layout::will_show_line_p): New method.
21174 (gcc_rich_location::add_location_if_nearby): New method.
21175 (selftest::test_add_location_if_nearby): New test function.
21176 (selftest::diagnostic_show_locus_c_tests): Call it.
21177 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
21180 2017-07-11 Tom de Vries <tom@codesourcery.com>
21182 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
21183 (bb_first_real_insn): New function.
21184 (nvptx_single): Add extra initialization of broadcasted condition
21187 2017-07-11 Nathan Sidwell <nathan@acm.org>
21189 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
21191 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
21193 * doc/extend.texi (AVR Function Attributes): Remove weblink to
21194 Binutils doc as TEXI will mess them up.
21195 * doc/invoke.texi (AVR Options): Same here.
21197 2017-07-11 Daniel Cederman <cederman@gaisler.com>
21199 * config/sparc/sparc.opt (mfix-ut700): New option.
21200 (mfix-gr712rc): Likewise.
21201 (sparc_fix_b2bst): New variable.
21202 * doc/invoke.texi (SPARC options): Document them.
21203 (ARM options): Fix warnings.
21204 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
21205 instructions to prevent sequences that can trigger the store-store
21206 errata for certain LEON3FT processors.
21207 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
21208 (sparc_option_override): Set sparc_fix_b2bst appropriately.
21209 * config/sparc/sparc.md (fix_b2bst): New attribute.
21210 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
21212 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
21215 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
21217 (*rsqrtsf2_sse): Ditto.
21219 (div<mode>3): Macroize insn from divdf3 and divsf3
21220 using MODEF mode iterator.
21222 2017-07-10 Martin Sebor <msebor@redhat.com>
21224 PR tree-optimization/80397
21225 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
21226 instead of testing for equality to INTEGER_TYPE.
21228 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
21230 * config.gcc: Remove uclibc from arc target spec.
21232 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
21234 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
21236 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21239 * lto-wrapper.c (remove_option): New function.
21240 (merge_and_complain): Merge PIC/PIE options more realistically.
21242 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21244 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
21248 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
21249 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
21250 * config.in: Regenerate.
21251 * configure: Regenerate.
21252 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
21253 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
21254 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
21255 (TARGET_GASISR_PROLOGUES): ...target mask.
21256 * common/config/avr/avr-common.c
21257 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
21258 Set -mgas-isr-prologues.
21259 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
21260 INSERT_PASS_BEFORE for it.
21261 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
21262 * config/avr/avr.c (avr_option_override)
21263 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
21264 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
21265 (avr_attribute_table) <no_gccisr>: Add new function attribute.
21266 (avr_set_current_function) <is_no_gccisr>: Init machine field.
21267 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
21269 (make_avr_pass_pre_proep): New function.
21270 (emit_push_sfr) <treg>: Add argument to function and use it
21271 instead of TMP_REG.
21272 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
21273 and set machine->gasisr.yes.
21274 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
21275 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
21276 __gcc_isr.n_pushed to .L__stack_usage.
21277 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
21278 (avr_asm_final_postscan_insn): ...this new static function.
21279 * config/avr/avr.h (machine_function)
21280 <is_no_gccisr, use_L__stack_usage>: New fields.
21281 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
21282 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
21283 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
21284 (gasisr, *gasisr): New expander and insn.
21285 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21286 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
21287 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
21289 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
21291 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
21294 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21296 Move jump-tables out of .text again.
21299 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
21300 (ASM_OUTPUT_ADDR_VEC): New function.
21301 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
21302 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
21303 INSN_ADDRESSes as asm comment.
21304 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
21305 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
21306 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
21307 * config/avr/avr.md (*tablejump): Adjust comment.
21308 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
21309 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
21311 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
21312 (avr_output_addr_vec): New proto.
21313 (avr_log_t) <insn_addresses>: New field.
21315 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
21318 * config/i386/i386.c (ix86_function_arg_advance): Set
21319 outgoing_args_on_stack to true if there are outgoing arguments
21321 (ix86_function_arg): Likewise.
21322 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
21323 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
21324 * config/i386/i386.h (machine_function): Add
21325 outgoing_args_on_stack.
21327 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
21329 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
21330 supporting pthreds.
21331 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
21333 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
21335 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
21336 (REAL_H): Remove $(MACHMODE_H).
21337 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
21339 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
21340 $(MACHMODE_H) and double-int.h.
21341 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
21343 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
21346 2017-07-07 Andrew Pinski <apinski@cavium.com>
21348 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
21349 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
21351 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21353 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
21354 Add warning if GCC was not configured to link against a GLIBC that
21355 exports the hardware capability bits.
21356 (make_resolver_func): Make resolver function private and not a
21357 COMDAT function. Create the name with clone_function_name instead
21358 of make_unique_name.
21361 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
21362 correct operand in doing the split.
21364 2017-07-07 Carl Love <cel@us.ibm.com>
21366 * config/rs6000/rs6000-c: Add support for built-in function
21367 vector unsigned short vec_pack_to_short_fp32 (vector float,
21369 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
21370 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
21371 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
21372 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
21373 (convert_4f32_8i16): Add define_expand.
21374 * doc/extend.texi: Update the built-in documentation file for the
21375 new built-in function.
21377 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21379 * config/sparc/m8.md: New file.
21380 * config/sparc/sparc.md: Include m8.md.
21382 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21384 * config/sparc/sparc.opt: New option -mvis4b.
21385 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
21386 (sparc_option_override): Handle VIS4B.
21387 (enum sparc_builtins): Define
21388 SPARC_BUILTIN_DICTUNPACK{8,16,32},
21389 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
21390 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
21391 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
21392 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
21393 (check_constant_argument): New function.
21394 (sparc_vis_init_builtins): Define builtins
21395 __builtin_vis_dictunpack{8,16,32},
21396 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
21397 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
21398 __builtin_vis_fpcmpde{8,16,32}shl and
21399 __builtin_vis_fpcmpur{8,16,32}shl.
21400 (sparc_expand_builtin): Check that the constant operands to
21401 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
21402 constant and in range.
21403 * config/sparc/sparc-c.c (sparc_target_macros): Handle
21405 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
21406 (SPARC_IMM5_P): Likewise.
21407 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
21408 (enabled): Handle vis4b.
21409 (UNSPEC_DICTUNPACK): New unspec.
21410 (UNSPEC_FPCMPSHL): Likewise.
21411 (UNSPEC_FPUCMPSHL): Likewise.
21412 (UNSPEC_FPCMPDESHL): Likewise.
21413 (UNSPEC_FPCMPURSHL): Likewise.
21414 (cpu_feature): New CPU feature `vis4b'.
21415 (dictunpack{8,16,32}): New insns.
21416 (FPCSMODE): New mode iterator.
21417 (fpcscond): New code iterator.
21418 (fpcsucond): Likewise.
21419 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
21420 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
21421 (fpcmpde{8,16,32}{si,di}shl): Likewise.
21422 (fpcmpur{8,16,32}{si,di}shl): Likewise.
21423 * config/sparc/constraints.md: Define constraints `q' for unsigned
21424 2-bit integer constants and `t' for unsigned 5-bit integer
21426 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
21428 (imm5_operand_dictunpack16): Likewise.
21429 (imm5_operand_dictunpack32): Likewise.
21430 (imm2_operand): Likewise.
21431 * doc/invoke.texi (SPARC Options): Document -mvis4b.
21432 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
21433 ditunpack* and fpcmp*shl builtins.
21435 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21437 * config.gcc: Handle m8 in --with-{cpu,tune} options.
21438 * config.in: Add HAVE_AS_SPARC6 define.
21439 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
21441 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
21443 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
21444 (CPP_CPU_SPEC): Handle m8.
21445 (ASM_CPU_SPEC): Likewise.
21446 * config/sparc/sparc-opts.h (enum processor_type): Add
21448 * config/sparc/sparc.c (m8_costs): New struct.
21449 (sparc_option_override): Handle TARGET_CPU_m8.
21450 (sparc32_initialize_trampoline): Likewise.
21451 (sparc64_initialize_trampoline): Likewise.
21452 (sparc_issue_rate): Likewise.
21453 (sparc_register_move_cost): Likewise.
21454 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
21455 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
21456 (ASM_CPU64_DEFAULT_SPEC): Likewise.
21457 (CPP_CPU_SPEC): Handle M8.
21458 (ASM_CPU_SPEC): Likewise.
21459 (AS_M8_FLAG): Define.
21460 * config/sparc/sparc.md: Add m8 to the cpu attribute.
21461 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
21462 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
21464 * configure: Regenerate.
21465 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
21468 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21470 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
21472 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
21473 ("*movdi_insn_sp32"): Do not set v3pipe.
21474 ("*movsi_insn"): Likewise.
21475 ("*movdi_insn_sp64"): Likewise.
21476 ("*movsf_insn"): Likewise.
21477 ("*movdf_insn_sp32"): Likewise.
21478 ("*movdf_insn_sp64"): Likewise.
21479 ("*zero_extendsidi2_insn_sp64"): Likewise.
21480 ("*sign_extendsidi2_insn"): Likewise.
21481 ("*mov<VM32:mode>_insn"): Likewise.
21482 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21483 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21484 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21485 ("<vlop:code><VL:mode>3"): Likewise.
21486 ("*not_<vlop:code><VL:mode>3"): Likewise.
21487 ("*nand<VL:mode>_vis"): Likewise.
21488 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
21489 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
21490 ("one_cmpl<VL:mode>2"): Likewise.
21491 ("faligndata<VM64:mode>_vis"): Likewise.
21492 ("alignaddrsi_vis"): Likewise.
21493 ("alignaddrdi_vis"): Likweise.
21494 ("alignaddrlsi_vis"): Likewise.
21495 ("alignaddrldi_vis"): Likewise.
21496 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21497 ("bmaskdi_vis"): Likewise.
21498 ("bmasksi_vis"): Likewise.
21499 ("bshuffle<VM64:mode>_vis"): Likewise.
21500 ("cmask8<P:mode>_vis"): Likewise.
21501 ("cmask16<P:mode>_vis"): Likewise.
21502 ("cmask32<P:mode>_vis"): Likewise.
21503 ("pdistn<P:mode>_vis"): Likewise.
21504 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21506 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21508 * config/sparc/sparc.md ("subtype"): New insn attribute.
21509 ("*wrgsr_sp64"): Set insn subtype.
21510 ("*rdgsr_sp64"): Likewise.
21511 ("alignaddrsi_vis"): Likewise.
21512 ("alignaddrdi_vis"): Likewise.
21513 ("alignaddrlsi_vis"): Likewise.
21514 ("alignaddrldi_vis"): Likewise.
21515 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21516 ("fexpand_vis"): Likewise.
21517 ("fpmerge_vis"): Likewise.
21518 ("faligndata<VM64:mode>_vis"): Likewise.
21519 ("bshuffle<VM64:mode>_vis"): Likewise.
21520 ("cmask8<P:mode>_vis"): Likewise.
21521 ("cmask16<P:mode>_vis"): Likewise.
21522 ("cmask32<P:mode>_vis"): Likewise.
21523 ("fchksm16_vis"): Likewise.
21524 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
21525 ("fmean16_vis"): Likewise.
21526 ("fp<plusminus_insn>64_vis"): Likewise.
21527 ("<plusminus_insn>v8qi3"): Likewise.
21528 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21529 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
21530 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
21531 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
21532 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
21533 ("*movqi_insn"): Likewise.
21534 ("*movhi_insn"): Likewise.
21535 ("*movsi_insn"): Likewise.
21536 ("movsi_pic_gotdata_op"): Likewise.
21537 ("*movdi_insn_sp32"): Likewise.
21538 ("*movdi_insn_sp64"): Likewise.
21539 ("movdi_pic_gotdata_op"): Likewise.
21540 ("*movsf_insn"): Likewise.
21541 ("*movdf_insn_sp32"): Likewise.
21542 ("*movdf_insn_sp64"): Likewise.
21543 ("*zero_extendhisi2_insn"): Likewise.
21544 ("*zero_extendqihi2_insn"): Likewise.
21545 ("*zero_extendqisi2_insn"): Likewise.
21546 ("*zero_extendqidi2_insn"): Likewise.
21547 ("*zero_extendhidi2_insn"): Likewise.
21548 ("*zero_extendsidi2_insn_sp64"): Likewise.
21549 ("ldfsr"): Likewise.
21550 ("prefetch_64"): Likewise.
21551 ("prefetch_32"): Likewise.
21552 ("tie_ld32"): Likewise.
21553 ("tie_ld64"): Likewise.
21554 ("*tldo_ldub_sp32"): Likewise.
21555 ("*tldo_ldub1_sp32"): Likewise.
21556 ("*tldo_ldub2_sp32"): Likewise.
21557 ("*tldo_ldub_sp64"): Likewise.
21558 ("*tldo_ldub1_sp64"): Likewise.
21559 ("*tldo_ldub2_sp64"): Likewise.
21560 ("*tldo_ldub3_sp64"): Likewise.
21561 ("*tldo_lduh_sp32"): Likewise.
21562 ("*tldo_lduh1_sp32"): Likewise.
21563 ("*tldo_lduh_sp64"): Likewise.
21564 ("*tldo_lduh1_sp64"): Likewise.
21565 ("*tldo_lduh2_sp64"): Likewise.
21566 ("*tldo_lduw_sp32"): Likewise.
21567 ("*tldo_lduw_sp64"): Likewise.
21568 ("*tldo_lduw1_sp64"): Likewise.
21569 ("*tldo_ldx_sp64"): Likewise.
21570 ("*mov<VM32:mode>_insn"): Likewise.
21571 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21572 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21574 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21576 * config/sparc/sparc.md ("type"): New insn type viscmp.
21577 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
21579 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
21580 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
21581 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21582 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
21584 ("n7_vis_logical_11cycle"): Likewise.
21585 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
21586 * config/sparc/niagara2.md ("niag3_vis": Likewise.
21587 * config/sparc/niagara.md ("niag_vis"): Likewise.
21588 * config/sparc/ultra3.md ("us3_fga"): Likewise.
21589 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
21591 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21593 * config/sparc/sparc.md: New instruction type `bmask'.
21594 (bmaskdi_vis): Use the `bmask' type.
21595 (bmasksi_vis): Likewise.
21596 * config/sparc/ultra3.md (us3_array): Likewise.
21597 * config/sparc/niagara7.md (n7_array): Likewise.
21598 * config/sparc/niagara4.md (n4_array): Likewise.
21599 * config/sparc/niagara2.md (niag2_vis): Likewise.
21600 (niag3_vis): Likewise.
21601 * config/sparc/niagara.md (niag_vis): Likewise.
21603 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21605 * ipa-comdats.c: Remove optimize check from gate.
21606 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
21607 for functions not optimized.
21608 (ipa_fn_summary_read): Skip optimize check.
21609 (ipa_fn_summary_write): Likewise.
21610 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
21612 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
21614 (can_inline_edge_p): Check flag_pcc_struct_return for match.
21615 (check_callers): Give up on caller which is not optimized.
21616 (inline_small_functions): Likewise.
21617 (ipa_inline): Do not give up when not optimizing.
21618 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
21619 away unoptimizes cdtors.
21620 (whole_program_function_and_variable_visibility): Do
21621 ipa_discover_readonly_nonaddressable_vars in LTO mode.
21622 * ipa.c (process_references): Do not check optimize.
21623 (symbol_table::remove_unreachable_nodes): Update optimize check.
21624 (set_writeonly_bit): Update optimize check.
21625 (pass_ipa_cdtor_merge::gate): Do not check optimize.
21626 (pass_ipa_single_use::gate): Remove.
21628 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21630 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
21631 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
21632 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
21633 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
21634 permute_load, permute_store, adjust_extract, adjust_splat,
21635 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
21636 replace_swap_with_copy, dump_swap_insn_table,
21637 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
21638 recombine_lvx_pattern, recombine_stvx_pattern,
21639 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
21640 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
21641 to file rs6000-p8swap.c.
21642 * config/rs6000/rs6000-p8swap.c: New file.
21643 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
21644 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
21645 and rs6000*-*-* targets.
21647 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21649 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
21651 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21653 * lto-wrapper.c (merge_and_complain): Do not merge
21654 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
21655 fsigned_zeros, ftrapping_math, fwrapv.
21656 (append_compiler_options): Do not track these options.
21657 (append_linker_options): Likewie
21659 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21661 * cgraphunit.c (cgraph_node::finalize_function): When
21662 !flag_toplevel_reorde set no_reorder flag.
21663 (varpool_node::finalize_decl): Likewise.
21664 (symbol_table::compile): Drop no toplevel reorder path.
21666 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21668 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
21669 edges; zero probability is not better than uninitialized.
21671 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
21673 * asan.h (asan_sanitize_allocas_p): Declare.
21674 * asan.c (asan_sanitize_allocas_p): New function.
21675 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
21676 (handle_builtin_alloca): Likewise.
21677 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
21678 if !asan_sanitize_allocas_p.
21679 * params.def (asan-instrument-allocas): Add new option.
21680 * params.h (ASAN_PROTECT_ALLOCAS): Define.
21681 * opts.c (common_handle_option): Disable allocas sanitization for
21684 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
21686 * asan.c: Include gimple-fold.h.
21687 (get_last_alloca_addr): New function.
21688 (handle_builtin_stackrestore): Likewise.
21689 (handle_builtin_alloca): Likewise.
21690 (asan_emit_allocas_unpoison): Likewise.
21691 (get_mem_refs_of_builtin_call): Add new parameter, remove const
21692 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
21693 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
21694 (instrument_builtin_call): Pass gimple iterator to
21695 get_mem_refs_of_builtin_call.
21696 (last_alloca_addr): New global.
21697 * asan.h (asan_emit_allocas_unpoison): Declare.
21698 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
21699 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
21700 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
21701 if function calls alloca.
21702 * gimple-fold.c (replace_call_with_value): Remove static keyword.
21703 * gimple-fold.h (replace_call_with_value): Declare.
21704 * internal-fn.c: Include asan.h.
21705 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
21706 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
21708 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21710 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
21711 (C_SELFTEST_FLAGS): New.
21712 (CPP_SELFTEST_FLAGS): New.
21713 (SELFTEST_DEPS): New, from deps of s-selftest.
21714 (C_SELFTEST_DEPS): New, from deps of s-selftest.
21715 (CPP_SELFTEST_DEPS): New.
21716 (selftest): Add dependency on s-selftest-c++.
21717 (s-selftest): Rename to...
21718 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
21719 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
21720 than SELFTEST_FLAGS.
21721 (selftest-gdb): Rename to...
21722 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
21724 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
21725 (selftest-valgrind): Rename to...
21726 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
21728 (selftest-valgrind): Reintroduce as an alias for
21729 selftest-c-valgrind.
21730 (s-selftest-c++): New.
21731 (selftest-c++-gdb): New.
21732 (selftest-c++-valgrind): New.
21734 2017-07-06 Olivier Hainque <hainque@adacore.com>
21736 * gcc.c (process_command): When deciding if undefined variables
21737 should be ignored when processing specs, accept "gcc -v" as well.
21739 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21741 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
21742 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
21744 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21746 * config/arm/arm-cpus.in (armv8-r): Add new entry.
21747 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
21748 * config/arm/arm-tables.opt: Regenerate.
21749 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
21751 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
21753 2017-07-06 Carl Love <cel@us.ibm.com>
21755 * ChangeLog: Clean up from mid air collision
21757 2017-07-06 Carl Love <cel@us.ibm.com>
21759 * config/rs6000/rs6000-c.c: Add support for built-in functions
21760 vector signed int vec_subc (vector signed int, vector signed int);
21761 vector signed __int128 vec_subc (vector signed __int128,
21762 vector signed __int128);
21763 vector unsigned __int128 vec_subc (vector unsigned __int128,
21764 vector unsigned __int128);
21765 vector signed int vec_sube (vector signed int, vector signed int,
21766 vector signed int);
21767 vector unsigned int vec_sube (vector unsigned int,
21768 vector unsigned int,
21769 vector unsigned int);
21770 vector signed __int128 vec_sube (vector signed __int128,
21771 vector signed __int128,
21772 vector signed__int128);
21773 vector unsigned __int128 vec_sube (vector unsigned __int128,
21774 vector unsigned __int128,
21775 vector unsigned __int128);
21776 vector signed int vec_subec (vector signed int, vector signed int,
21777 vector signed int);
21778 vector unsigned int vec_subec (vector unsigned int,
21779 vector unsigned int,
21780 vector unsigned int);
21781 vector signed __int128 vec_subec (vector signed __int128,
21782 vector signed __int128,
21783 vector signed__int128);
21784 vector unsigned __int128 vec_subec (vector unsigned __int128,
21785 vector unsigned __int128,
21786 vector unsigned __int128);
21787 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
21788 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
21789 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
21790 BU_ALTIVEC_OVERLOAD_X definitions.
21791 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
21792 * doc/extend.texi: Update the built-in documentation file for the new
21793 built-in functions.
21795 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21798 * diagnostic-show-locus.c (layout::layout): Use start and finish
21799 spelling location for the start and finish of each range.
21800 * genmatch.c (linemap_client_expand_location_to_spelling_point):
21801 Add unused aspect param.
21802 * input.c (expand_location_1): Add "aspect" param, and use it
21803 to access the correct part of the location.
21804 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
21806 (expand_location_to_spelling_point): Likewise.
21807 (linemap_client_expand_location_to_spelling_point): Add "aspect"
21808 param, and pass it to expand_location_1.
21810 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
21812 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
21813 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
21814 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
21815 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
21816 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
21817 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
21818 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
21819 _mm_maskz_getmant_ss): New intrinsics.
21820 (__builtin_ia32_getexpss128_mask): Changed to ...
21821 __builtin_ia32_getexpss128_round ... this.
21822 (__builtin_ia32_getexpsd128_mask): Changed to ...
21823 __builtin_ia32_getexpsd128_round ... this.
21824 * config/i386/i386-builtin-types.def
21825 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
21826 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
21827 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
21828 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
21829 __builtin_ia32_getmantss_mask_round): New builtins.
21830 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
21831 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
21832 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
21833 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
21834 * config/i386/sse.md
21835 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
21836 avx512f_sgetexp<mode><mask_scalar_name>
21837 <round_saeonly_scalar_name> ... this.
21838 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
21839 %0, %1, %2<round_saeonly_op3>}): Changed to ...
21840 vgetexp<ssescalarmodesuffix>
21841 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21842 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
21843 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
21844 avx512f_vgetmant<mode><mask_scalar_name>
21845 <round_saeonly_scalar_name> ... this.
21846 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
21847 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
21848 vgetmant<ssescalarmodesuffix>
21849 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
21850 %0<mask_scalar_operand4>, %1, %2
21851 <round_saeonly_scalar_mask_op4>, %3} ... this.
21852 * config/i386/subst.md (mask_scalar_operand4,
21853 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
21854 round_saeonly_scalar_nimm_predicate): New subst attributes.
21856 2017-07-06 Julia Koval <julia.koval@intel.com>
21858 * config/i386/i386.c (ix86_erase_embedded_rounding):
21859 Remove code for old rounding pattern.
21861 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
21863 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
21865 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
21867 * doc/sourcebuild.texi (Test Directives, Variants of
21868 dg-require-support): Add documentation for dg-require-stack-check.
21870 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
21872 * config/i386/subst.md (mask_scalar, round_scalar,
21873 round_saeonly_scalar): New meta-templates.
21874 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
21875 round_scalar_mask_operand3, round_scalar_mask_op3,
21876 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
21877 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
21878 round_saeonly_scalar_constraint,
21879 round_saeonly_scalar_prefix): New subst attribute.
21880 * config/i386/sse.md
21881 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
21882 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
21883 <round_scalar_name> ... this.
21884 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
21885 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
21886 <round_scalar_name> ... this.
21887 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
21888 <sse>_vm<code><mode>3<mask_scalar_name>
21889 <round_saeonly_scalar_name> ... this.
21890 (v<plusminus_mnemonic><ssescalarmodesuffix>
21891 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21892 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21893 v<plusminus_mnemonic><ssescalarmodesuffix>
21894 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21895 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21896 (v<multdiv_mnemonic><ssescalarmodesuffix>
21897 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21898 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21899 v<multdiv_mnemonic><ssescalarmodesuffix>
21900 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21901 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21902 (v<maxmin_float><ssescalarmodesuffix>
21903 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
21904 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
21905 v<maxmin_float><ssescalarmodesuffix>
21906 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21907 %0<mask_scalar_operand3>, %1, %<iptr>2
21908 <round_saeonly_scalar_mask_op3>} ... this.
21910 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
21912 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
21913 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
21915 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
21916 Alan Hayward <alan.hayward@arm.com>
21917 David Sherwood <david.sherwood@arm.com>
21919 * combine.c (simplify_if_then_else): Remove "enum" before
21921 * compare-elim.c (can_eliminate_compare): Likewise.
21922 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
21924 (aarch64_lookup_simd_builtin_type): Likewise.
21925 (aarch64_simd_builtin_type): Likewise.
21926 (aarch64_init_simd_builtin_types): Likewise.
21927 (aarch64_simd_expand_args): Likewise.
21928 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
21930 (aarch64_reverse_mask): Likewise.
21931 (aarch64_simd_emit_reg_reg_move): Likewise.
21932 (aarch64_gen_adjusted_ldpstp): Likewise.
21933 (aarch64_ccmp_mode_to_code): Likewise.
21934 (aarch64_operands_ok_for_ldpstp): Likewise.
21935 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21936 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
21938 (aarch64_min_divisions_for_recip_mul): Likewise.
21939 (aarch64_reassociation_width): Likewise.
21940 (aarch64_get_condition_code_1): Likewise.
21941 (aarch64_simd_emit_reg_reg_move): Likewise.
21942 (aarch64_simd_attr_length_rglist): Likewise.
21943 (aarch64_reverse_mask): Likewise.
21944 (aarch64_operands_ok_for_ldpstp): Likewise.
21945 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21946 (aarch64_gen_adjusted_ldpstp): Likewise.
21947 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
21949 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
21950 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
21951 (arm_lookup_simd_builtin_type): Likewise.
21952 (arm_simd_builtin_type): Likewise.
21953 (arm_init_simd_builtin_types): Likewise.
21954 (arm_expand_builtin_args): Likewise.
21955 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
21956 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
21957 (ft32_setup_incoming_varargs): Likewise.
21958 (ft32_function_arg): Likewise.
21959 (ft32_function_arg_advance): Likewise.
21960 (ft32_pass_by_reference): Likewise.
21961 (ft32_arg_partial_bytes): Likewise.
21962 (ft32_valid_pointer_mode): Likewise.
21963 (ft32_addr_space_pointer_mode): Likewise.
21964 (ft32_addr_space_legitimate_address_p): Likewise.
21965 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
21967 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
21968 (ix86_emit_outlined_ms2sysv_restore): Likewise.
21969 (iamcu_alignment): Likewise.
21970 (canonicalize_vector_int_perm): Likewise.
21971 (ix86_noce_conversion_profitable_p): Likewise.
21972 (ix86_mpx_bound_mode): Likewise.
21973 (ix86_operands_ok_for_move_multiple): Likewise.
21974 * config/microblaze/microblaze-protos.h
21975 (microblaze_expand_conditional_branch_reg): Likewise.
21976 * config/microblaze/microblaze.c
21977 (microblaze_expand_conditional_branch_reg): Likewise.
21978 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
21980 (rs6000_reassociation_width): Likewise.
21981 (rs6000_invalid_binary_op): Likewise.
21982 (fusion_p9_p): Likewise.
21983 (emit_fusion_p9_load): Likewise.
21984 (emit_fusion_p9_store): Likewise.
21985 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
21987 (riscv_hard_regno_mode_ok_p): Likewise.
21988 (riscv_address_insns): Likewise.
21989 (riscv_split_symbol): Likewise.
21990 (riscv_legitimize_move): Likewise.
21991 (riscv_function_value): Likewise.
21992 (riscv_hard_regno_nregs): Likewise.
21993 (riscv_expand_builtin): Likewise.
21994 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
21995 (riscv_build_integer): Likewise.
21996 (riscv_split_integer): Likewise.
21997 (riscv_legitimate_constant_p): Likewise.
21998 (riscv_cannot_force_const_mem): Likewise.
21999 (riscv_regno_mode_ok_for_base_p): Likewise.
22000 (riscv_valid_base_register_p): Likewise.
22001 (riscv_valid_offset_p): Likewise.
22002 (riscv_valid_lo_sum_p): Likewise.
22003 (riscv_classify_address): Likewise.
22004 (riscv_legitimate_address_p): Likewise.
22005 (riscv_address_insns): Likewise.
22006 (riscv_load_store_insns): Likewise.
22007 (riscv_force_binary): Likewise.
22008 (riscv_split_symbol): Likewise.
22009 (riscv_force_address): Likewise.
22010 (riscv_legitimize_address): Likewise.
22011 (riscv_move_integer): Likewise.
22012 (riscv_legitimize_const_move): Likewise.
22013 (riscv_legitimize_move): Likewise.
22014 (riscv_address_cost): Likewise.
22015 (riscv_subword): Likewise.
22016 (riscv_output_move): Likewise.
22017 (riscv_canonicalize_int_order_test): Likewise.
22018 (riscv_emit_int_order_test): Likewise.
22019 (riscv_function_arg_boundary): Likewise.
22020 (riscv_pass_mode_in_fpr_p): Likewise.
22021 (riscv_pass_fpr_single): Likewise.
22022 (riscv_pass_fpr_pair): Likewise.
22023 (riscv_get_arg_info): Likewise.
22024 (riscv_function_arg): Likewise.
22025 (riscv_function_arg_advance): Likewise.
22026 (riscv_arg_partial_bytes): Likewise.
22027 (riscv_function_value): Likewise.
22028 (riscv_pass_by_reference): Likewise.
22029 (riscv_setup_incoming_varargs): Likewise.
22030 (riscv_print_operand): Likewise.
22031 (riscv_elf_select_rtx_section): Likewise.
22032 (riscv_save_restore_reg): Likewise.
22033 (riscv_for_each_saved_reg): Likewise.
22034 (riscv_register_move_cost): Likewise.
22035 (riscv_hard_regno_mode_ok_p): Likewise.
22036 (riscv_hard_regno_nregs): Likewise.
22037 (riscv_class_max_nregs): Likewise.
22038 (riscv_memory_move_cost): Likewise.
22039 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
22040 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
22041 (rl78_addr_space_address_mode): Likewise.
22042 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22044 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
22045 (rs6000_reassociation_width): Likewise.
22046 (rs6000_invalid_binary_op): Likewise.
22047 (fusion_p9_p): Likewise.
22048 (emit_fusion_p9_load): Likewise.
22049 (emit_fusion_p9_store): Likewise.
22050 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
22051 (ok_for_simple_move_operands): Likewise.
22052 (ok_for_simple_move_strict_operands): Likewise.
22053 (ok_for_simple_arith_logic_operands): Likewise.
22054 (visium_legitimize_reload_address): Likewise.
22055 (visium_select_cc_mode): Likewise.
22056 (output_cbranch): Likewise.
22057 (visium_split_double_move): Likewise.
22058 (visium_expand_copysign): Likewise.
22059 (visium_expand_int_cstore): Likewise.
22060 (visium_expand_fp_cstore): Likewise.
22061 * config/visium/visium.c (visium_pass_by_reference): Likewise.
22062 (visium_function_arg): Likewise.
22063 (visium_function_arg_advance): Likewise.
22064 (visium_libcall_value): Likewise.
22065 (visium_setup_incoming_varargs): Likewise.
22066 (visium_legitimate_constant_p): Likewise.
22067 (visium_legitimate_address_p): Likewise.
22068 (visium_legitimize_address): Likewise.
22069 (visium_secondary_reload): Likewise.
22070 (visium_register_move_cost): Likewise.
22071 (visium_memory_move_cost): Likewise.
22072 (prepare_move_operands): Likewise.
22073 (ok_for_simple_move_operands): Likewise.
22074 (ok_for_simple_move_strict_operands): Likewise.
22075 (ok_for_simple_arith_logic_operands): Likewise.
22076 (visium_function_value_1): Likewise.
22077 (rtx_ok_for_offset_p): Likewise.
22078 (visium_legitimize_reload_address): Likewise.
22079 (visium_split_double_move): Likewise.
22080 (visium_expand_copysign): Likewise.
22081 (visium_expand_int_cstore): Likewise.
22082 (visium_expand_fp_cstore): Likewise.
22083 (visium_split_cstore): Likewise.
22084 (visium_select_cc_mode): Likewise.
22085 (visium_split_cbranch): Likewise.
22086 (output_cbranch): Likewise.
22087 (visium_print_operand_address): Likewise.
22088 * expmed.c (flip_storage_order): Likewise.
22089 * expmed.h (emit_cstore): Likewise.
22090 (flip_storage_order): Likewise.
22091 * genrecog.c (validate_pattern): Likewise.
22092 * hsa-gen.c (gen_hsa_addr): Likewise.
22093 * internal-fn.c (expand_arith_overflow): Likewise.
22094 * ira-color.c (allocno_copy_cost_saving): Likewise.
22095 * lra-assigns.c (find_hard_regno_for_1): Likewise.
22096 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
22097 (process_invariant_for_inheritance): Likewise.
22098 * lra-eliminations.c (move_plus_up): Likewise.
22099 * omp-low.c (lower_oacc_reductions): Likewise.
22100 * simplify-rtx.c (simplify_subreg): Likewise.
22101 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
22102 (TARGET_CHKP_BOUND_MODE): Likewise..
22103 * targhooks.c (default_chkp_bound_mode): Likewise.
22104 (default_setup_incoming_vararg_bounds): Likewise.
22105 * targhooks.h (default_chkp_bound_mode): Likewise.
22106 (default_setup_incoming_vararg_bounds): Likewise.
22107 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
22108 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
22109 (have_whole_vector_shift): Likewise.
22110 * tree-vect-stmts.c (vectorizable_load): Likewise.
22111 * doc/tm.texi: Regenerate.
22113 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22115 Graceful degrade if Binutils PR21472 is not available.
22118 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
22119 .rodata in flash test fails.
22120 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
22121 * confgure: Regenerate.
22122 * config.in: Regenerate.
22123 * config/avr/avr.c (avr_asm_named_section)
22124 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
22125 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
22126 (avr_asm_init_sections): Same.
22128 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22130 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
22131 (fma<VH:mode>4_intrinsic): Likewise.
22132 (*fmsub<VCVTF:mode>4): Likewise.
22133 (*fmsub<VH:mode>4_intrinsic): Likewise.
22135 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22138 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
22139 Don't depend on "optimize > 0".
22140 (out_movhi_r_mr, out_movqi_mr_r): Same.
22141 (out_movhi_mr_r, out_movqi_r_mr): Same.
22142 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
22143 io_address_operand on "optimize > 0".
22145 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22147 * tree-loop-distribution.c: Add general explanantion on the pass.
22148 (generate_loops_for_partition): Mark distributed loop.
22149 (pg_add_dependence_edges): New parameter. Handle alias data
22150 dependence specially and record it in the parameter if asked.
22151 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
22152 (init_partition_graph_vertices, add_partition_graph_edge): New.
22153 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
22154 (free_partition_graph_vdata, build_partition_graph): New.
22155 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
22156 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
22157 (data_ref_segment_size, latch_dominated_by_data_ref): New.
22158 (compute_alias_check_pairs, version_loop_by_alias_check): New.
22159 (version_for_distribution_p, finalize_partitions): New.
22160 (distribute_loop): Handle alias data dependence specially. Factor
22161 out loop fusion code as functions and call these functions.
22163 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22165 * tree-loop-distribution.c (classify_partition): New parameter and
22166 better handle reduction statement.
22167 (rdg_build_partitions): Revise comment.
22168 (distribute_loop): Compute statements in all partitions and pass it
22169 to classify_partition.
22171 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22173 * tree-loop-distribution.c (enum partition_type): New.
22174 (struct partition): New field type.
22175 (partition_merge_into): Add parameter. Update partition type.
22176 (data_dep_in_cycle_p, update_type_for_merge): New functions.
22177 (build_rdg_partition_for_vertex): Compute partition type.
22178 (rdg_build_partitions): Dump partition type.
22179 (distribute_loop): Update calls to partition_merge_into.
22181 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22183 * tree-loop-distribution.c (struct ddr_hasher): New.
22184 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
22186 (classify_partition): Call get_data_dependence.
22187 (pg_add_dependence_edges): Ditto.
22188 (distribute_loop): Release data dependence hash table.
22190 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22192 * tree-loop-distribution.c (ref_base_address): Delete.
22193 (similar_memory_accesses): Rename ...
22194 (share_memory_accesses): ... to this. Check if partitions access
22195 the same memory reference.
22196 (distribute_loop): Call share_memory_accesses.
22198 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22200 * tree-loop-distribution.c (struct partition): New field recording
22201 its data reference.
22202 (partition_alloc, partition_free): Init and release data refs.
22203 (partition_merge_into): Merge data refs.
22204 (build_rdg_partition_for_vertex): Collect data refs for partition.
22205 (pg_add_dependence_edges): Change parameters from vector to bitmap.
22207 (distribute_loop): Remve data refs from vertice data of partition
22210 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22212 * tree-loop-distribution.c (params.h): Include header file.
22213 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
22214 (datarefs_vec): New global var.
22215 (create_rdg_vertices): Use datarefs_vec directly.
22216 (free_rdg): Don't free data references.
22217 (build_rdg): Update use. Don't free data references.
22218 (distribute_loop): Compute global variable for data references.
22219 Bail out if there are too many data references.
22221 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22223 * tree-loop-distribution.c (loop_nest): New global var.
22224 (build_rdg): Use loop directly, rather than loop nest.
22225 (pg_add_dependence_edges): Remove loop nest parameter. Use global
22227 (distribute_loop): Compute global variable loop nest. Update use.
22229 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22231 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
22232 (partition_merge_into): New parameter. Dump reason for fusion.
22233 (distribute_loop): Update use of partition_merge_into.
22235 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22237 * tree-loop-distribution.c (bb_top_order_index): New.
22238 (bb_top_order_index_size, bb_top_order_cmp): New.
22239 (stmts_from_loop): Use topological order.
22240 (pass_loop_distribution::execute): Compute and release topological
22241 order for basic blocks.
22243 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22245 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
22248 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22250 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
22251 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
22252 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
22253 * internal-fn.def (LOOP_DIST_ALIAS): New.
22254 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
22255 (fold_loop_internal_call): ... this.
22256 (vect_loop_dist_alias_call): New function.
22257 (set_uid_loop_bbs): Call fold_loop_internal_call.
22258 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
22261 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22264 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
22265 Require dead FLAGS_REG at the beginning of a peephole.
22267 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22270 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
22271 arguments in the call to __builtin_ia32_sbb_u32.
22272 (_subborrow_u64): Swap _X and _Y arguments in the call to
22273 __builtin_ia32_sbb_u64.
22275 2017-07-04 Jakub Jelinek <jakub@redhat.com>
22278 * tree-vrp.c (compare_assert_loc): Turn into a function template
22279 with stable template parameter. Only test if a->e is NULL,
22280 !a->e == !b->e has been verified already. Use e == NULL or
22281 e != NULL instead of e or ! e tests. If stable is true, don't use
22282 iterative_hash_expr, on the other side allow a or b or both NULL
22283 and sort the NULLs last.
22284 (process_assert_insertions): Sort using compare_assert_loc<false>
22285 instead of compare_assert_loc, later sort using
22286 compare_assert_loc<true> before calling process_assert_insertions_for
22287 in a loop. Use break instead of continue once seen NULL pointer.
22289 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22291 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
22292 Cortex-R7 and Cortex-R8 processors.
22294 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22296 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
22297 uninitialized while src is not.
22299 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
22301 * common/config/arm/arm-common.c: Adjust include path for
22303 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
22304 (arm-cpu.h): Create in build directory. Adjust dependency rules.
22305 (arm-cpu-data.h): Likewise.
22306 (arm-cpu-cdata.h): Likewise.
22307 * config/arm/arm-cpu.h: Delete.
22308 * config/arm/arm-cpu-cdata.h: Delete.
22309 * config/arm/arm-cpu-data.h: Delete.
22311 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
22313 * config/arm/arm-cpus.in (cortex-a55): New.
22314 (cortex-a75): Likewise.
22315 (cortex-a75.cortex-a55): Likewise.
22316 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
22318 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
22319 * config/arm/arm-cpu-cdata.h: Regenerate.
22320 * config/arm/arm-cpu-data.h: Regenerate.
22321 * config/arm/arm-cpu.h: Regenerate.
22322 * config/arm/arm-tables.opt: Regenerate.
22323 * config/arm/arm-tune.md: Regenerate.
22325 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22327 * haifa-sched.c (sched_create_recovery_edges): Update profile.
22329 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22331 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
22334 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
22336 PR tree-optimization/81292
22337 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
22338 full_string_p, also call adjust_related_strinfos if the adjustment
22339 is simple, otherwise invalidate related strinfos.
22341 2017-07-04 Martin Liska <mliska@suse.cz>
22344 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
22345 newly created variable as DECL_IGNORED_P.
22347 2017-07-04 Martin Liska <mliska@suse.cz>
22350 * ipa-inline.c (inline_small_functions):
22351 Use xstrdup_for_dump.
22353 2017-07-04 Tom de Vries <tom@codesourcery.com>
22355 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
22357 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22360 * config/darwin.c (darwin_function_switched_text_sections):
22363 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
22365 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
22367 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22369 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
22371 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22373 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
22374 min_profitable_iters, and th as inclusive lower bounds.
22375 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
22376 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
22377 for min_profitable_iters and min_profitable_estimate.
22378 (vect_transform_loop): Treat th as an inclusive lower bound.
22379 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22381 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22384 * config/darwin.c (darwin_function_switched_text_sections):
22385 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
22386 in two pieces, and suppress the use of buf.
22388 2017-07-03 Nathan Sidwell <nathan@acm.org>
22390 * hash-table.h (hash_table_mod1): Fix indentation.
22392 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22394 PR middle-end/81290
22395 * predict.c (force_edge_cold): Be more careful about propagation
22397 * profile-count.h (profile_probability::guessed,
22398 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
22400 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
22402 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
22404 * doc/invoke.texi (rcpc architecture extension): Document it.
22406 2017-07-03 Richard Biener <rguenther@suse.de>
22408 PR tree-optimization/60510
22409 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
22410 the scalar reduction PHI and use it.
22411 (vectorizable_reduction): Properly guard the single_defuse_cycle
22412 path for non-SLP reduction chains where we cannot use it.
22413 Rework reduc_def/index and vector type deduction. Rework
22414 vector operand gathering during reduction op code-gen.
22415 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
22416 chains dissolve the chain and leave it to non-SLP reduction
22419 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22421 * tree-data-ref.h (dr_alignment): Declare.
22422 * tree-data-ref.c (dr_alignment): New function.
22423 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
22424 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
22426 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
22428 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22430 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
22431 and base_misalignment fields.
22432 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
22433 * tree-data-ref.c: Include builtins.h.
22434 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
22435 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
22436 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
22437 * tree-vect-data-refs.c: Include tree-cfg.h.
22438 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
22439 fields instead of calculating an alignment here.
22440 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
22441 innermost_loop_behavior fields.
22443 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22445 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
22447 (DR_STEP_ALIGNMENT): New macro.
22448 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
22449 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
22450 (create_data_ref): Print it.
22451 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
22452 to tell whether the step preserves vector (mis)alignment.
22453 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22454 Move the check for an integer step and generalise to all INTEGER_CST.
22455 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
22456 Print the outer step alignment.
22458 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22460 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
22461 with offset_alignment.
22462 (DR_ALIGNED_TO): Delete.
22463 (DR_OFFSET_ALIGNMENT): New macro.
22464 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
22465 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
22466 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
22467 (create_data_ref): Likewise.
22468 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22469 (vect_analyze_data_refs): Likewise.
22470 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
22471 creating dummy innermost behavior.
22473 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22475 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
22476 with a "innermost_loop_behavior *" and refeence tree.
22477 * tree-data-ref.c (dr_analyze_innermost): Likewise.
22478 (create_data_ref): Update call accordingly.
22479 * tree-predcom.c (find_looparound_phi): Likewise.
22481 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22483 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
22484 fields with dr_wrt_vec_loop.
22485 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
22486 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
22487 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
22488 (vect_dr_behavior): New function.
22489 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22490 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
22491 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
22492 track whether the step preserves the misalignment.
22493 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22494 Use vect_dr_behavior.
22495 (vect_setup_realignment): Update call accordingly.
22496 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
22497 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
22498 call to vect_create_addr_base_for_vector_ref.
22499 (vect_create_cond_for_align_checks): Likewise.
22500 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
22501 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
22502 (vect_recog_mask_conversion_pattern): Likewise.
22503 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
22504 (new_stmt_vec_info): Remove redundant zeroing.
22506 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22508 * common/config/arm/arm-common.c (arm_be8_option): New function.
22509 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
22510 (ISA_ARMv6): Add isa_bit_be8.
22511 * config/arm/arm.h (arm_be8_option): Add prototype.
22512 (BE8_SPEC_FUNCTION): New define.
22513 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
22514 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
22515 (mlittle-endian): Similarly.
22516 (mbe8, mbe32): New options.
22517 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
22518 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
22520 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22522 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
22524 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22526 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
22527 (cleanup_tree_cfg_bb): Use it.
22528 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
22530 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
22532 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22535 * loop-doloop.c (add_test): Update profile.
22537 2017-07-03 Martin Liska <mliska@suse.cz>
22540 * sanopt.c (rewrite_usage_of_param): New function.
22541 (sanitize_rewrite_addressable_params): Likewise.
22542 (pass_sanopt::execute): Call rewrite_usage_of_param.
22544 2017-07-03 Richard Biener <rguenther@suse.de>
22546 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
22547 back to using VIEW_CONVERT_EXPR.
22549 2017-07-03 Martin Liska <mliska@suse.cz>
22552 * doc/extend.texi: Document when a resolver function is
22553 generated for target_clones.
22555 2017-07-03 Martin Liska <mliska@suse.cz>
22557 * asan.c (asan_emit_stack_protection): Unpoison just red zones
22558 and shadow memory of auto variables which are subject of
22559 use-after-scope sanitization.
22560 (asan_expand_mark_ifn): Add do set only when is_poison.
22562 2016-07-03 Richard Biener <rguenther@suse.de>
22564 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
22566 (vect_force_simple_reduction): Record reduction def -> phi mapping.
22567 (vectorizable_reduction): Perform reduction PHI creation when
22568 visiting a reduction PHI and adjust and simplify code generation
22569 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
22570 (vect_transform_loop): Visit reduction PHIs.
22571 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
22572 defs into the SLP tree.
22573 (vect_build_slp_tree): Reduction defs terminate the recursion.
22574 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
22576 (vect_get_vec_defs_for_stmt_copy): Export.
22577 (vect_get_vec_defs): Likewise.
22578 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
22580 (vect_get_vec_defs_for_stmt_copy): Declare.
22581 (vect_get_vec_defs): Likewise.
22583 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22585 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
22586 parameter with a "loop" parameter and use it instead of the
22587 loop containing DR_STMT. Don't check simple_iv when doing
22588 BB analysis. Describe the two analysis modes in the comment.
22590 2017-07-03 Tom de Vries <tom@codesourcery.com>
22592 PR tree-optimization/69468
22593 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
22594 (find_same_succ_bb): Handle ignore_edge_flags.
22596 2017-07-03 Tom de Vries <tom@codesourcery.com>
22598 PR tree-optimization/81192
22599 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
22601 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
22603 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
22605 2017-07-03 Tom de Vries <tom@codesourcery.com>
22607 PR tree-optimization/81192
22608 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
22609 BB_SAME_SUCC (bb) == NULL.
22611 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22613 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
22616 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22618 * dumpfile.c: Include profile-count.h
22619 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
22621 (insert_cond_bb): Update profile.
22622 * tree-cfg.h (insert_cond_bb): Update prototype.
22623 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
22624 * tree-dump.c: Do not include tree-cfg.
22626 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22628 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
22630 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22632 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
22635 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22637 * tree-complex.c (expand_complex_div_wide): update profile.
22639 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22640 Alan Hayward <alan.hayward@arm.com>
22641 David Sherwood <david.sherwood@arm.com>
22643 * Makefile.in (MACHMODE_H): Remove insn-modes.h
22644 (CORETYPES_H): New define.
22645 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
22646 (insn-modes-inline.h, s-modes-inline-h): New rules.
22647 (generated_files): Add insn-modes-inline.h.
22648 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
22649 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
22650 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
22651 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
22652 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
22653 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
22654 (build/gencodes.o, build/genconditions.o): Likewise.
22655 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
22656 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
22657 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
22658 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
22659 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
22660 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
22661 * coretypes.h: Include everything up to real.h for generators.
22662 Include insn-modes.h first. Include wide-int-print.h after
22663 wide-int.h. Include insn-modes-inline.h and then machmode.h.
22664 * machmode.h: Don't include insn-modes.h here.
22665 * function-tests.c: Remove includes of signop.h, machmode.h,
22666 double-int.h and wide-int.h.
22668 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
22670 * optc-save-gen.awk: Likewise.
22671 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
22672 * godump.c: Remove include of wide-int-print.h.
22673 * pretty-print.h: Likewise.
22674 * wide-int-print.cc: Likewise.
22675 * wide-int.cc: Likewise.
22676 * hash-map-tests.c: Remove include of signop.h.
22677 * hash-set-tests.c: Likewise.
22678 * rtl-tests.c: Likewise.
22679 * mkconfig.sh: Remove include of machmode.h.
22680 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
22682 (emit_insn_modes_inline_h): ...this new function. Emit the code
22683 into an insn-modes-inline.h header file, adding appropriate
22684 include guards and end comments.
22685 (emit_insn_modes_c_header): Remove include of machmode.h.
22686 (emit_min_insn_modes_c_header): Include coretypes.h rather than
22688 (main): Handle -i flag and call emit_insn_modes_inline_h when
22691 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22693 * tree-ssa-strlen.c (strinfo): Rename the length field to
22694 nonzero_chars. Add a full_string_p field.
22695 (compare_nonzero_chars, zero_length_string_p): New functions.
22696 (get_addr_stridx): Add an offset_out parameter.
22697 Use compare_nonzero_chars.
22698 (get_stridx): Update accordingly. Use compare_nonzero_chars.
22699 (new_strinfo): Update after above changes to strinfo.
22700 (set_endptr_and_length): Set full_string_p.
22701 (get_string_length): Update after above changes to strinfo.
22702 (unshare_strinfo): Update call to new_strinfo.
22703 (maybe_invalidate): Likewise.
22704 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
22705 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
22706 as a uhwi instead of an shwi. Update after above changes to
22707 strinfo and new_strinfo.
22708 (zero_length_string): Assert that chainsi contains full strings.
22709 Use zero_length_string_p. Update call to new_strinfo.
22710 (adjust_related_strinfos): Update after above changes to strinfo.
22711 Copy full_string_p from origsi.
22712 (adjust_last_stmt): Use zero_length_string_p.
22713 (handle_builtin_strlen): Update after above changes to strinfo and
22714 new_strinfo. Install the lhs as the string length if the previous
22715 entry didn't describe a full string.
22716 (handle_builtin_strchr): Update after above changes to strinfo
22718 (handle_builtin_strcpy): Likewise.
22719 (handle_builtin_strcat): Likewise.
22720 (handle_builtin_malloc): Likewise.
22721 (handle_pointer_plus): Likewise.
22722 (handle_builtin_memcpy): Likewise. Track nonzero characters
22723 that aren't necessarily followed by a nul terminator.
22724 (handle_char_store): Likewise.
22726 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22728 PR tree-optimization/80769
22729 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
22730 for malloc and calloc. Document the new invariant that all related
22731 strinfos have delayed lengths or none do.
22732 (verify_related_strinfos): Move earlier in file.
22733 (set_endptr_and_length): New function, split out from...
22734 (get_string_length): ...here. Also set the lengths of related
22736 (zero_length_string): Assert that chainsi has known (rather than
22738 (adjust_related_strinfos): Likewise.
22740 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22742 PR tree-optimization/81136
22743 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
22744 assert that two references with the same misalignment have the same
22745 compile-time misalignment if those compile-time misalignments
22748 2017-07-01 Andi Kleen <ak@linux.intel.com>
22750 * print-tree.c (print_node): Print all attributes.
22752 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22754 * cfg.c (scale_bbs_frequencies): New function.
22755 * cfg.h (scale_bbs_frequencies): Declare it.
22756 * cfgloopanal.c (single_likely_exit): Cleanup.
22757 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
22759 (scale_loop_profile): Likewise.
22760 (loop_version): Likewise.
22761 (create_empty_loop_on_edge): Update.
22762 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
22763 scale_loop_frequencies, scale_loop_profile, loopify,
22764 loop_version): Update prototypes.
22765 * modulo-sched.c (sms_schedule): Update.
22766 * predict.c (unlikely_executed_edge_p): Also check probability.
22767 (probably_never_executed_edge_p): Fix typo.
22768 * tree-if-conv.c (version_loop_for_if_conversion): Update.
22769 * tree-parloops.c (gen_parallel_loop): Update.
22770 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
22771 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22772 * tree-ssa-loop-split.c (split_loop): Update.
22773 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
22774 * tree-vect-loop-manip.c (vect_do_peeling): Update.
22775 (vect_loop_versioning): Update.
22776 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22778 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22780 * trans-mem.c (split_bb_make_tm_edge): Update profile.
22782 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22784 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
22785 to keep profile consistent.
22787 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22789 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
22790 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
22791 * profile-count.h (max_safe_multiplier): Make unsigned.
22792 (profile_count::guessed_zero): New.
22794 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22796 * bb-reorder.c (fix_up_crossing_landing_pad,
22797 fix_crossing_conditional_branches): Use make_single_succ_edge
22798 to keep profile consistent.
22800 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22802 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
22805 2017-07-01 Jakub Jelinek <jakub@redhat.com>
22808 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
22809 the right scopes, make sure cond_jump isn't preserved between multiple
22810 iterations. Search for fallthru edge whenever there are 3+ edges and
22811 use find_fallthru_edge for it.
22813 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22815 Patch by Alexander Monakov <amonakov@ispras.ru>
22816 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
22817 probabilities consistently.
22819 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22821 * pa.c (pa_expand_compare_and_swap_loop): Update call of
22822 emit_cmp_and_jump_insns.
22824 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22827 * tree-inline.c (expand_call_inline): Combine profile statuses.
22829 2017-06-30 Andrew Pinski <apinski@cavium.com>
22831 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
22832 fold_stmt returned true.
22834 2017-06-30 Nathan Sidwell <nathan@acm.org>
22836 * ggc.h (empty_string): Delete.
22837 * cfgexpand.c (expand_asm_stmt): Use plain "".
22838 * optabs.c (expand_asm_memory_barrier): Likewise.
22839 * stringpool.c (empty_string): Delete.
22840 (digit_vector, digit_string): Delete.
22841 (ggc_alloc_string): Use plain "", don't optimize single digit
22842 strings. Use ggc_alloc_atomic.
22844 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
22846 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
22847 comparison set and one other set, use the cost of the non-comparison
22850 2017-06-30 Nathan Sidwell <nathan@acm.org>
22852 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
22855 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
22857 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
22858 loops. Remove now unneeded calls to gimple_switch_set_label() that
22859 just set removed labels to NULL_TREE.
22861 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
22863 * tree-ssanames.c (set_range_info_raw): Abstract from ...
22864 (set_range_info): ...here. Only call set_range_info_raw if domain
22866 (set_nonzero_bits): Call set_range_info_raw.
22867 * tree-ssanames.h (set_range_info_raw): New.
22869 2017-06-30 Jakub Jelinek <jakub@redhat.com>
22872 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
22873 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
22874 of nonimmediate_operand and <store_mask_constraint> instead of m
22875 for the input operand. For V8FI iterator, always split if input
22876 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
22877 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
22878 <store_mask_predicate> instead of register_operand and
22879 <store_mask_constraint> instead of v for the input operand. Make
22880 sure both operands aren't MEMs for if not <mask_applied>.
22882 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
22884 * lto-wrapper.c (copy_file) Close both file descriptors before
22887 2017-06-30 Martin Liska <mliska@suse.cz>
22890 * multiple_target.c (create_dispatcher_calls): Make ifunc
22891 also for function that don't have calls or are not referenced.
22893 2017-06-30 Richard Biener <rguenther@suse.de>
22895 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
22896 analyze the first scalar stmt. Move vector type computation
22897 for the BB case here from ...
22898 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
22899 live operation processing in the SLP case properly.
22901 2017-06-30 Richard Biener <rguenther@suse.de>
22903 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
22905 2017-06-30 Martin Liska <mliska@suse.cz>
22908 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
22909 before BUILT_IN_UNWIND_RESUME when ASAN is used.
22911 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
22913 * doc/invoke.texi (AArch64): Add missing options and remove redundant
22916 2017-06-30 Richard Biener <rguenther@suse.de>
22918 PR tree-optimization/81249
22919 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
22920 condition reduction result to original scalar type.
22922 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22924 * profile-count.h (enum profile_quality): Fix typos and whitespace
22927 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22929 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
22930 type for branch probabilities.
22932 2017-06-29 Julian Brown <julian@codesourcery.com>
22933 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22935 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
22936 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
22937 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
22938 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
22940 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22942 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
22943 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
22944 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
22945 CC usage from generic code to here.
22946 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
22947 CC usage into the target macros.
22949 2017-06-29 Maya Rashish <coypu@sdf.org>
22951 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
22954 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22956 * arm/arm-builtins.c: Include profile-count.h
22957 * except.c (sjlj_emit_function_enter): Use
22958 profile_probability::unlikely.
22960 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22962 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
22963 and tocrel_offset be pointer args rather than implicitly using
22965 (legitimate_constant_pool_address_p, rs6000_emit_move,
22966 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
22967 tocrel_offset and use in toc_relative_expr_p call.
22968 (print_operand, print_operand_address): Use static tocrel_base_oac
22969 and tocrel_offset_oac.
22970 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
22973 2017-06-29 Maya Rashish <coypu@sdf.org>
22975 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
22977 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
22979 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
22980 objects, take into account only the alignment of 'op0' and 'mode1' if
22983 2017-06-29 Steve Ellcey <sellcey@cavium.com>
22985 * ccmp.c (ccmp_tree_comparison_p): New function.
22986 (ccmp_candidate_p): Update to use above function.
22987 (get_compare_parts): New function.
22988 (expand_ccmp_next): Update to use new functions.
22989 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
22991 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
22992 take mode as argument.
22993 * ccmp.h (expand_ccmp_expr): Add mode as argument.
22994 * expr.c (expand_expr_real_1): Pass mode as argument.
22996 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
22998 * combine.c (combine_instructions): Print insns to dump_file, together
23001 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23003 * asan.c (asan_emit_stack_protection): Update.
23004 (create_cond_insert_point): Update.
23005 * auto-profile.c (afdo_propagate_circuit): Update.
23006 * basic-block.h (struct edge_def): Turn probability to
23007 profile_probability.
23008 (EDGE_FREQUENCY): Update.
23009 * bb-reorder.c (find_traces_1_round): Update.
23010 (better_edge_p): Update.
23011 (sanitize_hot_paths): Update.
23012 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
23013 (make_single_succ_edge): Update.
23014 (check_bb_profile): Update.
23015 (dump_edge_info): Update.
23016 (update_bb_profile_for_threading): Update.
23017 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
23018 probabilitycount to 0.
23019 * cfgbuild.c (compute_outgoing_frequencies): Update.
23020 * cfgcleanup.c (try_forward_edges): Update.
23021 (outgoing_edges_match): Update.
23022 (try_crossjump_to_edge): Update.
23023 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
23024 (expand_gimple_tailcall): Update.
23025 (construct_init_block): Use make_single_succ_edge.
23026 (construct_exit_block): Use make_single_succ_edge.
23027 * cfghooks.c (verify_flow_info): Update.
23028 (redirect_edge_succ_nodup): Update.
23029 (split_edge): Update.
23030 (account_profile_record): Update.
23031 * cfgloopanal.c (single_likely_exit): Update.
23032 * cfgloopmanip.c (scale_loop_profile): Update.
23033 (set_zero_probability): Remove.
23034 (duplicate_loop_to_header_edge): Update.
23035 * cfgloopmanip.h (loop_version): Update prototype.
23036 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
23037 (force_nonfallthru_and_redirect): Update.
23038 (update_br_prob_note): Update.
23039 (rtl_verify_edges): Update.
23040 (purge_dead_edges): Update.
23041 (rtl_lv_add_condition_to_bb): Update.
23042 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
23043 * cgraphunit.c (init_lowered_empty_function): Update.
23044 (cgraph_node::expand_thunk): Update.
23045 * cilk-common.c: Include profile-count.h
23046 * dojump.c (inv): Remove.
23047 (jumpifnot): Update.
23048 (jumpifnot_1): Update.
23049 (do_jump_1): Update.
23051 (do_jump_by_parts_greater_rtx): Update.
23052 (do_compare_rtx_and_jump): Update.
23053 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
23054 do_jump_1. do_compare_rtx_and_jump): Update prototype.
23055 * dwarf2cfi.c: Include profile-count.h
23056 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
23057 (sjlj_emit_dispatch_table): Likewise.
23058 * explow.c: Include profile-count.h
23059 * expmed.c (emit_store_flag_force): Update.
23060 (do_cmp_and_jump): Update.
23061 * expr.c (compare_by_pieces_d::generate): Update.
23062 (compare_by_pieces_d::finish_mode): Update.
23063 (emit_block_move_via_loop): Update.
23064 (store_expr_with_bounds): Update.
23065 (store_constructor): Update.
23066 (expand_expr_real_2): Update.
23067 (expand_expr_real_1): Update.
23068 * expr.h (try_casesi, try_tablejump): Update prototypes.
23069 * gimple-pretty-print.c (dump_probability): Update.
23070 (dump_profile): New.
23071 (dump_gimple_label): Update.
23072 (dump_gimple_bb_header): Update.
23073 * graph.c (draw_cfg_node_succ_edges): Update.
23074 * hsa-gen.c (convert_switch_statements): Update.
23075 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
23076 (find_if_case_1): Update.
23077 (find_if_case_2): Update.
23078 * internal-fn.c (expand_arith_overflow_result_store): Update.
23079 (expand_addsub_overflow): Update.
23080 (expand_neg_overflow): Update.
23081 (expand_mul_overflow): Update.
23082 (expand_vector_ubsan_overflow): Update.
23083 * ipa-cp.c (good_cloning_opportunity_p): Update.
23084 * ipa-split.c (split_function): Use make_single_succ_edge.
23085 * ipa-utils.c (ipa_merge_profiles): Update.
23086 * loop-doloop.c (add_test): Update.
23087 (doloop_modify): Update.
23088 * loop-unroll.c (compare_and_jump_seq): Update.
23089 (unroll_loop_runtime_iterations): Update.
23090 * lra-constraints.c (lra_inheritance): Update.
23091 * lto-streamer-in.c (input_cfg): Update.
23092 * lto-streamer-out.c (output_cfg): Update.
23093 * mcf.c (adjust_cfg_counts): Update.
23094 * modulo-sched.c (sms_schedule): Update.
23095 * omp-expand.c (expand_omp_for_init_counts): Update.
23096 (extract_omp_for_update_vars): Update.
23097 (expand_omp_ordered_sink): Update.
23098 (expand_omp_for_ordered_loops): Update.
23099 (expand_omp_for_generic): Update.
23100 (expand_omp_for_static_nochunk): Update.
23101 (expand_omp_for_static_chunk): Update.
23102 (expand_cilk_for): Update.
23103 (expand_omp_simd): Update.
23104 (expand_omp_taskloop_for_outer): Update.
23105 (expand_omp_taskloop_for_inner): Update.
23106 * omp-simd-clone.c (simd_clone_adjust): Update.
23107 * optabs.c (expand_doubleword_shift): Update.
23108 (expand_abs): Update.
23109 (emit_cmp_and_jump_insn_1): Update.
23110 (expand_compare_and_swap_loop): Update.
23111 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
23112 * predict.c (predictable_edge_p): Update.
23113 (edge_probability_reliable_p): Update.
23114 (set_even_probabilities): Update.
23115 (combine_predictions_for_insn): Update.
23116 (combine_predictions_for_bb): Update.
23117 (propagate_freq): Update.
23118 (estimate_bb_frequencies): Update.
23119 (force_edge_cold): Update.
23120 * profile-count.c (profile_count::dump): Add missing space into dump.
23121 (profile_count::debug): Add newline.
23122 (profile_count::differs_from_p): Explicitly convert to unsigned.
23123 (profile_count::stream_in): Update.
23124 (profile_probability::dump): New member function.
23125 (profile_probability::debug): New member function.
23126 (profile_probability::differs_from_p): New member function.
23127 (profile_probability::differs_lot_from_p): New member function.
23128 (profile_probability::stream_in): New member function.
23129 (profile_probability::stream_out): New member function.
23130 * profile-count.h (profile_count_quality): Rename to ...
23131 (profile_quality): ... this one.
23132 (profile_probability): New.
23133 (profile_count): Update.
23134 * profile.c (compute_branch_probabilities): Update.
23135 * recog.c (peep2_attempt): Update.
23136 * sched-ebb.c (schedule_ebbs): Update.
23137 * sched-rgn.c (find_single_block_region): Update.
23138 (compute_dom_prob_ps): Update.
23139 (schedule_region): Update.
23140 * sel-sched-ir.c (compute_succs_info): Update.
23141 * stmt.c (struct case_node): Update.
23142 (do_jump_if_equal): Update.
23143 (get_outgoing_edge_probs): Update.
23144 (conditional_probability): Update.
23145 (emit_case_dispatch_table): Update.
23146 (expand_case): Update.
23147 (expand_sjlj_dispatch_table): Update.
23148 (emit_case_nodes): Update.
23149 * targhooks.c: Update.
23150 * tracer.c (better_p): Update.
23151 (find_best_successor): Update.
23152 * trans-mem.c (expand_transaction): Update.
23153 * tree-call-cdce.c: Update.
23154 * tree-cfg.c (gimple_split_edge): Upate.
23155 (move_sese_region_to_fn): Upate.
23156 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
23157 * tree-eh.c (lower_resx): Upate.
23158 (cleanup_empty_eh_move_lp): Upate.
23159 * tree-if-conv.c (version_loop_for_if_conversion): Update.
23160 * tree-inline.c (copy_edges_for_bb): Update.
23161 (copy_cfg_body): Update.
23162 * tree-parloops.c (gen_parallel_loop): Update.
23163 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
23164 (gimple_gen_time_profiler): Update.
23165 * tree-ssa-dce.c (remove_dead_stmt): Update.
23166 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
23167 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
23168 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
23169 (unloop_loops): Update.
23170 (try_peel_loop): Update.
23171 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23172 * tree-ssa-loop-split.c (connect_loops): Update.
23173 (split_loop): Update.
23174 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
23175 (hoist_guard): Update.
23176 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
23177 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
23178 (value_replacement): Update.
23179 * tree-ssa-reassoc.c (branch_fixup): Update.
23180 * tree-ssa-tail-merge.c (replace_block_by): Update.
23181 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
23182 (create_edge_and_update_destination_phis): Update.
23183 (compute_path_counts): Update.
23184 (recompute_probabilities): Update.
23185 (update_joiner_offpath_counts): Update.
23186 (freqs_to_counts_path): Update.
23187 (duplicate_thread_path): Update.
23188 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
23189 (struct switch_conv_info): Update.
23190 (gen_inbound_check): Update.
23191 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23192 (vect_do_peeling): Update.
23193 (vect_loop_versioning): Update.
23194 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23195 (optimize_mask_stores): Update.
23196 * ubsan.c (ubsan_expand_null_ifn): Update.
23197 * value-prof.c (gimple_divmod_fixed_value): Update.
23198 (gimple_divmod_fixed_value_transform): Update.
23199 (gimple_mod_pow2): Update.
23200 (gimple_mod_pow2_value_transform): Update.
23201 (gimple_mod_subtract): Update.
23202 (gimple_mod_subtract_transform): Update.
23203 (gimple_ic): Update.
23204 (gimple_stringop_fixed_value): Update.
23205 (gimple_stringops_transform): Update.
23206 * value-prof.h: Update.
23208 2017-06-29 Carl Love <cel@us.ibm.com>
23210 * config/rs6000/rs6000-c.c: Add support for built-in functions
23211 vector signed int vec_signed (vector float);
23212 vector signed long long vec_signed (vector double);
23213 vector signed int vec_signed2 (vector double, vector double);
23214 vector signed int vec_signede (vector double);
23215 vector signed int vec_signedo (vector double);
23216 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
23217 instruction generator.
23218 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23219 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
23220 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
23222 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
23223 vunsignede_v2df): Add define_expands.
23224 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
23225 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
23226 VEC_UNSIGNEDO): Add definitions.
23227 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23228 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
23229 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
23230 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
23231 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
23232 * config/rs6000/altivec.h (vec_signed, vec_signed2,
23233 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
23234 vec_unsignede, vec_unsignedo): Add builtin defines.
23235 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
23237 * doc/extend.texi: Update the built-in documentation file for the
23238 new built-in functions.
23240 2017-06-29 Richard Biener <rguenther@suse.de>
23242 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
23243 reduction chains to LOOP_VINFO_REDUCTIONS.
23244 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
23245 SLP reductions after processing reduction chains.
23247 2017-06-29 Nathan Sidwell <nathan@acm.org>
23249 * builtins.c (fold_builtin_FUNCTION): Use
23250 lang_hooks.decl_printable_name.
23252 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
23254 PR middle-end/81194
23255 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
23256 with only one label.
23257 * stmt.c (expand_case): Assert NCASES is greater than one.
23259 2017-06-29 Richard Biener <rguenther@suse.de>
23261 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
23263 (group_case_labels): Likewise.
23264 (find_taken_edge): Push sanity checking on val to workers...
23265 (find_taken_edge_cond_expr): ... here
23266 (find_taken_edge_switch_expr): ... and here, handle cases
23267 with just a default label.
23268 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
23269 (group_case_labels): Likewise.
23270 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
23271 group_case_labels does anything cleanup the CFG again.
23273 2017-06-29 Bin Cheng <bin.cheng@arm.com>
23275 PR tree-optimization/81196
23276 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
23277 exit condition comparing two IVs.
23279 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
23281 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
23282 profile to the dummy entry at the end of the list of architectures.
23283 * config/arm/arm-cpu-cdata.h: Regenerated.
23285 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23286 Michael Collison <michael.collison@arm.com>
23289 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
23291 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
23292 (*aarch64_reg_<mode>3_minus_mask): New pattern.
23293 (*aarch64_<optab>_reg_di3_mask2): New pattern.
23294 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
23295 of shift when the shift amount is masked with constant equal to
23296 the size of the mode.
23297 * config/aarch64/predicates.md (subreg_lowpart_operator): New
23300 2017-06-29 Martin Liska <mliska@suse.cz>
23302 * config/i386/i386.opt: Change range from [1,5] to [0,5].
23304 2017-06-29 Yury Gribov <tetra2005@gmail.com>
23307 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
23310 (edge_growth_cache_entry::edge_growth_cache_entry): New
23312 (reset_edge_growth_cache): Update to use constructor.
23314 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23316 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
23317 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23318 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23320 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
23322 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
23323 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
23325 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
23327 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
23328 (*-linux-uclibc*): Add t-uclibc tmake_file.
23329 * config/t-musl: New.
23330 * config/t-uclibc: New.
23332 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
23334 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
23336 (gen_comm_data): Emit architectural setting of arch_prof.
23337 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
23339 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
23340 (armv8-m.base, armv8-m.main): Likewise.
23341 * arm-protos.h (arm_build_target): Add profile field.
23342 (arch_option): Likewise.
23343 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
23345 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
23346 arm_active_target.profile.
23348 2017-06-28 Richard Biener <rguenther@suse.de>
23350 PR middle-end/81227
23351 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
23352 TYPE_OVERFLOW_WRAPS.
23353 * match.pd (negate_expr_p): Likewise.
23354 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
23355 fold_build2, not fold_binary.
23357 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23359 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
23360 Convert memory address to Pmode.
23361 (aarch64_print_operand): Assert MEM operands are always Pmode.
23363 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23366 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
23367 Remove redundant if.
23368 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
23369 * config/arm/aarch-common-protos.h
23370 (aarch_forward_to_shift_is_not_shifted_re): Remove.
23371 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
23373 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
23376 * multiple_target.c (create_dispatcher_calls): Set the default
23377 clone to be static, not public.
23379 2017-06-28 Richard Biener <rguenther@suse.de>
23381 * tree-vect-loop.c (vectorizable_reduction): Move special
23382 cond reduction IV var creation ...
23383 (vect_create_epilog_for_reduction): ... here. Remove induction_index
23384 parameter. Use STMT_VINFO_VECTYPE.
23385 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
23388 2017-06-28 Martin Liska <mliska@suse.cz>
23391 * ipa-visibility.c (non_local_p): Handle visibility.
23393 2017-06-28 Martin Liska <mliska@suse.cz>
23396 * common.opt: Add IntegerRange to various options.
23397 * opt-functions.awk (integer_range_info): New function.
23398 * optc-gen.awk: Add integer_range_info to cl_options struct.
23399 * opts-common.c (decode_cmdline_option): Handle
23400 CL_ERR_INT_RANGE_ARG.
23401 (cmdline_handle_error): Likewise.
23402 * opts.c (print_filtered_help): Show valid interval in
23403 when --help is provided.
23404 * opts.h (struct cl_option): Add range_min and range_max fields.
23405 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
23407 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
23409 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
23410 (x * C EQ/NE y * C): New transformation.
23412 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
23414 * genmultilib (combination_space): Accept '+' in option names.
23416 2017-06-28 Martin Liska <mliska@suse.cz>
23419 * asan.c (instrument_derefs): Bail out inner references
23420 that are hard register variables.
23422 2017-06-28 Jakub Jelinek <jakub@redhat.com>
23425 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
23426 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
23428 2017-06-28 Richard Biener <rguenther@suse.de>
23430 * tree-vectorizer.h (vect_get_vec_defs): Remove.
23431 (vect_get_slp_defs): Adjust.
23432 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
23434 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
23436 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
23437 get_initial_defs_for_reduction instead of vect_get_vec_defs.
23438 (vectorizable_reduction): Adjust.
23439 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
23441 (vect_get_slp_defs): Likewise.
23442 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
23443 (vectorizable_bswap): Adjust.
23444 (vectorizable_call): Likewise.
23445 (vectorizable_conversion): Likewise.
23446 (vectorizable_assignment): Likewise.
23447 (vectorizable_shift): Likewise.
23448 (vectorizable_operation): Likewise.
23449 (vectorizable_store): Likewise.
23450 (vectorizable_condition): Likewise.
23451 (vectorizable_comparison): Likewise.
23453 2017-06-28 Michael Collison <michael.collison@arm.com>
23456 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
23458 (arm_gen_movmemqi): Removed unused variable 'i'.
23459 Convert 'for' loop into 'while' loop.
23460 (arm_expand_prologue): Remove last unnecessary set of insn.
23461 (thumb_pop): Remove unused variable 'pushed_words'.
23462 (thumb_exit): Remove last unnecessary set of regs_to_pop.
23464 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23466 * config/s390/predicates.md: Use s390_rel_address_ok_p.
23467 * config/s390/s390-protos.h: Add prototype of
23468 s390_rel_address_ok_p.
23469 * config/s390/s390.c (s390_got_symbol): New function.
23470 (s390_rel_address_ok_p): New function.
23471 (legitimize_pic_address): Use s390_rel_address_ok_p.
23472 (s390_load_got): Use s390_got_symbol.
23473 (s390_option_override): Issue error if
23474 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
23475 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
23477 * config/s390/s390.opt: New option mpic-data-is-text-relative.
23479 2017-06-27 Andrew Pinski <apinski@cavium.com>
23481 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
23482 (X * copysign (1.0, X)): New pattern.
23483 (X * copysign (1.0, -X)): New pattern.
23484 (copysign (-1.0, CST)): New pattern.
23486 2017-06-27 Joseph Myers <joseph@codesourcery.com>
23488 * genmultilib (combination_space): Remove variable.
23489 Validate reuse rules against regular expression for any sequence
23490 of multilib options in any order.
23492 2017-06-27 Michael Collison <michael.collison@arm.com>
23494 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
23495 call aarch64_split_simd_combine.
23496 * (aarch64_combine_internal<mode>): Delete pattern.
23497 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
23498 Allow register and subreg operands.
23500 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23502 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
23503 specific need, just fallback on defaults.
23504 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
23506 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23507 Olivier Hainque <hainque@adacore.com>
23509 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
23511 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
23512 targets. Pick a default if no particular attempt applied.
23513 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
23516 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23518 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
23519 (x86_64-wrs-vxworks7): Likewise.
23521 2017-06-27 Marek Polacek <polacek@redhat.com>
23524 * ubsan.c (instrument_null): Check get_base_address's result for null.
23526 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
23528 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
23530 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
23532 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
23533 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
23534 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
23535 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
23536 New function types.
23537 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
23538 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
23539 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
23540 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
23541 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
23542 BUILT_IN_FEUPDATEENV): New builtins.
23543 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
23544 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
23545 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
23546 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
23548 (builtin_structptr_types): Adjust size.
23549 * tree.c (builtin_structptr_types): Add four entries.
23551 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23552 Olivier Hainque <hainque@adacore.com>
23554 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
23555 (TLS_SYM): New local macro, forcing reference to __tls__ on
23556 link command lines for VxWorks 7 RTPs, triggering initialization
23558 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
23559 OS features TLS support, true for RTPs on VxWorks 7.
23560 * config/vxworks.c (vxworks_override_options): Setup emutls
23563 2017-06-27 Jakub Jelinek <jakub@redhat.com>
23565 * predict.c (test_prediction_value_range): Use -1U instead of -1
23566 to avoid narrowing conversion warning.
23567 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
23568 to avoid narrowing conversion warning.
23569 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
23571 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
23573 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23575 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
23576 64bit configurations.
23577 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
23578 (SIZE_TYPE): Likewise.
23579 * config/vxworks.c (vxworks_emutls_var_fields): Use
23580 long_unsigned_type_node instead of unsigned_type_node as the offset
23581 field type, which is "pointer" mode in emutls.c.
23583 2017-06-27 Jakub Jelinek <jakub@redhat.com>
23586 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
23588 PR middle-end/81207
23589 * gimple-fold.c (replace_call_with_call_and_fold): Handle
23590 gimple_vuse copying separately from gimple_vdef copying.
23592 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23594 * value-prof.c (free_hist): Remove call to memset and the enclosing if
23597 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
23598 Olivier Hainque <hainque@adacore.com>
23600 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
23601 for all vxworks7 targets.
23602 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
23603 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
23604 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
23605 variations for VX6/VX7 and 32/64bits later on in ...
23606 (VXWORKS_LIB_SPEC): Leverage new macros.
23607 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
23608 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
23610 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
23612 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
23613 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
23615 2017-06-26 Carl Love <cel@us.ibm.com>
23617 * config/rs6000/rs6000-c.c: Add support for built-in functions
23618 vector bool char vec_reve (vector bool char);
23619 vector signed char vec_reve (vector signed char);
23620 vector unsigned char vec_reve (vector unsigned char);
23621 vector bool int vec_reve (vector bool int);
23622 vector signed int vec_reve (vector signed int);
23623 vector unsigned int vec_reve (vector unsigned int);
23624 vector bool long long vec_reve (vector bool long long);
23625 vector signed long long vec_reve (vector signed long long);
23626 vector unsigned long long vec_reve (vector unsigned long long);
23627 vector bool short vec_reve (vector bool short);
23628 vector signed short vec_reve (vector signed short);
23629 vector double vec_reve (vector double);
23630 vector float vec_reve (vector float);
23631 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
23632 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
23633 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
23634 (altivec_vreve): New pattern.
23635 * config/rs6000/altivec.h (vec_reve): New define.
23636 * doc/extend.texi (vec_rev): Update the built-in documentation file
23637 for the new built-in functions.
23639 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23641 PR tree-optimization/71815
23642 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
23644 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
23646 (slsr_process_phi): Likewise.
23647 (replace_uncond_cands_and_profitable_phis): Don't replace a
23648 multiply candidate with a stride of 1 (copy or cast).
23649 (phi_incr_cost): Call uses_consumed_by_stmt rather than
23651 (lowest_cost_path): Likewise.
23652 (total_savings): Likewise.
23654 2017-06-26 Richard Biener <rguenther@suse.de>
23657 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
23658 Use def_builtin_pure for all gather builtins.
23660 2017-06-26 Richard Biener <rguenther@suse.de>
23662 PR tree-optimization/81203
23663 * tree-tailcall.c (find_tail_calls): Do not move stmts into
23664 non-dominating BBs.
23666 2017-06-26 Marek Polacek <polacek@redhat.com>
23669 * doc/invoke.texi: Document -Wmultistatement-macros.
23671 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
23673 * doc/sourcebuild.texi (ARM-specific attributes): Document new
23674 arm_neon_ok_no_float_abi effective target.
23676 2017-06-26 Richard Biener <rguenther@suse.de>
23678 PR tree-optimization/80928
23679 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
23680 (copy_bbs): Set BB_DUPLICATED flag early.
23681 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
23683 (execute_on_shrinking_pred): Likewise.
23684 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
23685 BB_DUPLICATED blocks.
23686 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
23687 iterate over all PHIs considering removal of *gsi.
23689 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
23691 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
23694 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23696 * config/rs6000/rs6000-string.c: (expand_block_clear,
23697 do_load_for_compare, select_block_compare_mode,
23698 compute_current_alignment, expand_block_compare,
23699 expand_strncmp_align_check, expand_strn_compare,
23700 expand_block_move, rs6000_output_load_multiple)
23701 Move functions related to string/block move/compare
23702 to a separate file.
23703 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
23704 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
23705 for this function which is now used in two files.
23706 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
23707 * config.gcc: Add rs6000-string.o to extra_objs for
23708 targets powerpc*-*-* and rs6000*-*-*.
23710 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
23713 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
23714 32-bit, since indexed is not valid for DImode.
23715 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
23716 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
23717 (define_peephole2 for Altivec d-form load): Add 32-bit support.
23718 (define_peephole2 for Altivec d-form store): Likewise.
23721 * multiple_target.c (create_dispatcher_calls): Only create the
23722 dispatcher call if the function is the default clone of a
23723 versioned function.
23725 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
23727 PR middle-end/80902
23728 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
23729 a call, force the call to not be a tail call.
23731 2017-06-23 Jeff Law <law@redhat.com>
23733 * doc/contrib.texi: Add entry for Steven Pemberton's work on
23736 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
23738 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
23739 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
23740 handling for early expansion of vector shifts (sl,sr,sra,rl).
23741 (builtin_function_type): Add vector shift right instructions
23742 to the unsigned argument list.
23744 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
23746 rtl-optimizatoin/79286
23747 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
23748 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
23749 trap. PIC register plus a const unspec without offset can never trap.
23751 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
23753 * tree.h (builtin_structptr_type): New type.
23754 (builtin_structptr_types): Declare new array.
23755 * tree.c (builtin_structptr_types): New array.
23756 (free_lang_data, build_common_tree_nodes): Use it.
23758 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
23761 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
23764 2017-06-22 Matt Turner <mattst88@gmail.com>
23766 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
23767 Lake models to skylake case. Assume skylake for unknown
23768 models with clflushopt.
23770 2017-06-22 Jeff Law <law@redhat.com>
23772 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
23773 frame sizes that do not satisfy aarch64_uimm12_shift.
23775 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
23777 * profile-count.h (apply_probability,
23778 apply_scale, probability_in): Fix checks for zero.
23780 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23782 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
23783 * doc/cppdiropts.texi (-I @var{dir}): Document it.
23785 2016-06-22 Richard Biener <rguenther@suse.de>
23787 * tree-vect-loop.c (vect_model_reduction_cost): Handle
23788 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
23789 REDUC_MAX_EXPR support.
23790 (vectorizable_reduction): Likewise.
23791 (vect_create_epilog_for_reduction): Likewise.
23793 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
23795 * match.pd (A / (1 << B) -> A >> B): New.
23796 * generic-match-head.c: Include optabs-tree.h.
23797 * gimple-match-head.c: Likewise.
23798 * optabs-tree.h (target_supports_op_p): New.
23799 * optabs-tree.c (target_supports_op_p): New.
23801 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23803 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
23804 $gcc_cv_ld --help output.
23805 (gcc_cv_ld_demangle): Likewise.
23806 (gcc_cv_ld_eh_frame_hdr): Likewise.
23807 (gcc_cv_ld_pie): Likewise.
23808 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
23809 (gcc_cv_ld_buildid): Likewise.
23810 (gcc_cv_ld_sysroot): Likewise.
23811 (ld_bndplt_support): Likewise.
23812 (ld_pushpopstate_support): Likewise.
23813 * configure: Regenerate.
23814 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
23816 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23819 * config/i386/sse.md (round<mode>2): Renumber match_dup and
23820 operands indexes to avoid gap between operands and match_dups.
23822 2017-06-21 Andrew Pinski <apinski@cavium.com>
23824 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
23825 Increment Arith_shift and Arith_shift_reg by 1.
23826 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
23828 * config/aarch64/aarch64.c (thunderx_tunings): Enable
23829 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
23830 (aarch64_strip_extend): Add new argument and test for it.
23831 (aarch64_cheap_mult_shift_p): New function.
23832 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
23833 add a cost if it is true.
23834 Update calls to aarch64_strip_extend.
23835 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
23837 2017-06-21 Andrew Pinski <apinski@cavium.com>
23839 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
23841 (thunderxt88): Likewise.
23842 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
23843 (thunderx_prefetch_tune): New variable.
23844 (thunderx2t99_prefetch_tune): Update for the correct values.
23845 (thunderxt88_tunings): New variable.
23846 (thunderx_tunings): Use thunderx_prefetch_tune instead of
23847 generic_prefetch_tune.
23848 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
23850 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23852 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
23853 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
23854 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
23855 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
23856 (aarch64_atomic_cas<mode>, GPI): Likewise.
23858 2017-06-21 Martin Liska <mliska@suse.cz>
23860 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
23861 statements on cold and hot labels.
23862 * predict.c (tree_estimate_probability_bb): Remove the
23863 prediction from this place.
23865 2017-06-21 Martin Liska <mliska@suse.cz>
23867 PR tree-optimization/79489
23868 * gimplify.c (maybe_add_early_return_predict_stmt): New
23870 (gimplify_return_expr): Call the function.
23871 * predict.c (tree_estimate_probability_bb): Remove handling
23873 * predict.def: Update comment about early return predictor.
23874 * gimple-predict.h (is_gimple_predict): New function.
23875 * predict.def: Change default value of early return to 66.
23876 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
23878 * passes.def: Put pass_strip_predict_hints to the beginning of
23881 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
23883 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
23884 FUNCTION_DECL declarations.
23885 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
23887 (dwaf2out_decl): Likewise.
23888 * godump.c (go_early_global_decl): Skip call to the real debug hook
23889 for FUNCTION_DECL declarations.
23890 * passes.c (rest_of_decl_compilation): Skip call to the
23891 early_global_decl debug hook for FUNCTION_DECL declarations, unless
23892 -fdump-go-spec is passed.
23894 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23896 * config/i386/i386.c (struct builtin_isa): New field pure_p.
23897 Reorder for compactness.
23898 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
23899 (def_builtin_pure, def_builtin_pure2): New functions.
23900 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
23902 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23904 * match.pd (nop_convert): New predicate.
23905 ((A +- CST1) +- CST2): Allow some NOP conversions.
23907 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23910 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
23911 with ctors/dtors if GOVD_SHARED is set.
23913 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23915 * config/aarch64/aarch64.md (movti_aarch64):
23916 Emit mov rather than orr.
23917 (movtf_aarch64): Likewise.
23918 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
23919 Emit mov rather than orr.
23921 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23923 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
23924 Swap alternatives, make integer dup more expensive.
23926 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23928 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
23929 Return true for non-tls symbols.
23931 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
23933 * config/aarch64/aarch64-cores.def (cortex-a55): New.
23934 (cortex-a75): Likewise.
23935 (cortex-a75.cortex-a55): Likewise.
23936 * config/aarch64/aarch64-tune.md: Regenerate.
23937 * doc/invoke.texi (-mtune): Document new values for -mtune.
23939 2017-06-21 Tom de Vries <tom@codesourcery.com>
23941 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
23942 stack_size feature.
23943 (Effective-Target Keywords, Other attributes): Suggest using
23944 dg-add-options stack_size feature to get stack limit in stack_size
23945 effective target documentation.
23947 2017-06-21 Julian Brown <julian@codesourcery.com>
23948 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23950 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
23951 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
23952 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
23954 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
23955 attribute type list for neon_multiply.
23956 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
23957 attribute type list for neon_multiply.
23958 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
23959 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
23960 attribute type list for neon_multiply.
23961 * config/arm/types.md (crypto_pmull): Add.
23962 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
23963 attribute type list.
23965 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
23967 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
23970 2017-06-20 Jakub Jelinek <jakub@redhat.com>
23972 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
23973 to make sure not to dereference a NULL cost_classes_ptr pointer.
23975 2017-06-20 Carl Love <cel@us.ibm.com>
23977 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23978 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23979 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23980 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23981 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23982 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23983 VMULOSW): New enum "unspec" values.
23984 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23985 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23986 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23987 altivec_vmulosw): New patterns.
23988 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23989 VMULOSW): Add definitions.
23991 2017-06-20 Julia Koval <julia.koval@intel.com>
23993 * config/i386/i386.c: Fix rounding expand for new pattern.
23994 * config/i386/subst.md: Fix pattern (parallel -> unspec).
23996 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23998 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
23999 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
24001 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24003 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
24006 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24008 * config/aarch64/aarch64-cores.def: Rearrange to sort by
24009 architecture, then by implementer ID.
24010 * config/aarch64/aarch64-tune.md: Regenerate.
24012 2017-06-20 Richard Biener <rguenther@suse.de>
24014 PR middle-end/81097
24015 * fold-const.c (split_tree): Fold to type before negating.
24017 2017-06-20 David Malcolm <dmalcolm@redhat.com>
24019 * diagnostic-show-locus.c
24020 (selftest::test_fixit_deletion_affecting_newline): New function.
24021 (selftest::diagnostic_show_locus_c_tests): Call it.
24023 2017-06-20 Andreas Schwab <schwab@suse.de>
24026 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
24029 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24031 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
24032 __ARM_FEATURE_COPROC according to support.
24034 2017-06-20 Jakub Jelinek <jakub@redhat.com>
24036 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
24037 Rewritten to avoid overflow for > 32-bit pointers.
24040 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
24041 by removing enum keyword.
24042 (ubsan_type_descriptor): Likewise. Formatting fix.
24045 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
24046 splitter): Require TARGET_SSE2 in the condition.
24048 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
24051 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
24052 for doing vector set of SFmode on ISA 3.0.
24053 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
24054 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
24056 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
24057 SFmode value into a V4SF variable that was extracted from another
24058 V4SF variable without converting the element to double precision
24059 and back to single precision vector format.
24060 (vsx_insert_extract_v4sf_p9_2): Likewise.
24062 2017-06-19 Jakub Jelinek <jakub@redhat.com>
24064 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
24065 in UWHI to avoid undefined overflow.
24068 * ubsan.h (enum ubsan_encode_value_phase): New.
24069 (ubsan_encode_value): Change second argument to
24070 enum ubsan_encode_value_phase with default value of
24071 UBSAN_ENCODE_VALUE_GENERIC.
24072 * ubsan.c (ubsan_encode_value): Change second argument to
24073 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
24074 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
24075 create_tmp_var_raw instead of create_tmp_var and use a
24077 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
24078 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
24079 ubsan_encode_value callers.
24082 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
24083 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
24084 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
24086 2017-06-19 Richard Biener <rguenther@suse.de>
24088 PR middle-end/81118
24089 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
24090 estimates if we changed anything.
24092 2017-06-19 Richard Biener <rguenther@suse.de>
24094 PR tree-optimization/80887
24095 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
24096 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
24097 simplified lookups, then reset mprts_hook.
24098 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
24100 (try_to_simplify): Likewise.
24102 2017-06-19 Martin Liska <mliska@suse.cz>
24105 * gimplify.c (gimplify_switch_expr):
24106 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
24108 2017-06-19 Martin Liska <mliska@suse.cz>
24110 * doc/install.texi: Document that PGO runs in 4 stages.
24112 2017-06-19 Martin Liska <mliska@suse.cz>
24115 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
24116 to dispatcher function name.
24117 * multiple_target.c (replace_function_decl): New function.
24118 (create_dispatcher_calls): Redirect both edges and references.
24120 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
24122 * profile-count.c (profile_count::dump): Dump quality.
24123 (profile_count::differs_from_p): Update for unsigned val.
24124 * profile-count.h (profile_count_quality): New enum.
24125 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
24127 2017-06-19 Richard Biener <rguenther@suse.de>
24129 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
24130 struct function as arg.
24131 (estimate_numbers_of_iterations): Export overload with loop arg.
24132 (free_numbers_of_iterations_estimates_loop): Use an overload of
24133 free_numbers_of_iterations_estimates instead.
24134 * tree-cfg.c (remove_bb): Adjust.
24135 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
24136 * tree-parloops.c (gen_parallel_loop): Likewise.
24137 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
24139 (tree_unroll_loops_completely): Likewise.
24140 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
24141 Use an overload instead and export.
24142 (estimated_loop_iterations): Adjust.
24143 (max_loop_iterations): Likewise.
24144 (likely_max_loop_iterations): Likewise.
24145 (estimate_numbers_of_iterations): Take struct function as arg
24147 (loop_exits_before_overflow): Adjust.
24148 (free_numbers_of_iterations_estimates_loop): Use an overload.
24149 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
24150 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
24152 2017-06-19 Richard Biener <rguenther@suse.de>
24155 * ipa-prop.c (find_constructor_constant_at_offset): Handle
24156 RANGE_EXPR conservatively.
24158 2017-06-16 Carl Love <cel@us.ibm.com>
24160 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24161 definitions for vec_float, vec_float2, vec_floato,
24162 vec_floate built-ins.
24163 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
24164 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
24166 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
24167 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
24168 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
24169 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
24170 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
24171 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
24172 vec_floato): Add builtin defines.
24173 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
24174 Update the built-in documentation file for the new built-in
24177 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24179 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
24180 (mthumb): Mark as the negative of -marm.
24182 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24184 * doc/invoke.texi (ARM Options, -mcpu): Document supported
24186 (ARM Options, -mtune): Document that this accepts the same
24187 extension options as -mcpu.
24188 (ARM Options, -mfpu): Document addition of -mfpu=auto.
24190 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24192 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
24193 permitted extensions.
24195 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24197 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
24198 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
24199 (armv8-m.main): Add option +nodsp.
24200 * config/arm/arm-cpu-cdata.h: Regenerated.
24202 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24204 * config/arm/t-fuchsia: New file.
24205 * config.gcc (arm*-*-fuchsia*): Use it.
24207 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24209 * config/arm/t-symbian: Rewrite for new option infrastructure.
24211 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24213 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
24214 (MULTILIB_REQUIRED): Likewise.
24216 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24218 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
24219 (MULTILIB_RESUE): Likewise.
24220 (MULTILIB_MATCHES): Likewise.
24221 (MULTLIB_REQUIRED): Likewise.
24223 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24225 * config/arm/t-rtems: Rewrite for new option framework.
24227 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24229 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
24230 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
24231 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
24232 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
24233 * config/arm/t-multilib: ... here.
24234 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
24235 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
24236 armv7-a and armv8*-a when A-profile libraries have not been built.
24237 * config/arm/t-rmprofile: Rewrite.
24239 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24241 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
24242 with a backslash. Remove the backslash after substituting unescaped
24244 * doc/fragments.texi (MULTILIB_REUSE): Document it.
24246 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24248 * config.gcc: (arm*-*-*): When building a-profile libraries, force
24249 the driver to pass through the default setting of -mfloat-abi.
24250 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
24252 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
24253 (all_feat_combs): New rule.
24254 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
24256 * config/arm/t-aprofile: Rewrite.
24258 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24260 * config/arm/arm.h (FPUTYPE_AUTO): Define.
24261 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
24262 fpu is not specified by the user/command-line.
24263 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
24264 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
24265 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
24266 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
24267 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
24268 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
24270 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24272 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
24273 * config/arm/t-arm-elf: Rewritten.
24275 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24277 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
24278 have some floating-point instructions.
24279 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
24280 (TARGET_MAYBE_HARD_FLOAT): New macro.
24281 * config/arm/arm-builtins.c (arm_init_builtins): Use
24282 TARGET_MAYBE_HARD_FLOAT.
24283 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
24285 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24287 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
24288 (configargs.h): Include it.
24289 (arm_print_hint_for_fpu_option): New function.
24290 (arm_parse_fpu_option): New function.
24291 (candidate_extension): New class.
24292 (arm_canon_for_multilib): New function.
24293 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
24294 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
24295 (ARCH_CANONICAL_SPECS): New macro.
24296 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
24298 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24300 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
24301 are set after handling multilib fragments. Set target_cpu_default2
24304 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24306 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
24308 (arm*-*-*): Set target_cpu_default2 to a quoted string.
24309 * config/arm/parsecpu.awk (check_cpu): Validate any extension
24311 (check_arch): Likewise.
24312 * config/arm/arm.c (arm_configure_build_target): Handle
24313 TARGET_CPU_DEFAULT being a string constant. Scan any feature
24314 options in the default.
24316 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24318 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
24319 when an option is an alias of another.
24320 * config/arm/parsecpu.awk (optalias): New parser token.
24321 (gen_comm_data): Mark non-alias options as such. Emit entries
24322 for extension aliases.
24323 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
24324 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
24325 (armv6kz, armv6zk, armv6t2): Likewise.
24326 (armv7): Make vfpv3-d16 an alias.
24327 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
24329 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
24330 Sort in canonical order.
24331 (armv8-a): Sort in canonical order.
24332 (armv8.1-a, armv8.2-a): Likewise.
24333 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
24335 (cortex-a9): Sort in canonical order.
24336 * config/arm/arm.c (selftests.h): Include it.
24337 (arm_test_cpu_arch_data): New function.
24338 (arm_run_self_tests): New function.
24339 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
24340 (targetm): Move declaration to the end of the file.
24341 * arm-cpu-cdata.h: Regenerated.
24343 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24345 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
24346 call to target_mode_check describing the type of option passed.
24347 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
24348 (arm_target_thumb_only): Use arm_parse_arch_option_name or
24349 arm_parse_cpu_option_name to match parameters against list of
24351 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
24352 arm_arch_core_flags data structure.
24353 * config/arm/arm-cpu_cdata.h: Regenerated.
24355 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24357 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
24359 (arm_print_hint_for_cpu_option): Likewise.
24360 (arm_print_hint_for_arch_option): Likewise.
24361 (arm_parse_cpu_option_name): Likewise.
24362 (arm_parse_arch_option_name): Likewise.
24363 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
24364 of entries in the all_fpus list.
24365 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
24366 (arm_parse_cpu_option_name): Declare.
24367 (arm_parse_arch_option_name): Declare.
24368 (arm_parse_option_features): Declare.
24369 (arm_intialize_isa): Declare.
24370 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
24372 (gen_comm_data): ... here. Make definitions non-static.
24373 * config/arm/arm-cpu-data.h: Regenerated.
24374 * config/arm/arm-cpu-cdata.h: Regenerated.
24376 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24378 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
24379 (cpu_arch_extension): New structure.
24380 (cpu_arch_option, arch_option, cpu_option): New structures.
24381 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
24382 architecture types.
24383 (gen_data): Generate new format data tables.
24384 * config/arm/arm.c (cpu_tune): New structure.
24385 (cpu_option, processors): Delete.
24386 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
24387 (arm_print_hint_for_cpu_option): ... this and ...
24388 (arm_print_hint_for_arch_option): ... this.
24389 (arm_parse_arch_cpu_name): Delete. Replace with ...
24390 (arm_parse_cpu_option_name): ... this and ...
24391 (arm_parse_arch_option_name): ... this.
24392 (arm_unrecognized_feature): Change type of target parameter to
24394 (arm_parse_arch_cpu_features): Delete. Replace with ...
24395 (arm_parse_option_features): ... this.
24396 (arm_configure_build_target): Rework to use new configuration data
24398 (arm_print_tune_info): Rework for new configuration data tables.
24399 * config/arm/arm-cpu-data.h: Regenerated.
24400 * config/arm/arm-cpu.h: Regenerated.
24402 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24404 * Makefile.in (OBJS): Move sbitmap.o from here ...
24405 (OBJS-libcommon): ... to here.
24407 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24409 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
24410 (ISA_ALL_CRYPTO): New macro.
24411 (ISA_ALL_SIMD): New macro
24412 (ISA_ALL_FP): New macro.
24413 * config/arm/arm.c (fpu_bitlist): Update initializer.
24414 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
24416 (arm9e): Add fpu. Add option for nofp
24417 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
24418 (arm926ej-s, arm1026ej-s): Likewise.
24419 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
24420 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
24421 neon-fp16, neon-vfpv4, nofp and nosimd.
24422 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
24423 (cortex-a8): Add fpu. Add option for nofp.
24424 (cortex-a9): Add fpu. Add options for nosimd and nofp.
24425 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
24426 (cortex-r4f): Add fpu.
24427 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
24428 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
24430 (cortex-r8): Likewise.
24431 (cortex-m4): Add fpu. Add option for nofp.
24432 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
24433 (cortex-a17.cortex-a7): Likewise.
24434 (cortex-a32): Add fpu. Add options for crypto and nofp.
24435 (cortex-a35, cortex-a53): Likewise.
24436 (cortex-a57): Add fpu. Add option for crypto.
24437 (cortex-a72, cortex-a73): Likewise.
24438 (exynos-m1): Likewise.
24439 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
24440 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
24441 (cortex-m33): Add fpu. Add option for nofp.
24442 * config/arm/arm-cpu-cdata.h: Regenerated
24443 * config/arm/arm-cpu-data.h: Regenerated.
24445 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24447 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
24448 (armv5te, armv5tej): Likewise.
24449 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
24450 (armv7): Add options fp and vfpv3-d16.
24451 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
24452 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
24454 (armv7ve): Likewise.
24455 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
24456 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
24457 (armv8-a): Add nocrypto option.
24458 (armv8.1-a, armv8.2-a): Likewise.
24459 (armv8-m.main): add options fp, fp.dp and nofp.
24461 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24463 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
24465 (armv8-a+crc): Delete.
24466 (armv8.1-a): Add options simd, crypto and nofp.
24467 (armv8.2-a): Add options fp16, simd, crypto and nofp.
24468 (armv8.2-a+fp16): Delete.
24469 (armv8-m.main): Add option dsp.
24470 (armv8-m.main+dsp): Delete.
24471 (cortex-a8): Add fpu. Add nofp option.
24472 (cortex-a9): Add fpu. Add nofp and nosimd options.
24473 * config/arm/parsecpu.awk (gen_data): Generate option tables and
24474 link to main cpu and architecture data structures.
24475 (gen_comm_data): Only put isa attributes from the main architecture
24477 (option): New statement for architecture and CPU entries.
24478 * arm.c (struct cpu_option): New structure.
24479 (struct processors): Add entry for options.
24480 (arm_unrecognized_feature): New function.
24481 (arm_parse_arch_cpu_name): Ignore any characters after the first
24483 (arm_parse_arch_cpu_feature): New function.
24484 (arm_configure_build_target): Separate out any CPU and architecture
24485 features and parse separately. Don't error out if -mfpu=auto is
24486 used with only an architecture string.
24487 (arm_print_asm_arch_directives): New function.
24488 (arm_file_start): Call it.
24489 * config/arm/arm-cpu-cdata.h: Regenerated.
24490 * config/arm/arm-cpu-data.h: Likewise.
24491 * config/arm/arm-tables.opt: Likewise.
24493 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24495 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
24496 assembler when it is not -mfpu=auto.
24498 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24500 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
24501 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
24502 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
24503 (ASM_CPU_SPEC): Rewrite.
24504 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
24505 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
24506 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
24507 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
24508 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
24509 copied string is NUL-terminated. Also strip any characters prefixed
24511 (arm_rewrite_selected_arch): New function.
24512 (arm_rewrite_march): New function.
24514 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24516 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
24517 (x_arm_cpu_string, x_arm_tune_string): Likewise.
24518 (march, mcpu, mtune): Convert to string-based options.
24519 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
24520 (arm_parse_arch_cpu_name): New function.
24521 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
24522 identify selected architecture or CPU.
24523 (arm_option_save): New function.
24524 (TARGET_OPTION_SAVE): Redefine.
24525 (arm_option_restore): Restore string options.
24526 (arm_option_print): Print string options.
24528 2017-06-16 Martin Sebor <msebor@redhat.com>
24530 PR tree-optimization/80933
24531 PR tree-optimization/80934
24532 * builtins.c (fold_builtin_3): Do not handle bcmp here.
24533 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
24534 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
24535 (gimple_fold_builtin): Call them.
24537 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24539 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
24540 as unlikely; update profile.
24542 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24544 * predict.c (force_edge_cold): Handle declaring edges impossible
24547 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24549 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
24551 (try_unroll_loop_completely): Fix reporting.
24553 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24555 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
24557 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
24560 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
24561 if given a non-constant argument for an intrinsic which requires a
24564 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24566 * profile.c (compare_freqs): New function.
24567 (branch_prob): Sort edge list.
24568 (find_spanning_tree): Assume that the list is priority sorted.
24570 2017-06-16 Richard Biener <rguenther@suse.de>
24572 PR tree-optimization/81090
24573 * passes.def (pass_record_bounds): Remove.
24574 * tree-pass.h (make_pass_record_bounds): Likewise.
24575 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
24576 make_pass_record_bounds): Likewise.
24577 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
24578 not free niter estimates at the beginning but at the end.
24579 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
24581 2017-06-16 Richard Biener <rguenther@suse.de>
24583 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
24584 initializer to workaround ICE in host GCC 4.8.
24586 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24588 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
24590 (clone_inlined_nodes): Update.
24592 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24594 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
24595 prefetch settings, and enable prefetching by default at -O3.
24597 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24599 * config/aarch64/aarch64.c (aarch64_override_options_internal):
24600 Set flag_prefetch_loop_arrays according to tuning data.
24602 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24604 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
24605 New tune structure.
24606 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
24607 [Unrelated to main purpose of the patch] Place the pointer field last
24608 to enable type checking errors when tune structure are wrongly merged.
24609 * config/aarch64/aarch64.c (generic_prefetch_tune,)
24610 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
24611 (thunderx2t99_prefetch_tune): New tune constants.
24612 (tune_params *_tunings): Update all tunings (no functional change).
24613 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
24614 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
24615 from tunings structures.
24617 2017-06-16 Jakub Jelinek <jakub@redhat.com>
24620 * ubsan.c (instrument_null): Add T argument, use it instead
24621 of computing it based on IS_LHS.
24622 (instrument_object_size): Likewise.
24623 (pass_ubsan::execute): Adjust instrument_null and
24624 instrument_object_size callers to pass gimple_get_lhs or
24625 gimple_assign_rhs1 result to it. Use instrument_null instead of
24626 calling get_base_address and instrument_mem_ref. Handle
24627 aggregate call arguments for object-size sanitization.
24629 2017-06-16 Yury Gribov <tetra2005@gmail.com>
24631 PR tree-optimization/81089
24632 * tree-vrp.c (is_masked_range_test): Validate operands of
24635 2017-06-15 Martin Sebor <msebor@redhat.com>
24638 * dumpfile.c (dump_register): Avoid calling memset to initialize
24639 a class with a default ctor.
24640 * gcc.c (struct compiler): Remove const qualification.
24641 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
24642 * hash-table.h: Ditto.
24643 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
24645 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
24646 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
24648 * params.h (struct param_info): Make struct members non-const.
24649 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
24650 with default initialization.
24651 * vec.h (vec_copy_construct, vec_default_construct): New helper
24653 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
24654 with vec_copy_construct.
24655 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
24656 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
24657 * doc/invoke.texi (-Wclass-memaccess): Document.
24659 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24661 * emit-rtl.h (is_leaf): Update comment about local
24662 register allocator.
24664 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
24667 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
24668 for a variable to have a section before checking if the section has a
24670 Set section to.persistent if persistent attribute is set.
24671 Warn if .persistent attribute is used on an automatic variable.
24673 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
24675 PR rtl-optimization/80474
24676 * reorg.c (update_block): Do not ignore instructions in a delay slot.
24678 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
24680 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
24683 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
24685 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
24686 (casesi): Emit bounds checking as RTL.
24687 (casesi_internal_mips16_<mode>): Remove bounds checking.
24689 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
24691 * config/xtensa/xtensa.c (xtensa_option_override): Append
24692 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
24693 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
24694 xtensa_doloop_hooks): Define unconditionally.
24695 (xtensa_reorg_loops): Only call reorg_loops in the presence of
24697 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
24698 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
24699 for it in xtensa_option_override.
24700 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
24701 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
24703 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
24705 * doc/cppopts.texi: Document '-' special value to -MF.
24707 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
24709 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
24710 (cortex_a53_fconst): Likewise.
24711 (cortex_a53_fpmul): Likewise.
24712 (cortex_a53_f_load_64): Likewise.
24713 (cortex_a53_f_load_many): Likewise.
24714 (cortex_a53_advsimd_alu): Likewise.
24715 (cortex_a53_advsimd_alu_q): Likewise.
24716 (cortex_a53_advsimd_mul): Likewise.
24717 (cortex_a53_advsimd_mul_q): Likewise.
24718 (fpmac bypass): Add new bypass for fpmac-fpmac case.
24719 Add missing fmul, r2f_cvt and fconst cases.
24721 2017-06-14 Richard Biener <rguenther@suse.de>
24723 PR middle-end/81088
24724 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
24726 (fold_binary_loc): When associating do not treat pre-existing
24727 TREE_OVERFLOW on literal constants as a reason to allow
24728 TREE_OVERFLOW on associated literal constants.
24730 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24732 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
24733 (MASK_FEATURES): New macro.
24734 * config/sparc/sparc.c (sparc_option_override): Remove the special
24735 handling of -mfpu and generalize it to all MASK_FEATURES switches.
24737 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24739 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
24740 a division of 0 if non-call exceptions are enabled.
24742 2017-06-14 Andrew Pinski <apinski@cavium.com>
24743 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24746 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
24747 Improve vector initialization code gen for only variable case.
24749 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24751 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
24753 2017-06-14 Richard Biener <rguenther@suse.de>
24755 PR tree-optimization/81083
24756 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
24759 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24761 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
24762 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
24763 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
24764 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
24765 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
24766 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
24768 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24770 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
24771 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
24773 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24775 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
24777 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24779 * config/rs6000/t-rtems: Don't handle SPE.
24781 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24783 * config/rs6000/t-linux: Don't handle SPE.
24785 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24787 * config/rs6000/eabispe.h: Delete file.
24789 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24791 * config/rs6000/t-spe: Delete file.
24793 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24795 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
24796 (rs6000_legitimate_offset_address_p): Return false for anything in
24797 V2SImode or V2SFmode.
24799 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24801 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
24802 except V2SF and V2SI. Rearrange the vector modes, and add comments.
24803 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
24805 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
24806 (rs6000_legitimate_offset_address_p): Ditto.
24807 (rs6000_emit_move): Ditto.
24808 (rs6000_init_builtins): Remove V4HI_type_node.
24810 2017-06-13 Martin Liska <mliska@suse.cz>
24813 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
24814 (gate_asan): Likewise.
24815 * asan.h (asan_no_sanitize_address_p): Remove the function.
24816 (sanitize_flags_p): New function.
24817 * builtins.def: Fix coding style.
24818 * common.opt: Use renamed enum value.
24819 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
24820 * doc/extend.texi: Document no_sanitize attribute.
24821 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
24822 to SANITIZE_UNDEFINED_NONDEFAULT.
24823 * gcc.c (sanitize_spec_function): Use the renamed enum value.
24824 * gimple-fold.c (optimize_atomic_compare_exchange_p):
24825 Use sanitize_flags_p.
24826 * gimplify.c (gimplify_function_tree): Likewise.
24827 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
24828 * opts.c (parse_no_sanitize_attribute): New function.
24829 (common_handle_option): Use renamed enum value.
24830 * opts.h (parse_no_sanitize_attribute): Declare.
24831 * tree.c (sanitize_flags_p): New function.
24832 * tree.h: Declared here.
24833 * tsan.c: Use sanitize_flags_p.
24834 * ubsan.c (ubsan_expand_null_ifn): Likewise.
24835 (instrument_mem_ref): Likewise.
24836 (instrument_bool_enum_load): Likewise.
24837 (do_ubsan_in_current_function): Remove the function.
24838 (pass_ubsan::execute): Use sanitize_flags_p.
24839 * ubsan.h: Remove do_ubsan_in_current_function
24840 * tree-cfg.c (print_no_sanitize_attr_value): New function.
24841 (dump_function_to_file): Use it here.
24843 2017-06-13 Martin Jambor <mjambor@suse.cz>
24845 PR tree-optimization/80803
24846 PR tree-optimization/81063
24847 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
24848 (propagate_subaccesses_across_link): Enqueue subtree whenever
24849 necessary instead of relying on the caller.
24851 2017-06-13 Martin Jambor <mjambor@suse.cz>
24853 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
24854 that have a first_link.
24855 (sort_and_splice_var_accesses): Do not check first_link before
24857 (subtree_mark_written_and_enqueue): Likewise.
24858 (propagate_all_subaccesses): Likewise and do not stop at first
24859 parent with a first_link.
24861 2017-06-13 Martin Jambor <mjambor@suse.cz>
24863 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
24866 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24868 * match.pd: New pattern.
24870 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24872 * tree-vrp.c (is_masked_range_test): New function.
24873 (register_edge_assert_for): Determine ranges for
24876 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24878 PR tree-optimization/67328
24879 * fold-const.c (maskable_range_p): New function.
24880 (build_range_check): Generate bittests if possible.
24882 2017-06-13 Martin Liska <mliska@suse.cz>
24884 * gimple-pretty-print.c (dump_probability): Add new argument.
24885 (dump_edge_probability): Dump both probability and count.
24886 (dump_gimple_label): Likewise.
24887 (dump_gimple_bb_header): Likewise.
24889 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
24892 * config/avr/avr-devices.c: Fix indentation.
24893 * config/avr/gen-avr-mmcu-specs.c: Dito.
24895 2017-06-13 Richard Biener <rguenther@suse.de>
24897 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
24898 instead get vector type from stmt_info.
24899 (vectorizable_reduction): Adjust. Remove dead code.
24901 2017-06-13 Richard Biener <rguenther@suse.de>
24903 PR middle-end/81065
24904 * fold-const.c (extract_muldiv_1): Remove bogus distribution
24905 case of C * (x * C2 + C3).
24906 (fold_addr_of_array_ref_difference): Properly fold index difference.
24908 2017-06-12 David S. Miller <davem@davemloft.net>
24911 * config/sparc/sparc.md (return expander): Emit frame blockage if
24912 function uses alloca.
24914 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
24916 * combine.c (make_field_assignment): Check len rather than the mode
24917 precision when calling force_to_mode.
24919 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
24921 Support multilibs and devices that see flash in RAM address range.
24924 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
24925 (avr_mcu_t) <flash_pm_offset>: New field.
24926 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
24927 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
24928 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
24929 (AVR_TINY_PM_OFFSET): Remove macro.
24930 * config/avr/avr.opt (-mshort-calls): New option.
24931 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24932 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
24933 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
24934 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
24935 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
24936 instead of avr_arch->have_jmp_call.
24937 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
24938 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
24939 avr_arch->flash_pm_offset to define.
24940 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
24941 new field flash_pm_offset. Add entry for avrxmega3.
24942 (avr_texinfo): Add entry for avrxmega3.
24943 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
24944 attiny212, attiny214,
24945 attiny412, attiny414, attiny416, attiny417,
24946 attiny814, attiny816, attiny817,
24947 attiny1614, attiny1616, attiny1617,
24948 attiny3214, attiny3216, attiny3217.
24949 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
24950 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
24951 (avr_print_operand_address) [AVR_TINY]: Same.
24952 (avr_asm_init_sections) <readonly_data_section>: Only patch
24953 callback if avr_arch->flash_pm_offset = 0.
24954 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
24955 for rodata if avr_arch->flash_pm_offset != 0.
24956 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
24957 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
24958 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
24959 (m_options): Append opt_rcall.
24960 (m_dirnames): Append dir_rcall.
24961 * config/avr/t-multilib: Regenerate.
24963 * configure.ac [target=avr]: Check whether avrxmega3 default
24964 linker description file works as needed.
24965 * configure: Regenerate.
24966 * doc/avr-mmcu.texi: Regenerate.
24967 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
24968 <__AVR_ARCH__>: Document avrxmega3 and 103.
24969 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
24970 <__AVR_SHORT_CALLS__>: Document it.
24971 <__AVR_PM_BASE_ADDRESS__>: Document it.
24972 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
24973 (AVR Variable Attributes) <progmem>: Document this is
24974 not needed for avrxmega3.
24975 (AVR Named Address Spaces) <__flash>: Dito.
24977 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
24979 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
24981 2017-06-12 Doug Rupp <rupp@adacore.com>
24983 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
24984 Append vxworks-stdint.h to the tm_file list.
24985 * config/vxworks-stdint.h: New file.
24987 2017-06-12 Martin Liska <mliska@suse.cz>
24989 PR tree-optimization/81041
24990 * tree-profile.c (gimple_gen_ic_func_profiler):
24991 Create an extra BB in profile-generate
24992 (gimple_gen_time_profiler): Likewise.
24994 2017-06-12 Jakub Jelinek <jakub@redhat.com>
24996 PR tree-optimization/81003
24997 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
24998 (update_range_test): Use it instead of force_gimple_operand_gsi.
25000 2017-06-12 Richard Biener <rguenther@suse.de>
25002 PR tree-optimization/81053
25003 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
25004 with backedge value not defined in loop. Simplify def stmt
25007 2017-06-11 Tom de Vries <tom@codesourcery.com>
25010 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
25012 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
25013 nvptx_cannot_force_const_mem.
25015 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25017 * opts.c (finish_options): Move test for flag_split_stack after
25018 it has been initialized.
25020 2017-06-11 Jason Merrill <jason@redhat.com>
25022 * tree.h (id_equal): New.
25023 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
25024 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
25025 instead of strcmp of IDENTIFIER_POINTER.
25027 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25029 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
25030 (mark_all_inlined_calls_cdtor): Fix formating.
25031 (inline_transform): Rescale profile before inlining.
25033 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25035 * cgraph.h (cgraph_edge::clone): Update prototype.
25036 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
25037 (cgraph_node::create_clone): Update.
25038 (cgraph_node::create_version_clone): Update.
25039 * tree-inline.c (copy_bb): Update.
25040 (expand_call_inline): Update.
25042 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25044 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
25045 factored out from ...
25046 (rs6000_emit_prologue): ... here.
25048 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25050 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
25051 factored out from ...
25052 (rs6000_emit_prologue): ... here.
25054 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25056 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
25058 (handle_missing_profiles): Fix computation of tp_first_run.
25059 (counts_to_freqs): Do not touch freqs when count is 0.
25061 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25063 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
25066 2017-06-10 Tom de Vries <tom@codesourcery.com>
25068 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
25069 attributes): Document signal effective target.
25071 2017-06-10 Tom de Vries <tom@codesourcery.com>
25073 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
25074 Document effective target stack_size.
25076 2017-06-09 David Malcolm <dmalcolm@redhat.com>
25078 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
25079 to the edit_context if they can be auto-applied.
25081 2017-06-9 Ian Lance Taylor <iant@golang.org>
25083 * opts.c (finish_options): If -fsplit-stack, disable implicit
25084 -forder-blocks-and-partition.
25085 * doc/invoke.texi (Optimize Options): Document that when using
25086 -fsplit-stack -forder-blocks-and-partition is not implicitly
25089 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25091 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
25092 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
25093 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
25094 * builtins.def (abort, trap, unreachable): Declare cold.
25095 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
25096 * tree-core.h (ECF_COLD): New.
25097 * tree.c (set_call_expr_flags): Handle ECF_COLD.
25098 (build_common_builtin_nodes): Mark unreachable and abort as cold.
25100 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25102 * predict.c (unlikely_executed_stmt_p): Cleanup.
25104 2017-06-09 Richard Biener <rguenther@suse.de>
25106 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
25107 model if the ref is always written to.
25109 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25111 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
25113 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25115 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
25118 2017-06-09 Tom de Vries <tom@codesourcery.com>
25121 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
25122 "target cannot support label values" when encountering LABEL_REF.
25124 2017-06-09 Martin Liska <mliska@suse.cz>
25126 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
25127 (gimple_gen_ic_func_profiler): Emit direct comparison
25128 of __gcov_indirect_call_callee with NULL.
25129 (gimple_gen_time_profiler): Change probability from
25130 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
25132 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25134 * profile.c (edge_gcov_counts): Turn to pointer.
25135 (compute_branch_probabilities, compute_branch_probabilities): Update.
25136 (branch_prob): Do not clear edge_gcov_count.
25137 * profile.h (edge_gcov_counts): Turn to pointer.
25138 (edge_gcov_count): Update.
25140 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25142 * gimple.h (gimple_check_failed): Mark cold.
25144 2017-06-09 Richard Biener <rguenther@suse.de>
25146 PR tree-optimization/66623
25147 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
25148 refactor check_reduction into two parts, properly computing
25149 whether we have to check reduction validity for outer loop
25152 2017-06-09 Richard Biener <rguenther@suse.de>
25154 PR tree-optimization/79483
25155 * graphite-scop-detection.c (order): New global.
25156 (get_order): Compute bb to order mapping that satisfies code
25157 generation constraints.
25158 (cmp_pbbs): New helper.
25159 (build_scops): Start domwalk at entry block, sort generated
25162 2017-06-09 Richard Biener <rguenther@suse.de>
25164 PR middle-end/81007
25165 * ipa-polymorphic-call.c
25166 (ipa_polymorphic_call_context::restrict_to_inner_class):
25167 Skip FIELD_DECLs with error_mark_node type.
25168 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
25171 2017-06-09 Martin Liska <mliska@suse.cz>
25173 * predict.c (struct branch_predictor): New struct.
25174 (test_prediction_value_range): New test.
25175 (predict_c_tests): New function.
25176 * selftest-run-tests.c (selftest::run_tests): Run the function.
25177 * selftest.h: Declare new tests.
25179 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
25182 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
25183 gen_add3_insn did not fail.
25184 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
25185 r0, construct that number in a temporary reg and add that reg to r0.
25186 If asked to put the result in r0 as well, fail.
25188 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
25190 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25191 for early expansion of vec_eqv.
25193 2017-06-08 Jakub Jelinek <jakub@redhat.com>
25195 PR middle-end/81005
25196 * ubsan.c (instrument_null): Avoid pointless code temporary.
25197 (pass_ubsan::execute): Instrument aggregate arguments of calls.
25199 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
25203 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
25206 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
25209 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25211 * predict.c (unlikely_executed_edge_p): Move ahead.
25212 (probably_never_executed_edge_p): Use it.
25214 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25216 PR middle-end/79988
25217 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
25218 gimple_call_builtin_p call.
25220 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25222 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
25223 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
25224 rtl_check_failed_type2, rtl_check_failed_code1,
25225 rtl_check_failed_code2, rtl_check_failed_code_mode,
25226 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
25227 rtvec_check_failed_bounds, rtl_check_failed_flag,
25228 _fatal_insn_not_found, _fatal_insn): Likewise.
25229 * tree.h (tree_contains_struct_check_failed,
25230 tree_check_failed, tree_not_check_failed,
25231 tree_class_check_failed, tree_range_check_failed,
25232 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
25233 tree_vec_elt_check_failed, phi_node_elt_check_failed,
25234 tree_operand_check_failed, omp_clause_check_failed,
25235 omp_clause_operand_check_failed, omp_clause_range_check_failed):
25238 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25240 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
25241 flag_branch_probabilities.
25242 * ipa-inline.c (edge_badness): Likewise.
25243 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
25244 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
25245 * predict.c (maybe_hot_frequency_p): Likewise.
25246 (probably_never_executed): Likewise.
25247 * sched-ebb.c (schedule_ebbs): Likewise.
25248 * sched-rgn.c (find_single_block_region): Likewise.
25249 * tracer.c (tail_duplicate): Likewise.
25251 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25253 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
25254 longer requires x_flag_profile_use.
25256 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25258 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
25259 instead of flag_reorder_blocks_and_partition.
25260 * dbxout.c (dbxout_function_end): Likewise.
25261 * dwarf2out.c (gen_subprogram_die): Likewise.
25262 * haifa-sched.c (sched_create_recovery_edges): Likewise.
25263 * hw-doloop.c (reorg_loops): Likewise.
25264 * varasm.c (assemble_start_function,
25265 assemble_end_function): Likewise.
25266 (decide_function_section): Do not check for
25267 flag_reorder_blocks_and_partition.
25269 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25271 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
25273 (chkp_get_hard_register_fake_addr_expr): Ditto.
25274 (chkp_build_addr_expr): Add check for hard reg case.
25275 (chkp_parse_array_and_component_ref): Ditto.
25276 (chkp_find_bounds_1): Ditto.
25277 (chkp_process_stmt): Don't generate bounds store for
25280 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25282 * predict.c (maybe_hot_bb_p): Do not check profile status.
25283 (maybe_hot_edge_p): Likewise.
25284 (probably_never_executed): Check for zero counts even if profile
25286 (unlikely_executed_edge_p): New function.
25287 (unlikely_executed_stmt_p): New function.
25288 (unlikely_executed_bb_p): New function.
25289 (set_even_probabilities): Use unlikely predicates.
25290 (combine_predictions_for_bb): Likewise.
25291 (predict_paths_for_bb): Likewise.
25292 (predict_paths_leading_to_edge): Likewise.
25293 (determine_unlikely_bbs): New function.
25294 (estimate_bb_frequencies): Use it.
25295 (compute_function_frequency): Use zero counts even if profile is
25297 * profile-count.h: Fix typo.
25299 2017-08-08 Julia Koval <julia.koval@intel.com>
25301 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
25302 _mm512_mask_cvtsepi16_storeu_epi8,
25303 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
25304 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
25305 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
25306 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
25307 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
25308 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
25309 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
25310 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
25311 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
25312 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
25313 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
25314 __builtin_ia32_pmovuswb256mem_mask,
25315 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
25316 __builtin_ia32_pmovwb512mem_mask): New builtins.
25318 2017-08-08 Julia Koval <julia.koval@intel.com>
25320 PR target/73350,80862
25321 * config/i386/subst.md (round): Fix round pattern.
25322 * config/i386/i386.c (ix86_erase_embedded_rounding):
25323 Fix erasing rounding for the fixed pattern.
25325 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25327 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
25329 2017-06-08 Martin Liska <mliska@suse.cz>
25331 PR gcov-profile/80911
25332 * gcov.c (block_info::block_info): New constructor.
25334 2017-06-07 Carl Love <cel@us.ibm.com>
25336 * config/rs6000/rs6000-c: The return type of the following
25337 built-in functions was implemented as int not long long. Fix sign
25338 of return value for the unsigned version of vec_mulo and vec_mule.
25339 vector unsigned long long vec_bperm (vector unsigned long long,
25340 vector unsigned char)
25341 vector signed long long vec_mule (vector signed int,
25343 vector unsigned long long vec_mule (vector unsigned int,
25344 vector unsigned int)
25345 vector signed long long vec_mulo (vector signed int,
25347 vector unsigned long long vec_mulo (vector unsigned int,
25348 vector unsigned int)
25349 * doc/extend.texi: Fix the documentation for the built-in
25352 2017-06-07 Carl Love <cel@us.ibm.com>
25355 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
25358 2017-06-07 Carl Love <cel@us.ibm.com>
25360 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
25361 support, Generate doublehv for signed int/float for BE case only.
25363 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
25365 * doc/invoke.texi (mcx16): Rewrite.
25367 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25369 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
25370 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
25371 *mov<mode>_softfloat, and an anonymous splitter): Use
25372 nonimmediate_operand instead of rs6000_nonimmediate_operand.
25374 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25376 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
25378 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25379 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
25380 (rs6000_debug_reg_global): Adjust.
25381 (rs6000_init_hard_regno_mode_ok): Adjust.
25382 (rs6000_dbx_register_number): Adjust.
25383 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
25384 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25385 Remove SPE_ACC and SPEFSCR.
25386 (REG_ALLOC_ORDER): Ditto.
25387 (FRAME_POINTER_REGNUM): Change to 111.
25388 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
25389 (REG_CLASS_NAMES): Ditto.
25390 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
25391 (REGISTER_NAMES): Ditto.
25392 (ADDITIONAL_REG_NAMES): Ditto.
25393 (rs6000_reg_names): Ditto.
25394 * config/rs6000/rs6000.md: Renumber some register number
25397 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25399 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
25401 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25402 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
25404 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
25405 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25406 Delete the SPE high registers.
25407 (REG_ALLOC_ORDER): Ditto.
25408 (enum reg_class): Remove SPE_HIGH_REGS.
25409 (REG_CLASS_NAMES): Ditto.
25410 (REG_CLASS_CONTENTS): Delete the SPE high registers.
25411 (REGISTER_NAMES): Ditto.
25412 (rs6000_reg_names): Ditto.
25413 * doc/tm.texi.in: Remove SPE as example.
25414 * doc/tm.texi: Regenerate.
25416 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25418 * config/rs6000/8540.md (ppc8540_brinc): Delete.
25419 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
25420 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
25421 * config/rs6000/rs6000.md (type): Remove "brinc".
25423 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25425 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
25426 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
25427 * config/rs6000/linuxspe.h: Delete file.
25428 * config/rs6000/rs6000.md: Don't include spe.md.
25429 * config/rs6000/spe.h: Delete file.
25430 * config/rs6000/spe.md: Delete file.
25431 * config/rs6000/t-rs6000: Remove spe.md.
25433 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25435 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
25436 (reg_or_none500mem_operand): Delete.
25437 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
25438 instead of reg_or_none500mem_operand.
25440 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25442 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
25443 handling of SPE flags.
25444 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
25446 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25448 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
25450 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
25451 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
25452 paired_divv2sf3): Similar.
25453 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
25454 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
25455 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
25457 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
25458 Rename the paired_* instruction patterns.
25459 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
25461 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
25462 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
25463 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
25464 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
25465 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
25466 PAIRED_VECTOR_MODE.
25467 (struct machine_function): Delete field spe_insn_chain_scanned_p.
25468 (spe_func_has_64bit_regs_p): Delete.
25469 (spe_expand_predicate_builtin): Delete.
25470 (spe_expand_evsel_builtin): Delete.
25471 (TARGET_DWARF_REGISTER_SPAN): Do not define.
25472 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
25473 (invalid_e500_subreg): Delete.
25474 (rs6000_legitimize_address): Always force_reg op2 as well, for
25475 paired single memory accesses.
25476 (rs6000_member_type_forces_blk): Delete.
25477 (rs6000_spe_function_arg): Delete.
25478 (rs6000_expand_unop_builtin): Delete SPE handling.
25479 (rs6000_expand_binop_builtin): Ditto.
25480 (spe_expand_stv_builtin): Delete.
25481 (bdesc_2arg_spe): Delete.
25482 (spe_expand_builtin): Delete.
25483 (spe_expand_predicate_builtin): Delete.
25484 (spe_expand_evsel_builtin): Delete.
25485 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
25486 (spe_init_builtins): Delete.
25487 (spe_func_has_64bit_regs_p): Delete.
25488 (savres_routine_name): Delete "info" parameter. Adjust callers.
25489 (rs6000_emit_stack_reset): Ditto.
25490 (rs6000_dwarf_register_span): Delete.
25491 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
25492 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
25493 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
25495 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
25497 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
25498 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
25499 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
25500 mulv2sf3, divv2sf3): Delete expanders.
25502 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25504 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
25506 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25508 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
25509 * config/rs6000/rs6000.c: Ditto.
25511 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25513 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
25514 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
25515 comparison_operator.
25517 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25519 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
25520 * config/rs6000/rs6000.opt: Ditto.
25521 * config/rs6000/t-rtems: Ditto.
25523 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25525 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
25526 TARGET_E500_SINGLE by 0, simplify.
25527 * config/rs6000/rs6000.c: Ditto.
25528 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
25529 (spe_build_register_parallel): Delete.
25530 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
25531 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
25532 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
25533 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
25534 (E500_CONVERT): Delete.
25535 * config/rs6000/spe.md: Remove many patterns and all define_constants.
25537 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25539 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
25540 * config/rs6000/dfp.md: Ditto.
25541 (negdd2, *negdd2_fpr): Merge.
25542 (absdd2, *absdd2_fpr): Merge.
25543 (negtd2, *negtd2_fpr): Merge.
25544 (abstd2, *abstd2_fpr): Merge.
25545 * config/rs6000/e500.h: Delete file.
25546 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
25547 TARGET_FPRS by 1 and simplify.
25548 * config/rs6000/rs6000-c.c: Ditto.
25549 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
25550 TARGET_DF_SPE by 0.
25551 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
25553 * config/rs6000/rs6000.md: Ditto.
25554 (floatdidf2, *floatdidf2_fpr): Merge.
25555 (move_from_CR_gt_bit): Delete.
25556 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
25557 (E500_CR_IOR_COMPARE): Delete.
25558 (All patterns that require !TARGET_FPRS): Delete.
25559 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
25561 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25563 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
25565 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25567 * graphds.c (add_edge): Intitialize edge's attached data.
25568 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
25569 pointer parameter. Call pointed function on each edge during
25570 graph traversing. Skip traversing the edge when the function
25572 (graphds_dfs, graphds_scc): Ditto.
25573 (for_each_edge): New parameter. Pass the new parameter to callback
25575 * graphds.h (skip_edge_callback): New function pointer type.
25576 (graphds_dfs, graphds_scc): New function pointer parameter.
25577 (graphds_edge_callback, for_each_edge): New parameter.
25579 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25581 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
25582 out code checking if runtime alias check is possible to below ...
25583 Call the new function.
25584 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
25585 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
25587 2017-06-07 Marek Polacek <polacek@redhat.com>
25590 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
25591 TYPE_OVERFLOW_WRAPS check.
25593 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25595 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
25596 if versioning is required.
25597 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
25598 peeling with the check for versioning.
25600 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25602 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
25603 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
25604 Set true to new parameter if new ssa variable is defined.
25605 (vect_gen_vector_loop_niters): Refactor. Set range information
25606 for the new vector loop bound variable.
25607 (vect_do_peeling): Ditto.
25609 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25611 * tree-affine.c (ssa.h): Include header file.
25612 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
25613 has wrapping overflow behavior.
25615 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25617 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
25619 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25621 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
25622 (tree_to_aff_combination): ... here.
25624 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25626 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
25627 reg_pressure model function.
25628 (ivopts_global_cost_for_size): Delete.
25629 (determine_set_costs, iv_ca_recount_cost): Call new model function
25630 ivopts_estimate_reg_pressure.
25632 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25634 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
25635 expensive than udiv. Remove floating point cases from mod.
25637 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25639 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
25640 Increase idiv cost.
25642 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25644 * config/aarch64/aarch64.md
25645 (copysignsf3): Fix mask generation.
25647 2017-06-07 Jakub Jelinek <jakub@redhat.com>
25649 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
25651 (class dump_manager): Add register_dumps method.
25652 * dumpfile.c: Include langhooks.h.
25653 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
25654 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
25655 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
25656 (dump_manager::dump_register): Start with 512 entries instead of 32.
25657 (dump_manager::register_dumps): New method.
25658 * toplev.c (general_init): Instead of invoking register_dumps
25659 langhook, invoke register_dumps method on the dump manager.
25660 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
25663 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
25665 * doc/md.texi: Clarify the restrictions on a define_insn condition.
25666 Say that # requires an associated define_split to exist, and that
25667 the define_split must be suitable for use after register allocation.
25669 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25671 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
25672 (compute_outgoing_frequencies): Also initialize zero counts.
25673 (find_many_sub_basic_blocks): Do not produce uninitialized profile
25674 around loops; preserve more of profile when nothing changes.
25676 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
25678 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
25680 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
25681 * config/arm/arm-cpu-cdata.h: Regenerate.
25682 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
25683 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
25684 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
25685 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
25686 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
25688 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
25690 * config/arm/t-rmprofile: Likewise.
25691 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
25693 2017-06-06 David S. Miller <davem@davemloft.net>
25696 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
25697 blockage if function uses alloca.
25699 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25701 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
25702 New "uid" fields to hold pretty-print IDs of group and ref.
25703 Memory references are now identified as <group_id>:<ref_id>
25704 instead of using [random] addresses.
25705 (dump_mem_details): Simplify, no functional change.
25706 (dump_mem_ref): Simplify and make output more concise.
25707 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
25708 (find_or_create_group): Initialize group uid.
25709 (record_ref): Initialize ref uid. Improve debug output.
25710 (prune_group_by_reuse, should_issue_prefetch_p,)
25711 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
25712 (mark_nontemporal_store, determine_loop_nest_reuse):
25713 Improve debug output.
25715 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25717 * dbgcnt.def (prefetch): New debug counter.
25718 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
25719 (schedule_prefetches): Stop issueing prefetches if debug counter
25722 2017-06-06 Tom de Vries <tom@codesourcery.com>
25724 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
25725 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
25727 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25729 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
25730 Use aarch64_reg_or_zero predicate for operand 4.
25731 (aarch64_compare_and_swap<mode> define_insn_and_split):
25732 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
25733 (aarch64_store_exclusive<mode>): Likewise for operand 2.
25735 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25737 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
25738 (arm_compute_save_core_reg_mask): This.
25739 (thumb1_compute_save_reg_mask): Rename into ...
25740 (thumb1_compute_save_core_reg_mask): This.
25741 (arm_compute_save_reg0_reg12_mask): Adapt comment.
25742 (arm_compute_frame_layout): Likewise.
25744 2017-06-06 Richard Biener <rguenther@suse.de>
25746 PR tree-optimization/80974
25747 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
25748 keep or clear leaders SSA info.
25750 2017-06-06 Tom de Vries <tom@codesourcery.com>
25752 * config/nvptx/nvptx.c (split_mode_p): New function.
25753 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
25755 2017-06-06 Tom de Vries <tom@codesourcery.com>
25757 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
25759 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25762 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
25765 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25767 * shrink-wrap.c (handle_simple_exit): Update profile.
25768 (try_shrink_wrapping): Upate profile.
25770 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25772 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
25773 (tree_guess_outgoing_edge_probabilities): New.
25774 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
25775 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
25777 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25779 * ipa-split.c (split_function): Initialize return bb profile.
25781 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25783 * profile.c (compute_branch_probabilities): Also initialize
25784 EXIT_BLOCK profile.
25786 2017-06-06 Richard Biener <rguenther@suse.de>
25788 PR tree-optimization/80928
25789 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
25790 (vect_analyze_loop_operations): Properly guard analysis for
25792 (vect_transform_loop): Likewise.
25793 (vect_analyze_loop_2): Also reset SLP type on PHIs.
25794 (vect_model_induction_cost): Do not cost for pure SLP.
25795 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
25796 of induction in inner loop vectorization.
25797 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
25798 (vect_get_and_check_slp_defs): Handle vect_induction_def.
25799 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
25801 (vect_analyze_slp_cost_1): Cost induction.
25802 (vect_detect_hybrid_slp_stmts): Handle PHIs.
25803 (vect_get_slp_vect_defs): Likewise.
25804 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
25805 (vect_transform_stmt): Handle SLP reductions.
25806 * tree-vectorizer.h (vectorizable_induction): Adjust.
25808 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25810 * config/rs6000/rs6000.c (make_resolver_func): Update
25811 init_lowered_empty_function call.
25813 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
25815 * doc/invoke.texi: Document the -fprofile-abs-path option.
25816 * common.opt (fprofile-abs-path): New option.
25817 * gcov-io.h (gcov_write_filename): Declare.
25818 * gcov-io.c (gcov_write_filename): New function.
25819 * coverage.c (coverage_begin_function): Use gcov_write_filename.
25820 * profile.c (output_location): Likewise.
25822 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25824 * shring-wrap.c: Revert accidental commit.
25826 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
25828 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
25830 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25832 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
25834 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
25835 profile in callgraph edge.
25836 * profile-count.h (apply_probability): If THIS is 0, then result is 0
25837 (apply_scale): Likewise.
25838 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
25839 Also scale profile when inlining function with zero profile.
25840 (initialize_cfun): Update exit block profile even when it is zero.
25841 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
25842 when profile is read.
25844 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25846 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
25847 (CLONE_*): New constants to define the processors we can generate
25848 code for with the target_clone attribute.
25849 (rs6000_clone_map): New array to identify which clone processors
25850 the current program is running on.
25851 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
25852 target_clone attribute.
25853 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
25854 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
25855 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
25856 (cpu_expand_builtin): Add support for target_clone attribute.
25857 (rs6000_valid_attribute_p): Allow "default" attribute.
25858 (get_decl_name): New debug function to simplify printing the
25859 current function name in debugging statements.
25860 (rs6000_clone_priority): New functions to support the target_clone
25861 attribute, and be able to generate code to switch between ISA 2.05
25862 through ISA 3.0 (power6 through power9).
25863 (rs6000_compare_version_priority): Likewise.
25864 (rs6000_get_function_versions_dispatcher): Likewise.
25865 (make_resolver_func): Likewise.
25866 (add_condition_to_bb): Likewise.
25867 (dispatch_function_versions): Likewise.
25868 (rs6000_generate_version_dispatcher_body): Likewise.
25869 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
25870 (fusion_gpr_load_p): Fix a spacing issue.
25871 * doc/extend.texi (Common Function Attributes): Document that the
25872 PowerPC supports the target_clone attribute.
25874 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25876 * config/arm/arm.h: explain F symbol found in description of ARM
25877 register allocation in its legend.
25879 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25881 * config/mips/frame-header-opt.c: Include profile-count.h.
25882 * config/riscv/riscv.c: Include profile-count.h
25884 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25886 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
25888 (sm_set_flag_if_changed): Add bbs field.
25889 (execute_sm_if_changed_flag_set): Pass BBS.
25890 (execute_sm): Update.
25892 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25894 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
25897 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25899 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
25900 (peephole2): New peephole2 to emit the above.
25901 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
25903 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25905 * config/aarch64/aarch64.c (define_peephole2 above
25906 *sub_<shift>_<mode>): New peephole.
25908 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25910 * config/i386/i386.c (make_resolver_func): Update.
25911 * Makefile.in: Add profile-count.h and profile-count.o
25912 * auto-profile.c (afdo_indirect_call): Update to new API.
25913 (afdo_set_bb_count): Update.
25914 (afdo_propagate_edge): Update.
25915 (afdo_propagate_circuit): Update.
25916 (afdo_calculate_branch_prob): Update.
25917 (afdo_annotate_cfg): Update.
25918 * basic-block.h: Include profile-count.h
25919 (struct edge_def): Turn count to profile_count.
25920 (struct basic_block_def): Likewie.
25921 (REG_BR_PROB_BASE): Move to profile-count.h
25922 (RDIV): Move to profile-count.h
25923 * bb-reorder.c (max_entry_count): Turn to profile_count.
25924 (find_traces): Update.
25925 (rotate_loop):Update.
25926 (connect_traces):Update.
25927 (sanitize_hot_paths):Update.
25928 * bt-load.c (migrate_btr_defs): Update.
25929 * cfg.c (RDIV): Remove.
25930 (init_flow): Use alloc_block.
25931 (alloc_block): Uninitialize count.
25932 (unchecked_make_edge): Uninitialize count.
25933 (check_bb_profile): Update.
25934 (dump_edge_info): Update.
25935 (dump_bb_info): Update.
25936 (update_bb_profile_for_threading): Update.
25937 (scale_bbs_frequencies_int): Update.
25938 (scale_bbs_frequencies_gcov_type): Update.
25939 (scale_bbs_frequencies_profile_count): New.
25940 * cfg.h (update_bb_profile_for_threading): Update.
25941 (scale_bbs_frequencies_profile_count): Declare.
25942 * cfgbuild.c (compute_outgoing_frequencies): Update.
25943 (find_many_sub_basic_blocks): Update.
25944 * cfgcleanup.c (try_forward_edges): Update.
25945 (try_crossjump_to_edge): Update.
25946 * cfgexpand.c (expand_gimple_tailcall): Update.
25947 (construct_exit_block): Update.
25948 * cfghooks.c (verify_flow_info): Update.
25949 (dump_bb_for_graph): Update.
25950 (split_edge): Update.
25951 (make_forwarder_block): Update.
25952 (duplicate_block): Update.
25953 (account_profile_record): Update.
25954 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
25955 (get_estimated_loop_iterations): Update.
25956 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
25957 (single_likely_exit): Update.
25958 * cfgloopmanip.c (scale_loop_profile): Update.
25960 (set_zero_probability): Update.
25961 (lv_adjust_loop_entry_edge): Update.
25962 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
25963 (purge_dead_edges): Update.
25964 (rtl_account_profile_record): Update.
25965 * cgraph.c (cgraph_node::create): Uninitialize count.
25966 (symbol_table::create_edge): Uninitialize count.
25967 (cgraph_update_edges_for_call_stmt_node): Update.
25968 (cgraph_edge::dump_edge_flags): Update.
25969 (cgraph_node::dump): Update.
25970 (cgraph_edge::maybe_hot_p): Update.
25971 * cgraph.h: Include profile-count.h
25972 (create_clone), create_edge, create_indirect_edge): Update.
25973 (cgraph_node): Turn count to profile_count.
25974 (cgraph_edge0: Likewise.
25975 (make_speculative, clone): Update.
25976 (create_edge): Update.
25977 (init_lowered_empty_function): Update.
25978 * cgraphclones.c (cgraph_edge::clone): Update.
25979 (duplicate_thunk_for_node): Update.
25980 (cgraph_node::create_clone): Update.
25981 * cgraphunit.c (cgraph_node::analyze): Update.
25982 (cgraph_node::expand_thunk): Update.
25983 * final.c (dump_basic_block_info): Update.
25984 * gimple-streamer-in.c (input_bb): Update.
25985 * gimple-streamer-out.c (output_bb): Update.
25986 * graphite.c (print_global_statistics): Update.
25987 (print_graphite_scop_statistics): Update.
25988 * hsa-brig.c: Include basic-block.h.
25989 * hsa-dump.c: Include basic-block.h.
25990 * hsa-gen.c (T sum_slice): Update.
25991 (convert_switch_statements):Update.
25992 * hsa-regalloc.c: Include basic-block.h.
25993 * ipa-chkp.c (chkp_produce_thunks): Update.
25994 * ipa-cp.c (struct caller_statistics): Update.
25995 (init_caller_stats): Update.
25996 (gather_caller_stats): Update.
25997 (ipcp_cloning_candidate_p): Update.
25998 (good_cloning_opportunity_p): Update.
25999 (get_info_about_necessary_edges): Update.
26000 (dump_profile_updates): Update.
26001 (update_profiling_info): Update.
26002 (update_specialized_profile): Update.
26003 (perhaps_add_new_callers): Update.
26004 (decide_about_value): Update.
26005 (ipa_cp_c_finalize): Update.
26006 * ipa-devirt.c (struct odr_type_warn_count): Update.
26007 (struct decl_warn_count): Update.
26008 (struct final_warning_record): Update.
26009 (possible_polymorphic_call_targets): Update.
26010 (ipa_devirt): Update.
26011 * ipa-fnsummary.c (redirect_to_unreachable): Update.
26012 * ipa-icf.c (sem_function::merge): Update.
26013 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
26014 * ipa-inline.c (compute_uninlined_call_time): Update.
26015 (compute_inlined_call_time): Update.
26016 (want_inline_small_function_p): Update.
26017 (want_inline_self_recursive_call_p): Update.
26018 (edge_badness): Update.
26019 (lookup_recursive_calls): Update.
26020 (recursive_inlining): Update.
26021 (inline_small_functions): Update.
26022 (dump_overall_stats): Update.
26023 (dump_inline_stats): Update.
26024 * ipa-profile.c (ipa_profile_generate_summary): Update.
26025 (ipa_propagate_frequency): Update.
26026 (ipa_profile): Update.
26027 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26028 * ipa-utils.c (ipa_merge_profiles): Update.
26029 * loop-doloop.c (doloop_modify): Update.
26030 * loop-unroll.c (report_unroll): Update.
26031 (unroll_loop_runtime_iterations): Update.
26032 * lto-cgraph.c (lto_output_edge): Update.
26033 (lto_output_node): Update.
26034 (input_node): Update.
26035 (input_edge): Update.
26036 (merge_profile_summaries): Update.
26037 * lto-streamer-in.c (input_cfg): Update.
26038 * lto-streamer-out.c (output_cfg): Update.
26039 * mcf.c (create_fixup_graph): Update.
26040 (adjust_cfg_counts): Update.
26041 (sum_edge_counts): Update.
26042 * modulo-sched.c (sms_schedule): Update.
26043 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
26044 * predict.c (maybe_hot_count_p): Update.
26045 (probably_never_executed): Update.
26046 (dump_prediction): Update.
26047 (combine_predictions_for_bb): Update.
26048 (propagate_freq): Update.
26049 (handle_missing_profiles): Update.
26050 (counts_to_freqs): Update.
26051 (rebuild_frequencies): Update.
26052 (force_edge_cold): Update.
26053 * predict.h: Include profile-count.h
26054 (maybe_hot_count_p, counts_to_freqs): UPdate.
26055 * print-rtl-function.c: Do not include cfg.h
26056 * print-rtl.c: Include basic-block.h
26057 * profile-count.c: New file.
26058 * profile-count.h: New file.
26059 * profile.c (is_edge_inconsistent): Update.
26060 (correct_negative_edge_counts): Update.
26061 (is_inconsistent): Update.
26062 (set_bb_counts): Update.
26063 (read_profile_edge_counts): Update.
26064 (compute_frequency_overlap): Update.
26065 (compute_branch_probabilities): Update; Initialize and deinitialize
26067 (branch_prob): Update.
26068 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
26069 (edge_gcov_count): New.
26070 (bb_gcov_count): New.
26071 * shrink-wrap.c (try_shrink_wrapping): Update.
26072 * tracer.c (better_p): Update.
26073 * trans-mem.c (expand_transaction): Update.
26074 (ipa_tm_insert_irr_call): Update.
26075 (ipa_tm_insert_gettmclone_call): Update.
26076 * tree-call-cdce.c: Update.
26077 * tree-cfg.c (gimple_duplicate_sese_region): Update.
26078 (gimple_duplicate_sese_tail): Update.
26079 (gimple_account_profile_record): Update.
26080 (execute_fixup_cfg): Update.
26081 * tree-inline.c (copy_bb): Update.
26082 (copy_edges_for_bb): Update.
26083 (initialize_cfun): Update.
26084 (freqs_to_counts): Update.
26085 (copy_cfg_body): Update.
26086 (expand_call_inline): Update.
26087 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26088 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
26089 (try_unroll_loop_completely): Update.
26090 (try_peel_loop): Update.
26091 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26092 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
26093 * tree-ssa-loop-split.c (connect_loops): Update.
26094 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
26095 * tree-ssa-reassoc.c (branch_fixup): Update.
26096 * tree-ssa-tail-merge.c (replace_block_by): Update.
26097 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
26098 (compute_path_counts): Update.
26099 (update_profile): Update.
26100 (recompute_probabilities): Update.
26101 (update_joiner_offpath_counts): Update.
26102 (estimated_freqs_path): Update.
26103 (freqs_to_counts_path): Update.
26104 (clear_counts_path): Update.
26105 (ssa_fix_duplicate_block_edges): Update.
26106 (duplicate_thread_path): Update.
26107 * tree-switch-conversion.c (case_bit_test_cmp): Update.
26108 (struct switch_conv_info): Update.
26109 * tree-tailcall.c (decrease_profile): Update.
26110 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26111 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26112 * value-prof.c (check_counter): Update.
26113 (gimple_divmod_fixed_value): Update.
26114 (gimple_mod_pow2): Update.
26115 (gimple_mod_subtract): Update.
26116 (gimple_ic_transform): Update.
26117 (gimple_stringop_fixed_value): Update.
26118 * value-prof.h (gimple_ic): Update.
26120 2017-06-02 Carl Love <cel@us.ibm.com>
26122 * config/rs6000/rs6000-c: Add support for built-in functions
26123 vector double vec_doublee (vector signed int);
26124 vector double vec_doublee (vector unsigned int);
26125 vector double vec_doublee (vector float);
26126 vector double vec_doubleh (vector signed int);
26127 vector double vec_doubleh (vector unsigned int);
26128 vector double vec_doubleh (vector float);
26129 vector double vec_doublel (vector signed int);
26130 vector double vec_doublel (vector unsigned int);
26131 vector double vec_doublel (vector float);
26132 vector double vec_doubleo (vector signed int);
26133 vector double vec_doubleo (vector unsigned int);
26134 vector double vec_doubleo (vector float);.
26135 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
26136 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
26138 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
26139 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
26140 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
26142 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
26143 vec_doublel, vec_doubleh.
26144 * doc/extend.texi: Update the built-in documentation file for the
26145 new built-in functions.
26147 2017-06-02 David Malcolm <dmalcolm@redhat.com>
26150 * ipa-inline-analysis.c (free_growth_caches): Set
26151 edge_removal_hook_holder to NULL after removing it.
26153 2017-06-02 Sudakshina Das <sudi.das@arm.com>
26155 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
26156 comparision with zero.
26158 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26159 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26160 for early expansion of vec_min and vec_max builtins.
26161 (builtin_function_type): Add min/max unsigned variants to those
26162 identified as having unsigned arguments.
26164 2017-06-02 Olivier Hainque <hainque@adacore.com>
26166 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
26168 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26170 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
26171 Use VALL_F16 iterator rather than VALL.
26173 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26175 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
26176 Emit CBNZ inside loop when doing a strong exchange and comparing
26177 against zero. Generate the CC flags after the loop.
26179 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
26181 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
26182 (dl_section_ref): New.
26183 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
26184 On AIX, append an expression to subtract the size of the
26185 section length to dl_section_ref.
26187 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26189 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26190 for early expansion of vector absolute builtins.
26192 2017-06-02 Richard Biener <rguenther@suse.de>
26194 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
26195 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
26197 2017-06-02 Richard Biener <rguenther@suse.de>
26199 PR tree-optimization/80948
26200 * tree-tailcall.c (find_tail_calls): Track stmts to move in
26201 stmt order as well.
26203 2017-06-02 Richard Biener <rguenther@suse.de>
26205 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
26207 * tree-vect-stmts.c (process_use): Do not mark backedge defs
26208 for inductions as relevant.
26210 2017-06-02 Richard Biener <rguenther@suse.de>
26212 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
26213 (vectorizable_induction): ... this. Remove dead code.
26215 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
26217 * builtins. (expand_builtin_alloca): Remove second parameter and
26218 infer its value from the first parameter instead.
26219 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
26221 2017-06-02 Jakub Jelinek <jakub@redhat.com>
26223 PR rtl-optimization/80903
26224 * loop-doloop.c (add_test): Unshare sequence.
26226 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
26228 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
26230 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
26232 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
26234 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
26235 xlogue_layout::get_instance, logue_layout::xlogue_layout,
26236 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
26237 (xlogue_layout::get_stub_rtx): Make static.
26238 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
26239 (xlogue_layout::compute_stub_managed_regs): Rename to...
26240 (xlogue_layout::count_stub_managed_regs): ...this.
26241 (xlogue_layout::is_stub_managed_reg): New function.
26242 (xlogue_layout::m_stub_names): Rename to...
26243 (xlogue_layout::s_stub_names): ...this, make static.
26244 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
26245 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
26246 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
26247 xlogue_layout::s_stub_names): Instantiate statics.
26248 (stub_managed_regs): Remove.
26249 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
26250 (disable_call_ms2sysv_xlogues): Rename to...
26251 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
26252 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
26254 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
26255 change after reload_completed.
26256 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
26258 (ix86_expand_prologue): Likewise.
26259 (ix86_expand_epilogue): Likewise.
26260 (ix86_expand_split_stack_prologue): Likewise.
26261 (ix86_compute_frame_layout): Remove frame parameter ...
26262 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
26263 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
26265 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
26266 (ix86_frame): Move from here ...
26267 * config/i386/i386.h (ix86_frame): ... to here.
26268 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
26269 complete ix86_frame data structure instead. Remove some_ld_name.
26271 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26273 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
26274 symbols that hold a DECL_VALUE_EXPR.
26276 2017-06-01 Martin Jambor <mjambor@suse.cz>
26278 PR tree-optimization/80898
26279 * tree-sra.c (process_subtree_disqualification): Removed.
26280 (disqualify_candidate): Do not acll
26281 process_subtree_disqualification.
26282 (subtree_mark_written_and_enqueue): New function.
26283 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
26284 RHS has been disqualified and re-queue LHS if necessary. Apart
26285 from that, ignore disqualified RHS.
26287 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26289 * config/s390/s390.c (s390_emit_epilogue): Disable early return
26290 address fetch for z10 or later.
26292 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26294 * config/arc/arc.md (tst_movb): Add guard when splitting.
26296 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26298 * config/arc/arc.c (arc_can_eliminate): Test against
26299 arc_frame_pointer_needed.
26301 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26303 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
26304 to prevent store reordering.
26305 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
26306 (type): Add block type.
26307 (stack_tie): Define special instruction to be used in
26310 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26312 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
26313 constraint. It is not valid for the pattern.
26314 (noncommutative_binary_comparison): Likewise.
26316 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26318 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
26321 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26323 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
26324 be used by the reg-alloc.
26326 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26328 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
26329 reg-alloc when having mul64 or mul32x16 instructions.
26330 (mulsidi3): Likewise.
26331 (umulsidi3): Likewise.
26332 (mulsi32x16): New pattern.
26333 (mulsi64): Likewise.
26334 (mulsidi64): Likewise.
26335 (umulsidi64): Likewise.
26336 (MUL32x16_REG): Define.
26337 (mul64_600): Use MUL32x16_REG.
26338 (mac64_600): Likewise.
26339 (umul64_600): Likewise.
26340 (umac64_600): Likewise.
26342 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26344 * config/arc/arc.md (mulsi3_700): Make it commutative.
26346 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
26348 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
26350 (*sign_extendsidi2_insn): Likewise for movstosw.
26352 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26354 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
26355 the type of the input discriminant value. Convert the
26356 discriminant value of signedness vary.
26358 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
26360 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
26361 Add to -Wall section.
26363 2017-06-01 Richard Biener <rguenther@suse.de>
26365 PR middle-end/66313
26366 * fold-const.c (fold_plusminus_mult_expr): If the factored
26367 factor may be zero use a wrapping type for the inner operation.
26368 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
26369 and handle moved defs.
26370 (process_assignment): Properly guard the unary op case. Return a
26371 tri-state indicating that moving the stmt before the call may allow
26372 to continue. Pass through to_move.
26373 (find_tail_calls): Handle moving unrelated defs before
26376 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
26379 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
26380 splitter result in the canonical way.
26382 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
26384 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
26385 also for 32bit target. Update insn attributes.
26386 (zero-extendsidi2 splitter): Allow all registers for operand 1.
26388 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
26390 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
26391 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
26392 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
26393 (_mm_maskz_min_ss): New intrinsics.
26395 2017-05-31 Martin Liska <mliska@suse.cz>
26397 * tree-vect-loop.c (vect_create_epilog_for_reduction):
26398 Change comment style to one we normally use.
26399 (vectorizable_reduction): Likewise.
26400 (vectorizable_induction): Likewise.
26401 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
26402 (vectorizable_call): Likewise.
26403 (vectorizable_simd_clone_call): Likewise.
26404 (vectorizable_conversion): Likewise.
26405 (vectorizable_assignment): Likewise.
26406 (vectorizable_shift): Likewise.
26407 (vectorizable_operation): Likewise.
26408 (vectorizable_store): Likewise.
26409 (vectorizable_load): Likewise.
26410 * tree-vectorizer.h: Likewise.
26412 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
26414 * passes.c (emergency_dump_function): New.
26415 * tree-pass.h (emergency_dump_function): Declare.
26416 * plugin.c (plugins_internal_error_function): Remove.
26417 * plugin.h (plugins_internal_error_function): Remove declaration.
26418 * toplev.c (internal_error_function): New static function. Use it...
26419 (general_init): ...here.
26421 2017-05-31 Graham Markall <graham.markall@embecosm.com>
26423 * config/arc/arc.c (arc_print_operand): Handle constant operands.
26424 (arc_rtx_costs): Add costs for new patterns.
26425 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
26426 * config/arc/predicates.md: Add _1_2_3_operand predicate.
26428 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26430 * tree-ssa-strlen.c (get_next_strinfo): New function.
26431 (get_stridx_plus_constant): Use it.
26432 (zero_length_string): Likewise.
26433 (adjust_related_strinfos): Likewise.
26434 (adjust_last_stmt): Likewise.
26436 2017-05-31 Richard Biener <rguenther@suse.de>
26439 * config/i386/i386.c (ix86_expand_builtin): Remove assert
26440 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
26442 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26444 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
26445 loop_vinfo argument and use of dependence distance vectors.
26446 Check instead whether the two references differ only in their
26447 initial value and assume that they have the same alignment if the
26448 difference is a multiple of the vector alignment.
26449 (vect_analyze_data_refs_alignment): Update call accordingly.
26451 2017-05-31 Martin Liska <mliska@suse.cz>
26454 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
26456 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26458 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
26459 (create_intersect_range_checks): Move from ...
26460 * tree-data-ref.c (create_intersect_range_checks_index)
26461 (create_intersect_range_checks): ... to here.
26462 (create_runtime_alias_checks): New function factored from ...
26463 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
26464 here. Call above function.
26465 * tree-data-ref.h (create_runtime_alias_checks): New function.
26467 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26469 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
26470 segment length for dr_b and compute it in wide_int.
26472 2017-05-31 Richard Biener <rguenther@suse.de>
26474 PR tree-optimization/80906
26475 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
26476 and pass through iv_map.
26477 (copy_bb_and_scalar_dependences): Adjust.
26478 (translate_pending_phi_nodes): Likewise.
26479 (copy_loop_close_phi_args): Handle code-generating IVs instead
26482 2017-05-30 David Malcolm <dmalcolm@redhat.com>
26484 * diagnostic-color.c (color_dict): Add "type-diff".
26485 (parse_gcc_colors): Update comment.
26486 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
26487 -fdiagnostics-show-template-tree and -fno-elide-type.
26488 (GCC_COLORS): Add type-diff to example.
26490 (-fdiagnostics-show-template-tree): New.
26491 (-fno-elide-type): New.
26492 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
26493 the pp_format_decoder callback. Call any m_format_postprocessor's
26495 (pretty_printer::pretty_printer): Initialize
26496 m_format_postprocessor.
26497 (pretty_printer::~pretty_printer): Delete any
26498 m_format_postprocessor.
26499 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
26500 (class format_postprocessor): New class.
26501 (struct pretty_printer::format_decoder): Document the new parameters.
26502 (struct pretty_printer::m_format_postprocessor): New field.
26503 * tree-diagnostic.c (default_tree_printer): Update for new
26504 bool and const char ** params.
26505 * tree-diagnostic.h (default_tree_printer): Likewise.
26507 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
26509 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
26510 (lwa_operand): Delete rs6000_gen_cell_microcode test.
26511 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26512 rs6000_gen_cell_microcode code.
26513 (rs6000_final_prescan_insn): Delete.
26514 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
26515 "warn-cell-microcode" entries.
26516 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
26517 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
26518 throughout. Change cc_reg_not_micro_cr0_operand to
26519 cc_reg_not_cr0_operand throughout.
26520 (*extendhi<mode>2_noload): Delete.
26521 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
26522 (mwarn-cell-microcode): Delete.
26523 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
26524 -mgen-cell-microcode and -mwarn-cell-microcode.
26526 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
26529 * config/i386/constraints.md (Yd): New constraint.
26531 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
26532 and (?Yd, r) alternatives. Update insn attributes.
26533 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
26534 and (?*Yd, r) alternatives. Update insn attributes.
26535 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
26537 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
26539 * gimplify.c (gimplify_modify_expr): Don't create a
26540 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
26543 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
26545 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
26547 2017-05-30 Richard Biener <rguenther@suse.de>
26549 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
26550 and reduc_def fields.
26551 (STMT_VINFO_REDUC_TYPE): New define.
26552 (STMT_VINFO_REDUC_DEF): Likewise.
26553 (vect_force_simple_reduction): Adjust prototype.
26554 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
26555 (vect_is_simple_reduction): Remove check_reduction argument.
26556 (vect_force_simple_reduction): Adjust and set
26557 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
26558 (vectorizable_reduction): Do not re-do reduction analysis
26559 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
26560 * tree-parloops.c (gather_scalar_reductions): Adjust.
26562 2017-05-30 Richard Biener <rguenther@suse.de>
26564 PR middle-end/80901
26565 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
26568 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
26570 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
26571 Introduce unknown_misalignment parameter and remove vf.
26572 (vect_peeling_hash_get_lowest_cost):
26573 Pass unknown_misalignment parameter.
26574 (vect_enhance_data_refs_alignment):
26575 Fix unsupportable data ref treatment.
26577 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26579 * tree-vect-data-refs.c (vect_get_data_access_cost):
26580 Workaround for SLP handling.
26581 (vect_enhance_data_refs_alignment):
26582 Compute costs for doing no peeling at all, compare to the best
26583 peeling costs so far and avoid peeling if cheaper.
26585 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26587 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
26588 Return peeling info and set costs to zero for unlimited cost
26590 (vect_enhance_data_refs_alignment): Also inspect all datarefs
26591 with unknown misalignment. Compute and costs for unknown
26592 misalignment, compare them to the costs for known misalignment
26593 and choose the cheapest for peeling.
26595 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26597 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
26598 (vect_get_peeling_costs_all_drs): Create function.
26599 (vect_peeling_hash_get_lowest_cost):
26600 Use vect_get_peeling_costs_all_drs.
26601 (vect_peeling_supportable): Create function.
26602 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
26604 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26606 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
26607 DR_HAS_NEGATIVE_STEP.
26608 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
26609 (vect_enhance_data_refs_alignment): Use.
26610 (vect_duplicate_ssa_name_ptr_info): Use.
26611 * tree-vectorizer.h (dr_misalignment): Use.
26612 (known_alignment_for_access_p): Use.
26614 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
26617 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
26619 (has_section_name): New function.
26621 2017-05-30 Martin Liska <mliska@suse.cz>
26624 * auto-profile.c (get_function_decl_from_block): Fix
26627 2017-05-30 Richard Biener <rguenther@suse.de>
26629 PR middle-end/80876
26630 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
26632 2017-05-30 Martin Liska <mliska@suse.cz>
26634 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
26635 * dumpfile.h (struct dump_file_info): Remove ctors.
26637 2017-05-30 Martin Liska <mliska@suse.cz>
26639 * predict.def: Fix GNU coding style.
26641 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
26643 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
26644 Mark 'to' argument with ATTRIBUTE_UNUSED.
26646 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
26648 * config/xtensa/xtensa.c (xtensa_emit_call): Use
26649 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
26650 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
26653 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
26655 * doc/install.texi (Options specification): Restore entry of
26656 --enable-sjlj-exceptions.
26658 2017-05-27 Michael Eager <eager@eagercon.com>
26661 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
26663 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
26665 * config/microblaze/microblaze.h
26666 (FIXED_REGISTERS): Update in macro.
26667 (CALL_USED_REGISTERS): Update in macro.
26669 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
26671 * doc/install.texi: Add links to macOS binary distributions.
26673 2017-05-27 Jakub Jelinek <jakub@redhat.com>
26677 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26679 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
26681 2017-05-26 Martin Liska <mliska@suse.cz>
26683 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
26685 2017-05-26 Martin Liska <mliska@suse.cz>
26687 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
26688 always leading ';; '.
26689 (dump_bb_info): Likewise.
26690 (brief_dump_cfg): Likewise.
26691 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
26692 * dumpfile.c: Remove usage of TDF_VERBOSE.
26693 * dumpfile.h (enum dump_kind): Likewise.
26694 (dump_gimple_bb_header): Do not use TDF_COMMENT.
26695 * print-tree.c (debug_verbose): Remove.
26696 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
26697 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
26698 * tree-diagnostic.c (default_tree_printer): Replace
26699 TDF_DIAGNOSTIC with TDF_SLIM.
26701 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26703 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
26704 in parameter loop, rather than loop_vinfo.
26705 (create_intersect_range_checks): Ditto.
26706 (vect_create_cond_for_alias_checks): Update call to above functions.
26708 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26710 PR tree-optimization/80815
26711 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
26712 for merging runtime alias checks. Handle negative DR_STEPs.
26714 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26716 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
26718 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
26719 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
26720 out code pruning runtime alias checks.
26721 * tree-data-ref.c (prune_runtime_alias_test_list): New function
26722 factored out from above.
26723 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
26725 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
26727 (prune_runtime_alias_test_list): New decalaration.
26729 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26731 * tree-vect-data-refs.c (compare_tree): Rename and move ...
26732 * tree-data-ref.c (data_ref_compare_tree): ... to here.
26733 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
26734 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
26735 (operator==, comp_dr_with_seg_len_pair): Ditto.
26736 (vect_prune_runtime_alias_test_list): Ditto.
26738 2017-05-26 Martin Liska <mliska@suse.cz>
26741 * params.def: Bound partial-inlining-entry-probability param.
26743 2017-05-26 Marek Polacek <polacek@redhat.com>
26746 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
26749 2017-05-26 Richard Biener <rguenther@suse.de>
26751 PR tree-optimization/80842
26752 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
26755 2017-05-26 Richard Biener <rguenther@suse.de>
26757 PR tree-optimization/80844
26758 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
26760 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
26762 * doc/md.texi (Machine Constraints): Update x86 family
26763 machine constraints section to match 'config/i386/constraints.md'.
26765 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
26767 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
26769 2017-05-25 Nathan Sidwell <nathan@acm.org>
26771 * doc/invoke.texi (--enable-languages): Update documentation.
26773 2017-05-25 Martin Liska <mliska@suse.cz>
26775 * dumpfile.c: Add TDF_FOLDING.
26776 * dumpfile.h (enum dump_kind): Likewise.
26777 * genmatch.c (dt_simplify::gen_1): Use it.
26779 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26781 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
26783 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26785 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
26786 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
26788 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26790 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
26791 * match.pd (X == C): Rewrite it here.
26792 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
26793 with_certain_nonzero_bits2): New predicates.
26794 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
26796 2017-05-24 Nathan Sidwell <nathan@acm.org>
26798 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
26801 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
26804 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26806 * config/powerpcspe: New port. Files are copied from the rs6000
26807 port, with "rs6000" in filenames replaced by "powerpcspe".
26809 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
26811 PR rtl-optimization/80754
26812 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
26814 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
26816 * config/sparc/sparc.md (length): Return the correct value for -mflat
26817 sibcalls to match output_sibcall.
26819 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26823 * config/rs6000/rs6000.c (struct machine_function): Add new field
26825 (rs6000_get_separate_components): Init that field, use it.
26826 (rs6000_components_for_bb): Use the field.
26828 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
26830 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
26832 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
26834 PR middle-end/80823
26835 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
26837 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26840 * config/s390/s390.c (s390_check_qrst_address): Check incoming
26841 address against address_operand predicate.
26842 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
26844 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
26846 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
26847 parameters passed indirectly.
26849 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
26851 * config/i386/i386.md (*movdi_internal): Remove SSE4
26852 alternative 18 (?r, *v). Update insn attributes.
26853 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
26854 Update insn attributes.
26855 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
26856 Update insn attributes.
26857 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
26858 alternative 1 (r, v). Remove isa attribute.
26859 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
26860 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
26861 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
26863 2017-05-23 Tom de Vries <tom@codesourcery.com>
26865 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
26868 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26870 * cgraphunit.c (symbol_table::process_new_functions): Update.
26871 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
26872 (inline_generate_summary): Rename to ...
26873 (ipa_fn_summary_generate): ... this one.
26874 (inline_read_summary): Rename to ...
26875 (ipa_fn_summary_read): ... this one.
26876 (inline_write_summary): Rename to ...
26877 (ipa_fn_summary_write): ... this one.
26878 (inline_free_summary): Rename to ...
26879 (ipa_free_fn_summary): ... this one.
26880 (pass_data_local_fn_summary, pass_local_fn_summary,
26881 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
26882 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
26883 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
26884 make_pass_ipa_fn_summary): New.
26885 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
26886 inline_write_summary, inline_free_summary): Remove.
26887 (ipa_free_fn_summary) : New.
26888 * ipa-inline.c (ipa_inline): Update.
26889 (pass_ipa_inline): Do not generate summaries.
26890 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
26892 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
26893 and add pass_ipa_fn_summary.
26894 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
26896 (make_pass_inline_parameters): Remove.
26898 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
26900 * omp-low.c (struct omp_context): Remove "default_kind" member.
26903 * omp-offload.c (execute_oacc_device_lower): Remove the
26904 parallelism dimensions function attributes for unparallelized
26905 OpenACC kernels constructs.
26907 2017-05-23 Martin Liska <mliska@suse.cz>
26909 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
26911 (cgraph_edge::make_speculative): Likewise.
26912 (cgraph_edge::resolve_speculation): Likewise.
26913 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
26914 (cgraph_node::dump): Likewise.
26915 * cgraph.h: Likewise.
26916 * cgraphunit.c (analyze_functions): Likewise.
26917 (symbol_table::compile): Likewise.
26918 * ipa-cp.c (print_all_lattices): Likewise.
26919 (determine_versionability): Likewise.
26920 (initialize_node_lattices): Likewise.
26921 (ipcp_verify_propagated_values): Likewise.
26922 (estimate_local_effects): Likewise.
26923 (update_profiling_info): Likewise.
26924 (create_specialized_node): Likewise.
26925 (perhaps_add_new_callers): Likewise.
26926 (decide_about_value): Likewise.
26927 (decide_whether_version_node): Likewise.
26928 (identify_dead_nodes): Likewise.
26929 (ipcp_store_bits_results): Likewise.
26930 * ipa-devirt.c (dump_targets): Likewise.
26931 (ipa_devirt): Likewise.
26932 * ipa-icf.c (sem_item::dump): Likewise.
26933 (sem_function::equals): Likewise.
26934 (sem_variable::equals): Likewise.
26935 (sem_item_optimizer::read_section): Likewise.
26936 (sem_item_optimizer::execute): Likewise.
26937 (congruence_class::dump): Likewise.
26938 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
26939 (dump_inline_summary): Likewise.
26940 (estimate_node_size_and_time): Likewise.
26941 (inline_analyze_function): Likewise.
26942 * ipa-inline-transform.c (inline_call): Likewise.
26943 * ipa-inline.c (report_inline_failed_reason): Likewise.
26944 (want_early_inline_function_p): Likewise.
26945 (edge_badness): Likewise.
26946 (update_edge_key): Likewise.
26947 (inline_small_functions): Likewise.
26948 * ipa-profile.c (ipa_profile): Likewise.
26949 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
26950 (ipa_make_edge_direct_to_target): Likewise.
26951 (remove_described_reference): Likewise.
26952 (ipa_impossible_devirt_target): Likewise.
26953 (propagate_controlled_uses): Likewise.
26954 (ipa_print_node_params): Likewise.
26955 (ipcp_transform_function): Likewise.
26956 * ipa-pure-const.c (pure_const_read_summary): Likewise.
26957 (propagate_pure_const): Likewise.
26958 * ipa-reference.c (generate_summary): Likewise.
26959 (read_write_all_from_decl): Likewise.
26960 (propagate): Likewise.
26961 (ipa_reference_read_optimization_summary): Likewise.
26962 * ipa-utils.c (ipa_merge_profiles): Likewise.
26963 * ipa.c (walk_polymorphic_call_targets): Likewise.
26964 (symbol_table::remove_unreachable_nodes): Likewise.
26965 (ipa_single_use): Likewise.
26966 * passes.c (execute_todo): Likewise.
26967 * predict.c (drop_profile): Likewise.
26968 * symtab.c (symtab_node::get_dump_name): New function.
26969 (symtab_node::dump_name): Likewise.
26970 (symtab_node::dump_asm_name): Likewise.
26971 (symtab_node::dump_references): Likewise.
26972 (symtab_node::dump_referring): Likewise.
26973 (symtab_node::dump_base): Likewise.
26974 (symtab_node::debug_symtab): Likewise.
26975 * tree-sra.c (convert_callers_for_node): Likewise.
26976 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
26977 * value-prof.c (init_node_map): Likewise.
26979 2017-05-23 Martin Liska <mliska@suse.cz>
26981 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
26982 and symtab_node::debug_symtab to symbol_table::debug.
26983 * cgraphunit.c (analyze_functions): Use the renamed function.
26984 (symbol_table::compile): Likewise.
26985 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
26986 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
26987 * passes.c (execute_todo): Likewise.
26988 * symtab.c (symbol_table::dump): New function.
26989 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
26991 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26993 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
26994 that nonconst implies exec.
26996 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26998 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
26999 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27000 (inline_edge_summary_vec): Turn into ...
27001 (ipa_call_summaries): ... this one.
27002 (redirect_to_unreachable, edge_set_predicate,
27003 evaluate_properties_for_edge, inline_summary_alloc,
27004 reset_ipa_call_summary, reset_inline_summary,
27005 inline_summary_t::duplicate): Update.
27006 (inline_edge_duplication_hook): Turn to ...
27007 (ipa_call_summary_t::duplicate): ... this one.
27008 (inline_edge_removal_hook): Turn to ...
27009 (ipa_call_summary_t::remove): ... this one.
27010 (dump_inline_edge_summary): Turn to ...
27011 (dump_ipa_call_summary): ... this one.
27012 (estimate_function_body_sizes): Update.
27013 (inline_update_callee_summaries): Update.
27014 (remap_edge_change_prob): Update.
27015 (remap_edge_summaries): Update.
27016 (inline_merge_summary): Update.
27017 (do_estimate_edge_time): Update.
27018 (inline_generate_summary): Update.
27019 (inline_read_section): Update.
27020 (inline_read_summary): Update.
27021 (inline_free_summary): Update.
27022 * ipa-inline.c (can_inline_edge_p): Update.
27023 (compute_inlined_call_time): Update.
27024 (want_inline_small_function_p): Update.
27025 (edge_badness): Update.
27026 (early_inliner): Update.
27027 * ipa-inline.h (inline_edge_summary): Turn to ...
27028 (ipa_call_summary): ... this one.
27029 (ipa_call_summary_t): New class.
27030 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27031 (ipa_call_summaries): New.
27032 (inline_edge_summary): Remove.
27033 (estimate_edge_growth): Update.
27034 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27035 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27036 * ipa-split.c (execute_split_functions): Update.
27037 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27039 2017-05-23 Tom de Vries <tom@codesourcery.com>
27041 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
27042 attributes): Document rdrand effective target.
27044 2017-05-23 Tom de Vries <tom@codesourcery.com>
27046 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
27047 attributes): Sort alphabetically.
27049 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
27051 * config/avr/genmultilib.awk: Use gsub instead of gensub.
27053 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
27056 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
27057 V2DF/V2DI splat into two separate patterns, one that handles
27058 registers, and the other that only handles memory. Drop support
27059 for splatting from a GPR on ISA 2.07 and then splitting the
27060 splat into direct move and splat.
27061 (vsx_splat_<mode>_reg): Likewise.
27062 (vsx_splat_<mode>_mem): Likewise.
27064 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
27066 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
27068 2017-05-22 Jakub Jelinek <jakub@redhat.com>
27070 PR middle-end/80809
27071 * omp-low.c (finish_taskreg_remap): New function.
27072 (finish_taskreg_scan): If unit size of ctx->record_type
27073 is non-constant, unshare the size expression and replace
27074 decls in it with possible outer var refs.
27076 PR middle-end/80809
27077 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
27078 GOVD_SHARED rather than GOVD_PRIVATE with it.
27079 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
27080 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
27082 PR middle-end/80853
27083 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
27084 as last argument to build_outer_var_ref for pointer bases of array
27085 section reductions.
27087 2017-05-19 Martin Sebor <msebor@redhat.com>
27089 * print-tree.c (print_node): Print DECL_READ_P flag.
27091 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27093 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
27094 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
27095 * cgraph.c: Likewise.
27096 * cgraphunit.c: Likewise.
27097 * gengtype.c: Likewise.
27098 * ipa-cp.c: Likewise.
27099 * ipa-devirt.c: Likewise.
27100 * ipa-icf.c: Likewise.
27101 * ipa-predicate.c: Likewise.
27102 * ipa-profile.c: Likewise.
27103 * ipa-prop.c: Likewise.
27104 * ipa-split.c: Likewise.
27106 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
27107 edge_predicate_pool, dump_inline_hints,
27108 inline_summary::account_size_time, redirect_to_unreachable,
27109 edge_set_predicate, set_hint_predicate,
27110 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
27111 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
27112 inline_summary_t::remove, remap_hint_predicate_after_duplication,
27113 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
27114 ipa_call_summary_t::remove, initialize_growth_caches,
27115 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
27116 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
27117 mark_modified, unmodified_parm_1, unmodified_parm,
27118 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
27119 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
27120 compute_bb_predicates, will_be_nonconstant_expr_predicate,
27121 will_be_nonconstant_predicate, record_modified_bb_info,
27122 get_minimal_bb, record_modified, param_change_prob,
27123 phi_result_unknown_predicate, predicate_for_phi_result,
27124 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
27125 estimate_function_body_sizes, compute_inline_parameters,
27126 compute_inline_parameters_for_curren, pass_data_inline_parameters,
27127 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
27128 inline_update_callee_summaries, remap_edge_change_prob,
27129 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
27130 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
27131 inline_analyze_function, inline_summary_t::insert,
27132 inline_generate_summary, read_ipa_call_summary, inline_read_section,
27133 inline_read_summary, write_ipa_call_summary, inline_write_summary,
27134 inline_free_summary): Move to ipa-fnsummary.h
27135 (predicate_t): Remove.
27136 * ipa-fnsummary.c: New file.
27137 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
27138 (enum inline_hints_vals, inline_hints, agg_position_info,
27139 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
27140 inline_summaries, ipa_call_summary, ipa_call_summary_t,
27141 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
27142 dump_inline_summary, dump_inline_hints, inline_generate_summary,
27143 inline_read_summary, inline_write_summary, inline_free_summary,
27144 inline_analyze_function, initialize_inline_failed,
27145 inline_merge_summary, inline_update_overall_summary,
27146 compute_inline_parameters): Move to ipa-fnsummary.h
27147 * ipa-fnsummary.h: New file.
27148 * ipa-inline-transform.h: Include ipa-inline.h.
27149 * ipa-inline.c: LIkewise.
27151 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27153 * ipa-inline.c (edge_badness): Use inlined_time instead of
27154 inline_summaries->get.
27156 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27158 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
27160 2017-05-22 Nathan Sidwell <nathan@acm.org>
27162 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
27163 (fdump-lang): Document 'raw' option.
27164 * dumpfile.h (TDI_tu): Delete.
27165 * dumpfile.c (dump_files): Remove translation-unit.
27166 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
27168 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
27170 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
27171 command option from $(AWK) call.
27172 * config/avr/genmultilib.awk: Simplify and rewrite so that it
27173 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
27174 [FORMAT]: Remove handling of variable.
27175 * config/avr/t-multilib: Regenerate.
27177 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27179 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
27181 (dump_inline_summary): Do not print self_time.
27182 (estimate_function_body_sizes): Do not set self_time.
27183 (compute_inline_parameters): Likewise.
27184 (inline_read_section, inline_write_summary): Do not stream self_time.
27185 * ipa-inline.h (inline_summary): Drop self_time.
27187 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27189 * ipa-inline-analysis.c (account_size_time): Rename to ...
27190 (inline_summary::account_size_time): ... this one.
27191 (reset_ipa_call_summary): Turn to ...
27192 (ipa_call_summary::reset): ... this one.
27193 (reset_inline_summary): Turn to ...
27194 (inline_summary::reset): ... this one.
27195 (inline_summary_t::remove): Update.
27196 (inline_summary_t::duplicate): Update.
27197 (ipa_call_summary_t::remove): Update.
27198 (dump_inline_summary): Update.
27199 (estimate_function_body_sizes): Update.
27200 (compute_inline_parameters): Update.
27201 (estimate_node_size_and_time): Update.
27202 (inline_merge_summary): Update.
27203 (inline_update_overall_summary): Update.
27204 (inline_read_section): Update.
27205 (inline_write_summary): Update.
27206 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
27207 add account_size_time and reset member functions.
27208 (ipa_call_summary): Add reset function.
27209 * ipa-predicate.h (predicate::operator &): Constify.
27211 2017-05-22 Richard Biener <rguenther@suse.de>
27213 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
27215 2017-05-19 Jason Merrill <jason@redhat.com>
27217 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
27219 2017-05-19 Marek Polacek <polacek@redhat.com>
27222 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
27223 TYPE_OVERFLOW_WRAPS checks.
27225 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
27227 * tree-core.h (enum omp_clause_default_kind): Add
27228 "OMP_CLAUSE_DEFAULT_PRESENT".
27229 * tree-pretty-print.c (dump_omp_clause): Handle it.
27230 * gimplify.c (enum gimplify_omp_var_data): Add
27231 "GOVD_MAP_FORCE_PRESENT".
27232 (gimplify_adjust_omp_clauses_1): Map it to
27233 "GOMP_MAP_FORCE_PRESENT".
27234 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
27236 * gimplify.c (oacc_default_clause): Clarify.
27238 2017-05-19 Nathan Sidwell <nathan@acm.org>
27240 LANG_HOOK_REGISTER_DUMPS
27241 * toplev.c (general_init): Call register dump lang hook.
27242 * doc/invoke.texi: Document -fdump-lang option family.
27243 * dumpfile.c (dump_files): Remove class dump here.
27244 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
27245 * dumpfile.h (tree_dump_index): Remove TDI_class.
27246 * langhooks-def.h (lhd_register_dumps): Declare.
27247 (LANG_HOOKS_REGISTER_DUMPS): Define.
27248 (LANG_HOOKS_INITIALIZER): Add it.
27249 * langhooks.c (lhd_register_dumps): Define.
27250 * langhooks.h (struct lang_hooks): Add register_dumps.
27252 2017-05-19 Nathan Sidwell <nathan@acm.org>
27254 * context.h (context::set_passes): New.
27255 * context.c (context::context): Do not create pass manager.
27256 * toplev.c (general_init): Create pass manager here.
27258 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
27260 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
27261 use this splitter if two add or or instructions would also work for
27262 the constant we want to generate.
27264 2017-05-19 Richard Biener <rguenther@suse.de>
27267 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
27268 predicate evaluation.
27270 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27272 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
27274 * ipa-inline.c (want_inline_small_function_p): Do not cast to
27277 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27279 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
27280 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27281 (inline_edge_summary_vec): Turn into ...
27282 (ipa_call_summaries): ... this one.
27283 (redirect_to_unreachable, edge_set_predicate,
27284 evaluate_properties_for_edge, inline_summary_alloc,
27285 reset_ipa_call_summary, reset_inline_summary,
27286 inline_summary_t::duplicate): Update.
27287 (inline_edge_duplication_hook): Turn to ...
27288 (ipa_call_summary_t::duplicate): ... this one.
27289 (inline_edge_removal_hook): Turn to ...
27290 (ipa_call_summary_t::remove): ... this one.
27291 (dump_inline_edge_summary): Turn to ...
27292 (dump_ipa_call_summary): ... this one.
27293 (estimate_function_body_sizes): Update.
27294 (inline_update_callee_summaries): Update.
27295 (remap_edge_change_prob): Update.
27296 (remap_edge_summaries): Update.
27297 (inline_merge_summary): Update.
27298 (do_estimate_edge_time): Update.
27299 (inline_generate_summary): Update.
27300 (inline_read_section): Update.
27301 (inline_read_summary): Update.
27302 (inline_free_summary): Update.
27303 * ipa-inline.c (can_inline_edge_p): Update.
27304 (compute_inlined_call_time): Update.
27305 (want_inline_small_function_p): Update.
27306 (edge_badness): Update.
27307 (early_inliner): Update.
27308 * ipa-inline.h (inline_edge_summary): Turn to ...
27309 (ipa_call_summary): ... this one.
27310 (ipa_call_summary_t): New class.
27311 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27312 (ipa_call_summaries): New.
27313 (inline_edge_summary): Remove.
27314 (estimate_edge_growth): Update.
27315 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27316 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27317 * ipa-split.c (execute_split_functions): Update.
27318 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27320 2017-05-19 Richard Biener <rguenther@suse.de>
27322 PR middle-end/80764
27323 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
27325 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
27327 * config/rs6000/rs6000.c (struct machine_function): Add field
27328 fpr_is_wrapped_separately.
27329 (rs6000_get_separate_components): Use 64 components. Handle the
27330 new FPR components.
27331 (rs6000_components_for_bb): Handle the FPR components.
27332 (rs6000_emit_prologue_components): Handle the FPR components.
27333 (rs6000_emit_epilogue_components): Handle the FPR components.
27334 (rs6000_set_handled_components): Handle the FPR components.
27335 (rs6000_emit_prologue): Don't output prologue code for those FPRs
27336 that are already separately shrink-wrapped.
27337 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
27338 that are already separately shrink-wrapped.
27340 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27343 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
27346 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
27347 (define_peephole2 for Altivec d-form load): Add peepholes to catch
27348 cases where the register allocator uses a move and an offsettable
27349 memory operation to/from a FPR register on ISA 2.06/2.07.
27350 (define_peephole2 for Altivec d-form store): Likewise.
27352 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
27355 * config/i386/mmx.md (*mov<mode>_internal): Enable
27356 alternatives 11, 12, 13 and 14 also for 32bit targets.
27357 Remove alternatives 15, 16, 17 and 18.
27358 * config/i386/sse.md (vec_concatv2di): Change
27359 alternative (!x, *y) to (x, ?!*Yn).
27361 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
27363 * dumpfile.h (enum dump_kind): Remove stray comma.
27365 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27367 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
27368 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
27369 predicate::num_conditions
27370 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
27371 (CHANGED): turn into predicate::changed.
27372 (agg_position_info): Move to ipa-predicate.h
27373 (add_condition, predicate::add_clause, predicate::operator &=,
27374 predicate::or_with, predicate::evaluate, predicate::probability,
27375 dump_condition, dump_clause, predicate::dump,
27376 predicate::remap_after_duplication, predicate::remap_after_inlining,
27377 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
27378 (evaluate_conditions_for_known_args): Update.
27379 (set_cond_stmt_execution_predicate): Update.
27380 * ipa-inline.h: Include ipa-predicate.h
27381 (condition, inline_param_summary, conditions, agg_position_info,
27382 predicate): Move to ipa-predicate.h
27383 * ipa-predicate.c: New file.
27384 * ipa-predicate.h: New file.
27386 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
27388 * final.c (leaf_function_p): Check we are not in a sequence.
27390 2017-05-18 Martin Liska <mliska@suse.cz>
27392 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
27393 * dumpfile.c (dump_register): Use new enum dump_kind.
27394 (get_dump_file_name): Likewise.
27395 (dump_enable_all): Likewise.
27396 (dump_switch_p_1): Likewise.
27397 (enable_rtl_dump_file): Remove usage of TDF_RTL.
27398 * dumpfile.h (enum dump_kind): New enum type.
27399 (struct dump_file_info): Create constructor and
27400 format fields and comments.
27401 * passes.c (pass_manager::register_one_dump_file):
27403 * statistics.c (statistics_early_init): Likewise.
27404 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
27405 TDF_TREE with TDF_SLIM.
27406 (gather_memory_references_ref): Likewise.
27408 2017-05-18 Martin Liska <mliska@suse.cz>
27410 * vec.h (struct vnull): Use it.
27412 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27414 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
27415 (true_predicate, false_predicate, true_predicate_p,
27416 false_predicate_p): Remove.
27417 (single_cond_predicate, not_inlined_predicate): Turn to member function
27419 (add_condition): Update.
27420 (add_clause): Turn to...
27421 (predicate::add_clause): ... this one; update; allow passing NULL
27423 (and_predicates): Turn to ...
27424 (predicate::operator &=): ... this one.
27425 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
27426 (or_predicates): Turn to ...
27427 (predicate::or_with): ... this one.
27428 (evaluate_predicate): Turn to ...
27429 (predicate::evaluate): ... this one.
27430 (predicate_probability): Turn to ...
27431 (predicate::probability): ... this one.
27432 (dump_condition): Update.
27433 (dump_predicate): Turn to ...
27434 (predicate::dump): ... this one.
27435 (account_size_time): Update.
27436 (edge_set_predicate): Update.
27437 (set_hint_predicate): UPdate.
27438 (evaluate_conditions_for_known_args): Update.
27439 (evaluate_properties_for_edge): Update.
27440 (remap_predicate_after_duplication): Turn to...
27441 (predicate::remap_after_duplication): ... this one.
27442 (remap_hint_predicate_after_duplication): Update.
27443 (inline_summary_t::duplicate): UPdate.
27444 (dump_inline_edge_summary): Update.
27445 (dump_inline_summary): Update.
27446 (set_cond_stmt_execution_predicate): Update.
27447 (set_switch_stmt_execution_predicate): Update.
27448 (compute_bb_predicates): Update.
27449 (will_be_nonconstant_expr_predicate): Update.
27450 (will_be_nonconstant_predicate): Update.
27451 (phi_result_unknown_predicate): Update.
27452 (predicate_for_phi_result): Update.
27453 (array_index_predicate): Update.
27454 (estimate_function_body_sizes): Update.
27455 (estimate_node_size_and_time): Update.
27456 (estimate_ipcp_clone_size_and_time): Update.
27457 (remap_predicate): Rename to ...
27458 (predicate::remap_after_inlining): ... this one.
27459 (remap_hint_predicate): Update.
27460 (inline_merge_summary): Update.
27461 (inline_update_overall_summary): Update.
27462 (estimate_size_after_inlining): Update.
27463 (read_predicate): Rename to ...
27464 (predicate::stream_in): ... this one.
27465 (read_inline_edge_summary): Update.
27466 (write_predicate): Rename to ...
27467 (predicate::stream_out): ... this one.
27468 (write_inline_edge_summary): Update.
27469 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
27470 (clause_t): Turn to uint32_t
27471 (predicate): Turn to class; implement constructor and operators
27473 (size_time_entry): Update.
27474 (inline_summary): Update.
27475 (inline_edge_summary): Update.
27477 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
27479 * fold-const.c (fold_binary_loc): Move transformation...
27480 * match.pd (C - X CMP X): ... here.
27482 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
27484 * config/sparc/sparc.c (sparc_option_override): Set function
27485 alignment for -mcpu=niagara7 to 64 to match the I$ line.
27486 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
27488 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
27490 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
27492 2017-05-18 Marek Polacek <polacek@redhat.com>
27495 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
27496 (pass_ubsan::execute): Call gimple_assign_single_p instead of
27497 gimple_assign_load_p.
27499 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
27501 PR middle-end/80692
27502 * real.c (do_compare): Give decimal_do_compare preference over
27503 comparing just the signs.
27505 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
27507 * doc/md.texi (Canonicalization of Instructions): Describe the
27508 canonical form of instructions that inherently set a condition
27511 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
27513 PR middle-end/80775
27514 * tree-cfg.c: Move deletion of unreachable case statements to after
27515 the merging of consecutive case labels.
27517 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
27519 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
27520 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
27521 restoring of callee-saved registers.
27523 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
27525 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
27526 * config/visium/visium.c (single_set_and_flags): Likewise.
27527 * config/visium/visium.md (Substitutions): Likewise.
27529 2017-05-17 Martin Liska <mliska@suse.cz>
27531 * cfg.c: Introduce dump_flags_t type and
27532 use it instead of int type.
27534 * cfghooks.c: Likewise.
27535 * cfghooks.h (struct cfg_hooks): Likewise.
27536 * cfgrtl.c: Likewise.
27537 * cfgrtl.h: Likewise.
27538 * cgraph.c (cgraph_node::get_body): Likewise.
27539 * coretypes.h: Likewise.
27540 * domwalk.c: Likewise.
27541 * domwalk.h: Likewise.
27542 * dumpfile.c (struct dump_option_value_info): Likewise.
27543 (dump_enable_all): Likewise.
27544 (dump_switch_p_1): Likewise.
27545 (opt_info_switch_p): Likewise.
27546 * dumpfile.h (enum tree_dump_index): Likewise.
27547 (struct dump_file_info): Likewise.
27548 * genemit.c: Likewise.
27549 * generic-match-head.c: Likewise.
27550 * gengtype.c (open_base_files): Likewise.
27551 * gimple-pretty-print.c: Likewise.
27552 * gimple-pretty-print.h: Likewise.
27553 * graph.c (print_graph_cfg): Likewise.
27554 * graphite-scop-detection.c (dot_all_sese): Likewise.
27555 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
27556 * loop-unroll.c (report_unroll): Likewise.
27557 * passes.c (pass_manager::register_one_dump_file): Likewise.
27558 * print-tree.c: Likewise.
27559 * statistics.c: Likewise.
27560 * tree-cfg.c: Likewise.
27561 * tree-cfg.h: Likewise.
27562 * tree-dfa.c: Likewise.
27563 * tree-dfa.h: Likewise.
27564 * tree-dump.c (dump_function): Likewise.
27565 * tree-dump.h (struct dump_info): Likewise.
27566 * tree-pretty-print.c: Likewise.
27567 * tree-pretty-print.h: Likewise.
27568 * tree-ssa-live.c: Likewise.
27569 * tree-ssa-live.h: Likewise.
27570 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
27571 * tree-vect-loop.c: Likewise.
27572 * tree-vect-slp.c: Likewise.
27574 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
27575 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27577 PR tree-optimization/80457
27578 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
27579 of all arguments to a statement as scalar_to_vec operations.
27580 (vectorizable_call): Adjust call to vect_model_simple_cost for
27582 (vectorizable_conversion): Likewise.
27583 (vectorizable_assignment): Likewise.
27584 (vectorizable_shift): Likewise.
27585 (vectorizable_operation): Likewise.
27586 (vectorizable_comparison): Likewise.
27587 (vect_is_simple_cond): Record the def types for operands.
27588 (vectorizable_condition): Likewise, call vect_model_simple_cost.
27589 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
27590 for statement argument count.
27592 2017-05-16 Carl Love <cel@us.ibm.com>
27594 * config/rs6000/rs6000-c: Add support for built-in functions
27595 vector unsigned long long vec_bperm (vector unsigned long long,
27596 vector unsigned char)
27597 vector signed long long vec_mule (vector signed int,
27599 vector unsigned long long vec_mule (vector unsigned int,
27600 vector unsigned int)
27601 vector signed long long vec_mulo (vector signed int,
27603 vector unsigned long long vec_mulo (vector unsigned int,
27604 vector unsigned int)
27605 vector signed char vec_sldw (vector signed char,
27606 vector signed char,
27608 vector unsigned char vec_sldw (vector unsigned char,
27609 vector unsigned char,
27611 vector signed short vec_sldw (vector signed short,
27612 vector signed short,
27614 vector unsigned short vec_sldw (vector unsigned short,
27615 vector unsigned short,
27617 vector signed int vec_sldw (vector signed int,
27620 vector unsigned int vec_sldw (vector unsigned int,
27621 vector unsigned int,
27623 vector signed long long vec_sldw (vector signed long long,
27624 vector signed long long,
27626 vector unsigned long long vec_sldw (vector unsigned long long,
27627 vector unsigned long long,
27629 * config/rs6000/rs6000-c: Add support for built-in functions
27630 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
27631 * config/rs6000/altivec.h: Add defintion for vec_sldw.
27632 * doc/extend.texi: Update the built-in documentation for the
27633 new built-in functions.
27635 2017-05-16 Marek Polacek <polacek@redhat.com>
27639 * tree.c (save_expr): Don't fold the expression.
27641 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
27643 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
27644 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
27645 and (?*y,m). Update insn attributes.
27647 2017-05-16 Martin Liska <mliska@suse.cz>
27649 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
27650 flags argument of print_gimple_stmt, print_gimple_expr,
27651 print_generic_stmt and print_generic_expr.
27652 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
27653 * coretypes.h: Likewise.
27654 * except.c (dump_eh_tree): Likewise.
27655 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
27656 * gimple-pretty-print.h: Likewise.
27657 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
27658 (backprop::push_to_worklist): Likewise.
27659 (backprop::pop_from_worklist): Likewise.
27660 (backprop::process_use): Likewise.
27661 (backprop::intersect_uses): Likewise.
27662 (note_replacement): Likewise.
27663 * gimple-ssa-store-merging.c
27664 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
27665 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
27666 (pass_store_merging::execute): Likewise.
27667 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
27668 (ssa_base_cand_dump_callback): Likewise.
27669 (dump_incr_vec): Likewise.
27670 (replace_refs): Likewise.
27671 (replace_mult_candidate): Likewise.
27672 (create_add_on_incoming_edge): Likewise.
27673 (create_phi_basis): Likewise.
27674 (insert_initializers): Likewise.
27675 (all_phi_incrs_profitable): Likewise.
27676 (introduce_cast_before_cand): Likewise.
27677 (replace_one_candidate): Likewise.
27678 * gimplify.c (gimplify_expr): Likewise.
27679 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
27680 (set_rename): Likewise.
27681 (rename_uses): Likewise.
27682 (copy_loop_phi_nodes): Likewise.
27683 (add_close_phis_to_merge_points): Likewise.
27684 (copy_loop_close_phi_args): Likewise.
27685 (copy_cond_phi_args): Likewise.
27686 (graphite_copy_stmts_from_block): Likewise.
27687 (translate_pending_phi_nodes): Likewise.
27688 * graphite-poly.c (print_pdr): Likewise.
27689 (dump_gbb_cases): Likewise.
27690 (dump_gbb_conditions): Likewise.
27691 (print_scop_params): Likewise.
27692 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
27693 (build_cross_bb_scalars_use): Likewise.
27694 (gather_bbs::before_dom_children): Likewise.
27695 * hsa-dump.c (dump_hsa_immed): Likewise.
27696 * ipa-cp.c (print_ipcp_constant_value): Likewise.
27697 (get_replacement_map): Likewise.
27698 * ipa-inline-analysis.c (dump_condition): Likewise.
27699 (estimate_function_body_sizes): Likewise.
27700 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
27701 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
27702 * ipa-prop.c (ipa_dump_param): Likewise.
27703 (ipa_print_node_jump_functions_for_edge): Likewise.
27704 (ipa_modify_call_arguments): Likewise.
27705 (ipa_modify_expr): Likewise.
27706 (ipa_dump_param_adjustments): Likewise.
27707 (ipa_dump_agg_replacement_values): Likewise.
27708 (ipcp_modif_dom_walker::before_dom_children): Likewise.
27709 * ipa-pure-const.c (check_stmt): Likewise.
27710 (pass_nothrow::execute): Likewise.
27711 * ipa-split.c (execute_split_functions): Likewise.
27712 * omp-offload.c (dump_oacc_loop_part): Likewise.
27713 (dump_oacc_loop): Likewise.
27714 * trans-mem.c (tm_log_emit): Likewise.
27715 (tm_memopt_accumulate_memops): Likewise.
27716 (dump_tm_memopt_set): Likewise.
27717 (dump_tm_memopt_transform): Likewise.
27718 * tree-cfg.c (gimple_verify_flow_info): Likewise.
27719 (print_loop): Likewise.
27720 * tree-chkp-opt.c (chkp_print_addr): Likewise.
27721 (chkp_gather_checks_info): Likewise.
27722 (chkp_get_check_result): Likewise.
27723 (chkp_remove_check_if_pass): Likewise.
27724 (chkp_use_outer_bounds_if_possible): Likewise.
27725 (chkp_reduce_bounds_lifetime): Likewise.
27726 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
27727 (chkp_mark_completed_bounds): Likewise.
27728 (chkp_register_incomplete_bounds): Likewise.
27729 (chkp_mark_invalid_bounds): Likewise.
27730 (chkp_maybe_copy_and_register_bounds): Likewise.
27731 (chkp_build_returned_bound): Likewise.
27732 (chkp_get_bound_for_parm): Likewise.
27733 (chkp_build_bndldx): Likewise.
27734 (chkp_get_bounds_by_definition): Likewise.
27735 (chkp_generate_extern_var_bounds): Likewise.
27736 (chkp_get_bounds_for_decl_addr): Likewise.
27737 * tree-chrec.c (chrec_apply): Likewise.
27738 * tree-data-ref.c (dump_data_reference): Likewise.
27739 (dump_subscript): Likewise.
27740 (dump_data_dependence_relation): Likewise.
27741 (analyze_overlapping_iterations): Likewise.
27742 * tree-inline.c (expand_call_inline): Likewise.
27743 (tree_function_versioning): Likewise.
27744 * tree-into-ssa.c (dump_defs_stack): Likewise.
27745 (dump_currdefs): Likewise.
27746 (dump_names_replaced_by): Likewise.
27747 (dump_update_ssa): Likewise.
27748 (update_ssa): Likewise.
27749 * tree-object-size.c (pass_object_sizes::execute): Likewise.
27750 * tree-parloops.c (build_new_reduction): Likewise.
27751 (try_create_reduction_list): Likewise.
27752 (ref_conflicts_with_region): Likewise.
27753 (oacc_entry_exit_ok_1): Likewise.
27754 (oacc_entry_exit_single_gang): Likewise.
27755 * tree-pretty-print.h: Likewise.
27756 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
27757 (get_scalar_evolution): Likewise.
27758 (add_to_evolution): Likewise.
27759 (get_loop_exit_condition): Likewise.
27760 (analyze_evolution_in_loop): Likewise.
27761 (analyze_initial_condition): Likewise.
27762 (analyze_scalar_evolution): Likewise.
27763 (instantiate_scev): Likewise.
27764 (number_of_latch_executions): Likewise.
27765 (gather_chrec_stats): Likewise.
27766 (final_value_replacement_loop): Likewise.
27767 (scev_const_prop): Likewise.
27768 * tree-sra.c (dump_access): Likewise.
27769 (disqualify_candidate): Likewise.
27770 (create_access): Likewise.
27771 (reject): Likewise.
27772 (maybe_add_sra_candidate): Likewise.
27773 (create_access_replacement): Likewise.
27774 (analyze_access_subtree): Likewise.
27775 (analyze_all_variable_accesses): Likewise.
27776 (sra_modify_assign): Likewise.
27777 (initialize_constant_pool_replacements): Likewise.
27778 (find_param_candidates): Likewise.
27779 (decide_one_param_reduction): Likewise.
27780 (replace_removed_params_ssa_names): Likewise.
27781 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
27782 * tree-ssa-copy.c (dump_copy_of): Likewise.
27783 (copy_prop_visit_cond_stmt): Likewise.
27784 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
27785 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
27786 (record_equivalences_from_stmt): Likewise.
27787 * tree-ssa-dse.c (compute_trims): Likewise.
27788 (delete_dead_call): Likewise.
27789 (delete_dead_assignment): Likewise.
27790 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
27791 (forward_propagate_into_cond): Likewise.
27792 (pass_forwprop::execute): Likewise.
27793 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
27794 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
27796 (move_computations_worker): Likewise.
27797 (execute_sm): Likewise.
27798 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
27799 (remove_exits_and_undefined_stmts): Likewise.
27800 (remove_redundant_iv_tests): Likewise.
27801 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
27802 (adjust_iv_update_pos): Likewise.
27803 * tree-ssa-math-opts.c (bswap_replace): Likewise.
27804 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
27805 (value_replacement): Likewise.
27806 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
27807 * tree-ssa-pre.c (print_pre_expr): Likewise.
27808 (get_representative_for): Likewise.
27809 (create_expression_by_pieces): Likewise.
27810 (insert_into_preds_of_block): Likewise.
27811 (eliminate_insert): Likewise.
27812 (eliminate_dom_walker::before_dom_children): Likewise.
27813 (eliminate): Likewise.
27814 (remove_dead_inserted_code): Likewise.
27815 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
27816 * tree-ssa-reassoc.c (get_rank): Likewise.
27817 (eliminate_duplicate_pair): Likewise.
27818 (eliminate_plus_minus_pair): Likewise.
27819 (eliminate_not_pairs): Likewise.
27820 (undistribute_ops_list): Likewise.
27821 (eliminate_redundant_comparison): Likewise.
27822 (update_range_test): Likewise.
27823 (optimize_range_tests_var_bound): Likewise.
27824 (optimize_vec_cond_expr): Likewise.
27825 (rewrite_expr_tree): Likewise.
27826 (rewrite_expr_tree_parallel): Likewise.
27827 (linearize_expr): Likewise.
27828 (break_up_subtract): Likewise.
27829 (linearize_expr_tree): Likewise.
27830 (attempt_builtin_powi): Likewise.
27831 (attempt_builtin_copysign): Likewise.
27832 (transform_stmt_to_copy): Likewise.
27833 (transform_stmt_to_multiply): Likewise.
27834 (dump_ops_vector): Likewise.
27835 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
27836 (print_scc): Likewise.
27837 (set_ssa_val_to): Likewise.
27838 (visit_reference_op_store): Likewise.
27839 (visit_use): Likewise.
27840 (sccvn_dom_walker::before_dom_children): Likewise.
27841 (run_scc_vn): Likewise.
27842 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
27844 (expr_hash_elt::print): Likewise.
27845 (const_and_copies::pop_to_marker): Likewise.
27846 (const_and_copies::record_const_or_copy_raw): Likewise.
27847 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
27848 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
27849 (dump_predicates): Likewise.
27850 (find_uninit_use): Likewise.
27851 (warn_uninitialized_phi): Likewise.
27852 (pass_late_warn_uninitialized::execute): Likewise.
27853 * tree-ssa.c (verify_vssa): Likewise.
27854 (verify_ssa): Likewise.
27855 (maybe_optimize_var): Likewise.
27856 * tree-vrp.c (dump_value_range): Likewise.
27857 (dump_all_value_ranges): Likewise.
27858 (dump_asserts_for): Likewise.
27859 (register_edge_assert_for_2): Likewise.
27860 (vrp_visit_cond_stmt): Likewise.
27861 (vrp_visit_switch_stmt): Likewise.
27862 (vrp_visit_stmt): Likewise.
27863 (vrp_visit_phi_node): Likewise.
27864 (simplify_cond_using_ranges_1): Likewise.
27865 (fold_predicate_in): Likewise.
27866 (evrp_dom_walker::before_dom_children): Likewise.
27867 (evrp_dom_walker::push_value_range): Likewise.
27868 (evrp_dom_walker::pop_value_range): Likewise.
27869 (execute_early_vrp): Likewise.
27871 2017-05-16 Richard Biener <rguenther@suse.de>
27873 * dwarf2out.c (loc_list_from_tree_1): Do not create
27874 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
27876 2017-05-16 Richard Biener <rguenther@suse.de>
27878 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
27880 (note_variable_value_in_expr): If we resolved the decl ref
27881 do not push to the stack.
27883 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
27885 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
27886 operations in fast-math mode.
27887 (vaddq_f16): Likewise.
27888 (vmul_f16): Likewise.
27889 (vmulq_f16): Likewise.
27890 (vsub_f16): Likewise.
27891 (vsubq_f16): Likewise.
27892 * config/arm/neon.md (add<mode>3): New.
27894 (fma:<VH:mode>3): New. Also remove outdated comment.
27897 2017-05-16 Martin Liska <mliska@suse.cz>
27901 * ipa-devirt.c (warn_types_mismatch): Fix typo.
27902 (odr_types_equivalent_p): Likewise.
27904 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
27906 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
27908 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
27911 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
27912 non-interunit SSE move alternatives with '?'.
27913 (zero-extendsidi peephole2): New peephole to skip intermediate
27914 general register in SSE zero-extend sequence.
27916 2017-05-15 Jeff Law <law@redhat.com>
27918 * reorg.c (relax_delay_slots): Create a new variable to hold
27919 the temporary target rather than clobbering TARGET_LABEL.
27921 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
27922 missing argument to extract_bit_field call.
27923 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
27925 2017-05-15 Martin Liska <mliska@suse.cz>
27928 * gcc.c (process_command): Do not allow empty argument of -o option.
27930 2017-05-15 Renlin Li <renlin.li@arm.com>
27932 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
27933 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
27934 * config/aarch64/constraints.md (Usf): Add long call check.
27935 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
27936 (call_value): Likewise.
27937 (sibcall): Likewise.
27938 (sibcall_value): Likewise.
27940 (call_value_insn): New.
27941 (sibcall_insn): Update rtx pattern.
27942 (sibcall_value_insn): Likewise.
27943 (call_internal): Remove.
27944 (call_value_internal): Likewise.
27945 (sibcall_internal): Likewise.
27946 (sibcall_value_internal): Likewise.
27947 (call_reg): Likewise.
27948 (call_symbol): Likewise.
27949 (call_value_reg): Likewise.
27950 (call_value_symbol): Likewise.
27952 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
27955 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
27957 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
27959 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
27960 compatible with CCGOCmode and with CCZmode.
27962 2017-05-14 Martin Sebor <msebor@redhat.com>
27964 PR middle-end/77671
27965 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
27966 (gimple_fold_builtin_snprintf): Same.
27967 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
27968 (gimple_fold_builtin_snprintf): Same.
27969 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
27970 of character types.
27971 (is_call_safe): New function.
27972 (try_substitute_return_value): Call it.
27973 (try_simplify_call): New function.
27974 (pass_sprintf_length::handle_gimple_call): Call it.
27976 2017-05-14 Martin Sebor <msebor@redhat.com>
27978 PR middle-end/80669
27979 * builtins.c (expand_builtin_stpncpy): Simplify.
27981 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
27983 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
27984 * config/i386/i386.h
27985 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
27986 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
27987 (struct machine_function): Add new members call_ms2sysv,
27988 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
27989 (struct machine_frame_state): New fields sp_realigned and
27990 sp_realigned_offset.
27991 * config/i386/i386.c
27992 (enum xlogue_stub): New enum.
27993 (enum xlogue_stub_sets): New enum.
27994 (class xlogue_layout): New class.
27995 (struct ix86_frame): New fields stack_realign_allocate_offset,
27996 stack_realign_offset and outlined_save_offset. Modify comments to
27997 detail stack layout when using out-of-line stubs.
27998 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
27999 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
28000 -mcall-ms2sysv-xlogues.
28001 (stub_managed_regs): New static variable.
28002 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
28003 registers managed by out-of-line stub.
28004 (disable_call_ms2sysv_xlogues): New function.
28005 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
28006 m->call_ms2sysv when appropriate and compute frame layout for
28008 (sp_valid_at, fp_valid_at): New inline functions.
28009 (choose_basereg): New function.
28010 (choose_baseaddr): Add align parameter, use choose_basereg and modify
28012 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
28013 Use align parameter of choose_baseaddr to generated aligned SSE movs
28015 (pro_epilogue_adjust_stack): Modify to track
28016 machine_frame_state::sp_realigned.
28017 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
28018 (ix86_nsaved_sseregs): Likewise.
28019 (ix86_emit_save_regs): Likewise.
28020 (ix86_emit_save_regs_using_mov): Likewise.
28021 (ix86_emit_save_sse_regs_using_mov): Likewise.
28022 (get_scratch_register_on_entry): Likewise.
28023 (gen_frame_set): New function.
28024 (gen_frame_load): Likewise.
28025 (gen_frame_store): Likewise.
28026 (emit_outlined_ms2sysv_save): Likewise.
28027 (emit_outlined_ms2sysv_restore): Likewise.
28028 (ix86_expand_prologue): Modify stack re-alignment code and call
28029 emit_outlined_ms2sysv_save when appropriate.
28030 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
28031 parameter rtx_insn *insn, which allows the function to be used to only
28032 generate the notes.
28033 (ix86_expand_epilogue): Modify validity checks of frame and stack
28034 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
28035 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
28036 * config/i386/predicates.md
28037 (save_multiple): New predicate.
28038 (restore_multiple): Likewise.
28039 * config/i386/sse.md
28040 (save_multiple<mode>): New pattern.
28041 (save_multiple_realign<mode>): Likewise.
28042 (restore_multiple<mode>): Likewise.
28043 (restore_multiple_and_return<mode>): Likewise.
28044 (restore_multiple_leave_return<mode>): Likewise.
28045 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
28047 2017-05-14 Julia Koval <julia.koval@intel.com>
28049 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
28050 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
28051 (__builtin_ia32_xsetbv): New builtins.
28052 * config/i386/i386.c (ix86_expand_special_args_builtin):
28054 (ix86_expand_builtin): Special expand for new intrinsics.
28055 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
28056 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
28057 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
28059 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28061 * cfganal.c (inverted_post_order_compute): Change argument type
28063 * cfganal.h (inverted_post_order_compute): Adjust prototype.
28064 * df-core.c (rest_of_handle_df_initialize): Adjust.
28065 (rest_of_handle_df_finish): Likewise.
28066 (df_analyze_1): Likewise.
28067 (df_analyze): Likewise.
28068 (loop_inverted_post_order_compute): Change argument to be a vec *.
28069 (df_analyze_loop): Adjust.
28070 (df_get_n_blocks): Likewise.
28071 (df_get_postorder): Likewise.
28072 * df.h (struct df_d): Change field to be a vec.
28073 * lcm.c (compute_laterin): Adjust.
28074 (compute_available): Likewise.
28075 * lra-lives.c (lra_create_live_ranges_1): Likewise.
28076 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
28077 * tree-ssa-pre.c (compute_antic): Likewise.
28079 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28081 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
28082 (depth_first_search::depth_first_search): Change structure init
28083 function to this constructor.
28084 (depth_first_search::add_bb): Rename function to this member.
28085 (depth_first_search::execute): Likewise.
28086 (flow_dfs_compute_reverse_finish): Adjust.
28088 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28090 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
28091 (longest_simple_path): Likewise.
28092 * shrink-wrap.c (spread_components): Likewise.
28093 (disqualify_problematic_components): Likewise.
28094 (emit_common_heads_for_components): Likewise.
28095 (emit_common_tails_for_components): Likewise.
28096 (insert_prologue_epilogue_for_components): Likewise.
28098 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28100 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
28103 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28105 * df-core.c (df_set_blocks): Start using auto_bitmap.
28106 (df_compact_blocks): Likewise.
28107 * df-problems.c (df_rd_confluence_n): Likewise.
28108 * df-scan.c (df_insn_rescan_all): Likewise.
28109 (df_process_deferred_rescans): Likewise.
28110 (df_update_entry_block_defs): Likewise.
28111 (df_update_exit_block_uses): Likewise.
28112 (df_entry_block_bitmap_verify): Likewise.
28113 (df_exit_block_bitmap_verify): Likewise.
28114 (df_scan_verify): Likewise.
28115 * lra-constraints.c (lra_constraints): Likewise.
28116 (undo_optional_reloads): Likewise.
28117 (lra_undo_inheritance): Likewise.
28118 * lra-remat.c (calculate_gen_cands): Likewise.
28119 (do_remat): Likewise.
28120 * lra-spills.c (assign_spill_hard_regs): Likewise.
28121 (spill_pseudos): Likewise.
28122 * tree-ssa-pre.c (bitmap_set_and): Likewise.
28123 (bitmap_set_subtract_values): Likewise.
28125 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28127 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
28128 management with auto_bitmap.
28129 (fix_inter_tick): Likewise.
28130 (fix_recovery_deps): Likewise.
28131 * ira.c (add_store_equivs): Likewise.
28132 (find_moveable_pseudos): Likewise.
28133 (split_live_ranges_for_shrink_wrap): Likewise.
28134 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
28135 (rtx_reuse_manager::seen_def_p): Likewise.
28136 (rtx_reuse_manager::set_seen_def): Likewise.
28137 * print-rtl.h (class rtx_reuse_manager): Likewise.
28139 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28141 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
28143 (migrate_btr_def): Likewise.
28144 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
28145 * df-core.c (loop_post_order_compute): Likewise.
28146 (loop_inverted_post_order_compute): Likewise.
28147 * hsa-common.h: Likewise.
28148 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
28149 * init-regs.c (initialize_uninitialized_regs): Likewise.
28150 * ipa-inline.c (resolve_noninline_speculation): Likewise.
28151 (inline_small_functions): Likewise.
28152 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
28153 * ira.c (combine_and_move_insns): Likewise.
28154 (build_insn_chain): Likewise.
28155 * loop-invariant.c (find_invariants): Likewise.
28156 * lower-subreg.c (propagate_pseudo_copies): Likewise.
28157 * predict.c (tree_predict_by_opcode): Likewise.
28158 (predict_paths_leading_to): Likewise.
28159 (predict_paths_leading_to_edge): Likewise.
28160 (estimate_loops_at_level): Likewise.
28161 (estimate_loops): Likewise.
28162 * shrink-wrap.c (try_shrink_wrapping): Likewise.
28163 (spread_components): Likewise.
28164 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
28165 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
28166 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
28167 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
28168 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
28169 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
28170 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
28171 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
28172 (mark_threaded_blocks): Likewise.
28173 (thread_through_all_blocks): Likewise.
28174 * tree-ssa.c (verify_ssa): Likewise.
28175 (execute_update_addresses_taken): Likewise.
28176 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
28178 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28180 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
28182 (post_order_compute): Likewise.
28183 (inverted_post_order_compute): Likewise.
28184 (pre_and_rev_post_order_compute_fn): Likewise.
28186 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28188 * genrecog.c (int_set::int_set): Explicitly construct our
28189 auto_vec base class.
28190 * vec.h (auto_vec::auto_vec): New constructor.
28192 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28194 * bitmap.h (class auto_bitmap): New constructor taking
28195 bitmap_obstack * argument.
28197 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28199 * bitmap.h (class auto_bitmap): Change type of m_bits to
28200 bitmap_head, and adjust ctor / dtor and member operators.
28202 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
28204 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
28205 when returned register mode doesn't match original mode.
28207 2017-05-12 Jeff Law <law@redhat.com>
28208 Jakub Jelinek <jakub@redhat.com>
28210 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
28211 we look for cc setter after the compare-elim changes.
28212 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
28213 within the vector to match what compare-elim now expects.
28214 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
28215 (xorsi3_flags, one_cmplsi2_flags): Likewise.
28217 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
28218 after the compare-elim changes.
28219 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
28220 the vector to match what compare-elim now expects.
28221 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
28222 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
28223 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
28224 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
28225 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
28227 * config/visium/visium.c (single_set_and_flags): Fix where
28228 we look for cc setter after the compare-elim changes.
28229 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
28230 with the vector to match what compare-elim now expects.
28231 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
28232 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
28233 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
28234 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
28235 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
28236 (neg<mode>2_insn_set_overflow): Likewise.
28238 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
28240 PR middle-end/79794
28241 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
28242 maybe_expand_insn call, set ops[0].target. If still set after call,
28243 set alt_rtl. Add extra arg to recursive calls.
28244 (extract_bit_field): Add alt_rtl argument. Pass to
28246 * expmed.h (extract_bit_field): Fix prototype.
28247 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
28248 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
28249 to extract_bit_field_calls.
28250 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
28251 Pass alt_rtl to extract_bit_field calls.
28252 * calls.c (store_unaligned_arguments_into_psuedos)
28253 load_register_parameters): Pass extra NULL to extract_bit_field calls.
28254 * optabs.c (maybe_legitimize_operand): Clear op->target when call
28256 * optabs.h (struct expand_operand): Add target bitfield.
28258 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28260 * compare-elim.c (try_eliminate_compare): Canonicalize
28261 operation with embedded compare to
28262 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
28263 (set (reg) (operation)].
28265 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
28267 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28270 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
28271 cost of adding a carry flag for ADC instruction.
28272 [case MINUS]: Ignore the cost of subtracting a carry flag
28273 for SBB instruction.
28275 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
28277 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
28279 * config/rs6000/bmiintrin.h: New file.
28280 * config/rs6000/bmi2intrin.h: New file.
28281 * config/rs6000/x86intrin.h: New file.
28283 2017-05-12 Jeff Law <law@redhat.com>
28285 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
28288 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
28290 PR middle-end/80707
28291 * tree-cfg.c: Remove cfg edges of unreachable case statements.
28293 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28295 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28296 early expansion of vector divide builtins.
28297 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
28298 builtins identified as having unsigned arguments.
28300 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28302 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
28303 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
28304 expansion of vector logical operations (and, andc, or, xor,
28307 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28309 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
28310 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
28312 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28314 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28315 early GIMPLE expansion of vector multiplies.
28317 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28319 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
28320 TARGET_HAVE_MOVT conditional.
28321 (movt splitter): Likewise.
28323 2017-05-12 Richard Biener <rguenther@suse.de>
28325 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
28326 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28327 Fold all stmts not inplace.
28329 2017-05-12 Richard Biener <rguenther@suse.de>
28331 PR tree-optimization/80713
28332 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
28333 inserted_exprs bit for not removed stmts.
28335 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
28337 PR middle-end/69921
28338 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
28339 parallelized" attribute for parallelized OpenACC kernels.
28340 * omp-offload.c (execute_oacc_device_lower): Use it.
28342 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
28343 Set "oacc kernels" attribute.
28344 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
28345 parameter. Adjust all users.
28346 (oacc_fn_attrib_kernels_p): Remove function.
28347 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
28348 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
28349 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
28350 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
28351 assert "oacc kernels" attribute is set.
28353 2017-05-11 Carl Love <cel@us.ibm.com>
28355 * config/rs6000/rs6000-c: Add support for built-in functions
28356 vector unsigned char vec_popcnt (vector signed char)
28357 vector unsigned char vec_popcnt (vector unsigned char)
28358 vector unsigned short vec_popcnt (vector signed short)
28359 vector unsigned short vec_popcnt (vector unsigned short)
28360 vector unsigned int vec_popcnt (vector signed int)
28361 vector unsigned int vec_popcnt (vector unsigned int)
28362 vector unsigned long long vec_popcnt (vector signed long long)
28363 vector unsigned long long vec_popcnt (vector unsigned long long)
28364 vector signed long long vec_slo (vector signed long long,
28365 vector signed char)
28366 vector signed long long vec_slo (vector signed long long,
28367 vector unsigned char)
28368 vector unsigned long long vec_slo (vector unsigned long long,
28369 vector signed char)
28370 vector unsigned long long vec_slo (vector unsigned long long,
28371 vector unsigned char)
28372 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
28373 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
28374 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
28375 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
28376 * doc/extend.texi: Update the built-in documentation file for the
28377 new built-in functions.
28379 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
28381 * attribs.h (sorted_attr_string): Move machine independent
28382 functions for target clone support from the i386 port to common
28383 code. Rename ix86_function_versions to common_function_versions.
28384 Rename make_name to make_unique_name.
28385 (common_function_versions): Likewise.
28386 (make_unique_name): Likewise.
28387 (make_dispatcher_decl): Likewise.
28388 (is_function_default_version): Likewise.
28389 * attribs.c (attr_strcmp): Likewise.
28390 (sorted_attr_string): Likewise.
28391 (common_function_versions): Likewise.
28392 (make_unique_name): Likewise.
28393 (make_dispatcher_decl): Likewise.
28394 (is_function_default_version): Likewise.
28395 * config/i386/i386.c (attr_strcmp): Likewise.
28396 (sorted_attr_string): Likewise.
28397 (ix86_function_versions): Likewise.
28398 (make_name): Likewise.
28399 (make_dispatcher_decl): Likewise.
28400 (is_function_default_version): Likewise.
28401 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
28403 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28406 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
28407 Account for direct move costs for vec_construct of integer
28410 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
28413 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
28414 (UNSPEC_STX_ATOMIC): Ditto.
28415 (loaddi_via_sse): New insn.
28416 (storedi_via_sse): Ditto.
28417 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
28418 Update corresponding peephole2 patterns.
28419 (atomic_storedi_fpu): Ditto.
28421 2017-05-11 Julia Koval <julia.koval@intel.com>
28423 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
28424 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
28426 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
28427 (__builtin_ia32_rsqrt14ss_mask): New builtins.
28428 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
28430 2017-05-11 Nathan Sidwell <nathan@acm.org>
28432 * graphite-poly.c: Include dumpfile.h.
28434 * dumpfle.h (dump_function): Declare here ...
28435 * tree-dump.h (dump_function): ... not here.
28436 * dumpfile.c: #include tree-cfg.h.
28437 (dump_function): Move here from ...
28438 * tree-dump.c (dump_function): ... here.
28439 * gimplify.c: #include splay-tree.h, not tree-dump.h.
28440 * graphite-poly.c: Don't include tree-dump.h.
28441 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
28442 * print-tree.c: Likewise.
28443 * stor-layout.c: Likewise.
28444 * tree-nested.c: Likewise.
28446 * dumpfile.c (dump_start): Use TDF_FLAGS.
28447 (dump_enable_all): Fix TDF_KIND check thinko.
28449 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
28451 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
28452 array entries to represent two legal parameterizations of the
28453 overloaded __builtin_cmpb function, as represented by the
28454 P6_OV_BUILTIN_CMPB constant.
28455 (altivec_resolve_overloaded_builtin): Add special case handling
28456 for the __builtin_cmpb function, as represented by the
28457 P6_OV_BUILTIN_CMPB constant.
28458 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
28459 (BU_P6_64BIT_2): New macro.
28460 (BU_P6_OVERLOAD_2): New macro
28461 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
28462 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
28463 (CMPB): Add overload support to represent both 32-bit and 64-bit
28464 compare-bytes function.
28465 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
28466 support for TARGET_CMPB.
28467 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
28468 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28469 documentation of the __builtin_cmpb overloaded built-in function.
28471 2017-05-11 Richard Biener <rguenther@suse.de>
28473 PR tree-optimization/80705
28474 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
28475 bases are not vectorizable.
28477 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28479 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
28480 when counting register pressure.
28482 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28484 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
28485 (iv_ca_more_deps): Renamed to ...
28486 (iv_ca_compare_deps): ... this.
28487 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
28489 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28491 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
28493 (determine_group_iv_costs): ... here.
28494 (find_inv_vars_cb): Record inv var if it's not recorded before.
28496 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28498 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
28499 (get_shiftadd_cost): Ditto.
28501 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28503 * tree-ssa-address.c: Include header file.
28504 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
28506 (add_to_parts): Refactor.
28507 (addr_to_parts): New parameter. Update use of move_hint_to_base.
28508 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
28511 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28513 PR tree-optimization/53090
28514 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
28516 (extract_cond_operands): Detect condition with IV on both sides
28517 and return COMP_IV_EXPR_2.
28518 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
28519 (rewrite_use_compare): Simplify by removing call to function
28520 extract_cond_operands.
28522 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28524 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
28525 (extract_cond_operands): Detect condition comparing against non-
28526 invariant bound and return appropriate enum value.
28527 (find_interesting_uses_cond): Update use of extract_cond_operands.
28528 Handle its return value accordingly.
28529 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
28531 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28533 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
28534 nonlinear iv_use computation in loop invariant sensitive way.
28536 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28538 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
28539 (find_iv_candidates): Call relate_compare_use_with_all_cands.
28541 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28543 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
28544 (dump_cand): Support iv_cand.inv_exprs.
28545 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
28547 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
28550 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28552 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
28554 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
28555 as local function. Include necessary header files.
28556 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
28558 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28560 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
28562 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28564 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
28565 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
28566 RSHIFT_EXPR and BIT_NOT_EXPR.
28568 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28570 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
28571 (adjust_setup_cost): New parameter supporting round up adjustment.
28572 (struct address_cost_data): Delete.
28573 (force_expr_to_var_cost): Don't bound cost with spill_cost.
28574 (split_address_cost, ptr_difference_cost): Delete.
28575 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
28576 (struct ainc_cost_data): New struct.
28577 (get_address_cost_ainc): New function.
28578 (get_address_cost, get_computation_cost): Reimplement.
28579 (determine_group_iv_cost_address): Record inv_expr for all uses of
28581 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
28582 (iv_ca_has_deps): Reimplemented to ...
28583 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
28585 (iv_ca_extend): Call iv_ca_more_deps.
28587 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28589 * tree-ssa-address.c (struct mem_address): Move to header file.
28590 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
28591 * tree-ssa-address.h (struct mem_address): Move from C file.
28592 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
28594 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28596 * tree-affine.h (aff_combination_type): New interface.
28597 (aff_combination_zero_p): Remove static.
28598 (aff_combination_const_p): New interface.
28599 (aff_combination_singleton_var_p): New interfaces.
28601 2017-05-11 Richard Biener <rguenther@suse.de>
28603 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28604 Skip unreachable blocks and destinations.
28605 (eliminate): Move stmt removal and fixup ...
28606 (fini_eliminate): ... here. Skip inserted exprs.
28607 (pass_pre::execute): Move fini_pre after fini_eliminate.
28608 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
28609 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
28610 PRE to get rid of dead code that has invalid SSA form and
28611 split critical edges again.
28613 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28615 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28617 2017-05-11 Richard Biener <rguenther@suse.de>
28619 * passes.c (execute_function_todo): Verify loops if they are
28620 said to be up-to-date.
28621 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
28622 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
28624 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
28627 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
28628 handle calling assemble_external ourself.
28631 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
28632 modes with zero size. Enhance comment.
28634 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28636 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
28637 built-ins for vec_xl and vec_xst with short and char pointer
28640 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
28642 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
28643 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
28644 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
28645 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
28646 (_mm_maskz_min_round_ss): New intrinsics.
28647 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
28648 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
28649 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
28650 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
28651 (__builtin_ia32_minss_mask_round): New builtins.
28652 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28653 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28654 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
28656 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
28657 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
28659 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
28662 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
28664 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
28665 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
28666 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
28667 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
28668 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
28669 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
28670 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
28671 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28672 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
28673 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
28674 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
28675 (__builtin_ia32_mulss_mask_round): New builtins.
28676 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28677 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28678 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
28680 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
28681 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
28683 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
28686 2017-05-10 Julia Koval <julia.koval@intel.com>
28688 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
28689 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
28690 (_mm256_setr_m128i): New intrinsics.
28692 2017-05-10 Julia Koval <julia.koval@intel.com>
28694 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
28695 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
28696 (_mm_maskz_rcp14_ss): New intrinsics.
28697 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
28698 (__builtin_ia32_rcp14ss_mask): New builtins.
28699 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
28701 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
28703 PR tree-optimization/51513
28704 * tree-cfg.c (gimple_seq_unreachable_p): New function.
28705 (assert_unreachable_fallthru_edge_p): Use it.
28706 (group_case_labels_stmt): Likewise.
28707 * tree-cfg.h: Prototype it.
28708 * stmt.c: Include cfghooks.h and tree-cfg.h.
28709 (emit_case_dispatch_table) <gap_label>: New local variable.
28710 Use it to fill dispatch table gaps.
28711 Test for default_label before updating probabilities.
28712 (expand_case) <default_label>: Remove unneeded initialization.
28713 Test for unreachable default case statement and remove its edge.
28714 Set default_label accordingly.
28715 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
28717 2017-05-10 Carl Love <cel@us.ibm.com>
28719 * config/rs6000/rs6000-c: Add support for built-in functions
28720 vector signed char vec_neg (vector signed char)
28721 vector signed short int vec_neg (vector short int)
28722 vector signed int vec_neg (vector signed int)
28723 vector signed long long vec_neg (vector signed long long)
28724 vector float vec_neg (vector float)
28725 vector double vec_neg (vector double)
28726 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
28728 * config/rs6000/altivec.h: Add define for vec_neg
28729 * doc/extend.texi: Update the built-in documentation for the
28730 new built-in functions.
28732 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28734 PR tree-optimization/77644
28735 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
28737 2017-05-10 Nathan Sidwell <nathan@acm.org>
28739 * dumpfile.h (TDI_lang_all): New.
28740 (TDF_KIND): New. Renumber others
28741 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
28743 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
28745 (get_dump_file_name): Adjust suffix generation.
28746 (dump_enable_all): Use TDF_KIND.
28747 * doc/invoke.texi (-fdump-lang-all): Document.
28749 * dumpfile.h: Tabify.
28751 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
28754 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
28755 Move member access before delete.
28757 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
28759 * tree-inline.c (expand_call_inline): Split block at stmt
28762 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
28765 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
28766 are now unused after splitting mov{sf,sd}_hardfloat.
28767 (f32_lr2): Likewise.
28768 (f32_lm): Likewise.
28769 (f32_lm2): Likewise.
28770 (f32_li): Likewise.
28771 (f32_li2): Likewise.
28772 (f32_lv): Likewise.
28773 (f32_sr): Likewise.
28774 (f32_sr2): Likewise.
28775 (f32_sm): Likewise.
28776 (f32_sm2): Likewise.
28777 (f32_si): Likewise.
28778 (f32_si2): Likewise.
28779 (f32_sv): Likewise.
28780 (f32_dm): Likewise.
28781 (f32_vsx): Likewise.
28782 (f32_av): Likewise.
28783 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
28784 For movsf, order stores so the VSX stores occur before the GPR
28785 store which encourages the register allocator to use a traditional
28786 FPR instead of a GPR. For movsd, order the stores so that the GPR
28787 store comes before the VSX stores to allow the power6 to work.
28788 This is due to the power6 not having a 32-bit integer store
28789 instruction from a FPR.
28790 (movsf_hardfloat): Likewise.
28791 (movsd_hardfloat): Likewise.
28793 2017-05-09 Martin Sebor <msebor@redhat.com>
28795 PR translation/80280
28796 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
28799 PR translation/80280
28800 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
28801 data member added in r247778.
28802 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
28804 2017-05-09 Nathan Sidwell <nathan@acm.org>
28806 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
28808 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
28811 2017-05-09 Marek Polacek <polacek@redhat.com>
28813 * doc/invoke.texi: Fix typo.
28815 2017-05-09 Richard Biener <rguenther@suse.de>
28817 * tree-vrp.c (vrp_val_is_max): Adjust comment.
28818 (vrp_val_is_min): Likewise.
28819 (set_value_range_to_value): Likewise.
28820 (set_value_range_to_nonnegative): Likewise.
28821 (gimple_assign_nonzero_p): Likewise.
28822 (gimple_stmt_nonzero_p): Likewise.
28823 (vrp_int_const_binop): Likewise. Remove unreachable case.
28824 (adjust_range_with_scev): Adjust comments.
28825 (compare_range_with_value): Likewise.
28826 (extract_range_from_phi_node): Likewise.
28827 (test_for_singularity): Likewise.
28829 2017-05-09 Richard Biener <rguenther@suse.de>
28831 * tree-vrp.c (get_single_symbol): Add assert that we don't
28832 get overflowed constants as invariant part.
28833 (compare_values_warnv): Add comment before the TREE_NO_WARNING
28834 checks. Use wi::cmp instead of recursing for integer constants.
28835 (compare_values): Just ignore whether we assumed undefined
28836 overflow instead of failing the compare.
28837 (extract_range_for_var_from_comparison_expr): Add comment before the
28838 TREE_NO_WARNING sets.
28839 (test_for_singularity): Likewise.
28840 (extract_range_from_comparison): Do not disable optimization
28841 when we assumed undefined overflow.
28842 (extract_range_basic): Remove init of unused var.
28844 2017-05-09 Richard Biener <rguenther@suse.de>
28846 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
28847 (extract_range_from_multiplicative_op_1): Adjust.
28848 (extract_range_from_binary_expr_1): Use int_const_binop.
28850 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
28853 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
28854 rs6000_store_data_bypass_p in seven define_bypass directives and
28855 in several comments.
28856 * config/rs6000/rs6000-protos.h: Add prototype for
28857 rs6000_store_data_bypass_p function.
28858 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
28859 function implements slightly different (rs6000-specific) semantics
28860 than store_data_bypass_p, returning false rather than aborting
28861 with assertion error when arguments do not satisfy the
28862 requirements of store data bypass.
28863 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
28864 rs6000_store_data_bypass_p.
28866 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
28868 * config/xtensa/xtensa-protos.h
28869 (xtensa_initial_elimination_offset): New declaration.
28870 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28871 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
28872 macro definition, add case for FRAME_POINTER_REGNUM when
28873 FRAME_GROWS_DOWNWARD.
28874 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
28875 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
28876 xtensa_initial_elimination_offset.
28878 2017-05-08 Nathan Sidwell <nathan@acm.org>
28880 * doc/invoke.texi: Alphabetize -fdump options.
28882 2017-05-08 Martin Sebor <msebor@redhat.com>
28884 PR translation/80280
28885 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
28887 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
28889 * target.def (compute_frame_layout): New optional target hook.
28890 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
28891 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
28892 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
28894 * reload1.c (verify_initial_elim_offsets): Likewise.
28895 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
28896 (use_simple_return_p): Call arm_compute_frame_layout if needed.
28897 (arm_get_frame_offsets): Split up into this ...
28898 (arm_compute_frame_layout): ... and this function.
28900 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
28902 * config/aarch64/constraints.md (Usa): New constraint.
28903 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
28905 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
28907 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
28908 with_multilib_list after it has been checked.
28910 2017-05-08 Richard Biener <rguenther@suse.de>
28912 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
28913 (bitmap_set_subtract_values): Likewise.
28915 2017-05-08 Richard Biener <rguenther@suse.de>
28917 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
28918 (gimple_assign_nonzero): ... this and remove strict_overflow_p
28920 (gimple_stmt_nonzero_warnv_p): Rename to ...
28921 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
28923 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
28924 (extract_range_basic): Adjust, do not disable propagation on
28925 strict overflow sensitive simplification.
28926 (vrp_visit_cond_stmt): Likewise.
28928 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
28930 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
28931 body size unconditionally.
28933 2017-05-07 Jeff Law <law@redhat.com>
28936 2017-05-06 Jeff Law <law@redhat.com>
28937 PR tree-optimization/78496
28938 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28941 PR tree-optimization/78496
28942 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28943 (simplify_stmt_using_ranges): Call it.
28944 (vrp_dom_walker::before_dom_children): Extract equivalences
28945 from an ASSERT_EXPR with an equality comparison against a
28948 2017-05-06 Jeff Law <law@redhat.com>
28950 PR tree-optimization/78496
28951 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28954 PR tree-optimization/78496
28955 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28956 (simplify_stmt_using_ranges): Call it.
28957 (vrp_dom_walker::before_dom_children): Extract equivalences
28958 from an ASSERT_EXPR with an equality comparison against a
28961 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
28963 * lra-constraints.c (lra_copy_reg_equiv): New function.
28964 (split_reg): Use it to copy equivalence information from the
28965 original register to the spill register.
28967 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
28969 PR rtl-optimization/75964
28970 * simplify-rtx.c (simplify_const_relational_operation): Remove
28971 invalid handling of comparisons of integer ABS.
28973 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
28975 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
28976 initialize to zero.
28977 (init_regs): Remove declaration.
28978 (function_arg_advance_32): Initialize error_p as boolean variable.
28980 2017-05-05 Nathan Sidwell <nathan@acm.org>
28982 * store-motion.c (remove_reachable_equiv_notes): Reformat long
28983 lines. Use for (;;).
28985 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28987 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
28988 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
28989 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
28990 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
28991 VF=2 that require versioning.
28993 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28995 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
28998 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29000 * diagnostic.h (diagnostic_override_option_index): Convert from
29001 macro to inline function.
29003 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29005 * diagnostic.c (last_module_changed_p): New function.
29006 (set_last_module): New function.
29007 (diagnostic_report_current_module): Convert macro usage to
29008 the above functions.
29009 * diagnostic.h (diagnostic_context::last_module): Strengthen
29010 from const line_map * to const line_map_ordinary *.
29011 (diagnostic_last_module_changed): Delete macro.
29012 (diagnostic_set_last_module): Delete macro.
29014 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29016 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
29017 with diagnostic_report_diagnostic.
29018 (diagnostic_n_impl_richloc): Likewise.
29019 * diagnostic.h (report_diagnostic): Delete macro.
29020 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
29021 with diagnostic_report_diagnostic.
29022 * substring-locations.c (format_warning_va): Likewise.
29024 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29026 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
29027 save/restor of format_spec. Move option-printing code to...
29028 (print_option_information): ...this new function, and
29029 reimplement by simply printing to the pretty_printer,
29030 rather than appending to the format string.
29032 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29034 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
29035 handling logic into...
29036 (update_effective_level_from_pragmas): ...this new function.
29038 2017-05-04 Andrew Waterman <andrew@sifive.com>
29040 * config/riscv/riscv.opt (mstrict-align): New option.
29041 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
29042 (SLOW_UNALIGNED_ACCESS): Define.
29043 (riscv_slow_unaligned_access): Declare.
29044 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
29046 (riscv_slow_unaligned_access): New variable.
29047 (rocket_tune_info): Set slow_unaligned_access to true.
29048 (optimize_size_tune_info): Set slow_unaligned_access to false.
29049 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
29050 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
29051 (riscv_option_override): Set riscv_slow_unaligned_access.
29052 * doc/invoke.texi: Add -mstrict-align to RISC-V.
29054 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
29056 * config/riscv/riscv.md: Unify indentation.
29058 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29063 * config/rs6000/rs6000.md (u code attribute): Add FIX and
29065 (extendsi<mode>2): Add support for doing sign extension via
29066 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
29067 don't have ISA 3.0 instructions.
29068 (extendsi<mode>2 splitter): Likewise.
29069 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
29070 generate the normal insns since SImode can now go in vector
29071 registers. Disallow the special UNSPECs needed for previous
29072 machines to hide SImode being used. Add new insns
29073 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
29074 (fix_trunc<mode>si2_stfiwx): Likewise.
29075 (fix_trunc<mode>si2_internal): Likewise.
29076 (fixuns_trunc<mode>si2): Likewise.
29077 (fixuns_trunc<mode>si2_stfiwx): Likewise.
29078 (fctiw<u>z_<mode>_smallint): Likewise.
29079 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
29080 of floating point to 32-bit integer from doing a direct move to
29081 the GPR registers to do a store.
29082 (fctiwz_<mode>): Break long line.
29084 2017-05-05 Bin Cheng <bin.cheng@arm.com>
29086 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
29087 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
29088 (addr_list, addr_offset_valid_p): New.
29089 (split_address_groups): Check offset validity with above function.
29090 (gt-tree-ssa-loop-ivopts.h): Include header file.
29092 2017-05-05 Nathan Sidwell <nathan@acm.org>
29094 * config.gcc (arm*-*-*): Add missing 'fi'.
29096 2017-05-05 Steve Ellcey <sellcey@cavium.com>
29098 * doc/invoke.texi (-fopt-info): Explicitly say order of options
29099 included in -fopt-info does not matter.
29100 * doc/optinfo.texi (-fopt-info): Fix description of default
29101 behavour. Explicitly say order of options included in -fopt-info
29104 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29106 * config.gcc: Allow combinations of aprofile and rmprofile values for
29107 --with-multilib-list.
29108 * config/arm/t-multilib: New file.
29109 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
29110 variables. Remove setting of ISA and floating-point ABI in
29111 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
29112 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
29113 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
29114 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
29116 * config/arm/t-rmprofile: Likewise except for the matches changes.
29117 * doc/install.texi (--with-multilib-list): Document the combination of
29118 aprofile and rmprofile values and warn about pitfalls in doing that.
29120 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29122 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
29123 (movdi_aarch64): Likewise.
29125 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29127 PR tree-optimization/80632
29128 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
29130 (build_arrays): Initialize it for virtual phis.
29131 (fix_phi_nodes): Use it for virtual phis.
29133 PR tree-optimization/80558
29134 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
29135 [x, y] op z into [x op, y op z] for op & or | if conditions
29138 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
29139 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29142 * config/arm/arm.md (use_literal_pool): Remove.
29143 (64-bit immediate split): No longer takes cost into consideration
29144 if arm_disable_literal_pool is enabled.
29145 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
29146 used when arm_disable_literal_pool is enabled.
29147 (arm_max_const_double_inline_cost): Remove use of
29148 arm_disable_literal_pool.
29149 (push_minipool_fix): Add assert.
29150 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
29151 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
29152 (no_literal_pool_sf_immediate): New.
29154 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29156 PR tree-optimization/80613
29157 * tree-ssa-dce.c (propagate_necessity): Remove cases for
29158 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
29160 2017-05-05 Richard Biener <rguenther@suse.de>
29162 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
29164 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
29166 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
29167 of this flag from insn conditions due to removal from r247495.
29169 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29171 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
29173 (arm_early_store_addr_dep_ptr): Likewise.
29174 * config/arm/aarch-common-protos.h
29175 (arm_early_load_addr_dep_ptr): Add prototype.
29176 (arm_early_store_addr_dep_ptr): Likewise.
29177 * config/arm/cortex-a53.md: Add new bypasses.
29179 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29181 * tree.c (next_type_uid): Change type to unsigned.
29182 (type_hash_canon): Decrement back next_type_uid if
29183 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
29184 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
29187 2017-05-04 Martin Sebor <msebor@redhat.com>
29189 * builtins.c: Fix a trivial typo in a comment.
29191 PR middle-end/79234
29192 * builtins.c (check_sizes): Adjust to handle reading past the end.
29193 Avoid printing excessive upper bound of ranges. Use %E to print
29194 tree nodes instead of converting them to %wu.
29195 (expand_builtin_memchr): New function.
29196 (compute_dest_size): Rename...
29197 (compute_objsize): ...to this.
29198 (expand_builtin_memcpy): Adjust.
29199 (expand_builtin_mempcpy): Adjust.
29200 (expand_builtin_strcat): Adjust.
29201 (expand_builtin_strcpy): Adjust.
29202 (check_strncat_sizes): Adjust.
29203 (expand_builtin_strncat): Adjust.
29204 (expand_builtin_strncpy): Adjust and simplify.
29205 (expand_builtin_memset): Adjust.
29206 (expand_builtin_bzero): Adjust.
29207 (expand_builtin_memcmp): Adjust.
29208 (expand_builtin): Handle memcmp.
29209 (maybe_emit_chk_warning): Check strncat just once.
29211 2017-05-04 Martin Sebor <msebor@redhat.com>
29213 PR preprocessor/79214
29214 PR middle-end/79222
29215 PR middle-end/79223
29216 * builtins.c (check_sizes): Add inlining context and issue
29217 warnings even when -Wno-system-headers is set.
29218 (check_strncat_sizes): Same.
29219 (expand_builtin_strncat): Same.
29220 (expand_builtin_memmove): New function.
29221 (expand_builtin_stpncpy): Same.
29222 (expand_builtin): Handle memmove and stpncpy.
29224 2017-05-04 Bin Cheng <bin.cheng@arm.com>
29226 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
29227 which is not used any more.
29229 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29231 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
29233 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29235 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
29236 (cortexa53_tunings): Likewise.
29237 (cortexa57_tunings): Likewise.
29238 (cortexa72_tunings): Likewise.
29239 (cortexa73_tunings): Likewise.
29241 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29243 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
29244 Set loop alignment to 8.
29246 2017-05-04 Martin Sebor <msebor@redhat.com>
29248 PR translation/80280
29249 * builtins.c (expand_builtin_object_size): Add missing quoting to
29250 %D and like directives.
29251 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
29252 (hsa_type_for_tree_type): Same.
29253 (verify_function_arguments): Same.
29254 * symtab.c (symbol_table::change_decl_assembler_name): Same.
29255 * varasm.c (get_section): Same.
29258 2017-05-04 Martin Sebor <msebor@redhat.com>
29260 PR translation/80280
29261 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
29263 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29265 * config/aarch64/aarch64.c (generic_addrcost_table):
29266 Change HI/TI mode setting.
29268 2017-05-04 Martin Jambor <mjambor@suse.cz>
29270 PR tree-optimization/80622
29271 * tree-sra.c (comes_initialized_p): New function.
29272 (build_accesses_from_assign): Only set write lazily when
29273 comes_initialized_p is false.
29274 (analyze_access_subtree): Use comes_initialized_p.
29275 (propagate_subaccesses_across_link): Assert !comes_initialized_p
29276 instead of testing for PARM_DECL.
29278 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29280 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
29281 constraint on operand 0 to allow more general addressing modes.
29282 Adjust output template.
29283 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
29285 * config/aarch64/aarch64-protos.h
29286 (aarch64_address_valid_for_prefetch_p): Declare prototype.
29287 * config/aarch64/constraints.md (Dp): New address constraint.
29288 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
29291 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
29293 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
29294 update use of estimate_ipcp_clone_size_and_time.
29295 (estimate_local_effects): Update use of
29296 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
29297 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
29298 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
29299 Return nonspecialized time.
29301 2017-05-04 Richard Biener <rguenther@suse.de>
29303 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
29304 for the last VUSE which def dominates the PHI. Directly call
29306 (get_continuation_for_phi_1): Remove.
29308 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
29310 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
29311 to explain the use of truncating division. Cap the number of
29312 iterations to the maximum given by nb_iterations_upper_bound,
29315 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
29317 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
29318 * configure: Regenerate.
29319 * config.in: Regenerate.
29320 * config/i386/driver-mingw32.c: new file.
29321 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
29322 * config.host: Link driver-mingw32.o on MinGW host.
29323 * doc/install.texi: Document new --enable-mingw-wildcard configure
29326 2017-05-04 Marek Polacek <polacek@redhat.com>
29328 PR tree-optimization/80612
29329 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
29331 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29332 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
29334 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
29335 (movt splitter): Likewise.
29336 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
29337 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
29338 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
29339 block for Thumb-1 with MOVT.
29340 (thumb2_legitimate_address_p): Move code block ...
29341 (can_avoid_literal_pool_for_label_p): ... into this new function.
29342 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
29344 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
29345 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
29346 "M-profile targets with the MOVT instruction".
29348 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29350 * config/arm/arm-builtins.c (arm_init_builtins): Rename
29351 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
29352 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
29354 2017-05-04 Martin Liska <mliska@suse.cz>
29356 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
29357 variable cond_code.
29359 2017-05-04 Richard Biener <rguenther@suse.de>
29361 * tree.c (array_at_struct_end_p): Handle arrays at struct
29362 end with flexarrays more conservatively. Refactor and treat
29363 arrays of arrays or aggregates more strict. Fix
29364 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
29365 * tree.c (array_at_struct_end_p): Adjust prototype.
29366 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
29367 * gimple-fold.c (get_range_strlen): Likewise.
29368 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
29370 2017-05-04 Richard Biener <rguenther@suse.de>
29372 PR tree-optimization/31130
29373 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
29375 (supports_overflow_infinity): Likewise.
29376 (is_negative_overflow_infinity): Likewise.
29377 (is_positive_overflow_infinity): Likewise.
29378 (is_overflow_infinity): Likewise.
29379 (stmt_overflow_infinity): Likewise.
29380 (overflow_infinity_range_p): Likewise.
29381 (usable_range_p): Remove as always returning true.
29382 (make_overflow_infinity): Remove.
29383 (negative_overflow_infinity): Likewise.
29384 (positive_overflow_infinity): Likewise.
29385 (avoid_overflow_infinity): Likewise.
29386 (set_value_range): Adjust accordingly.
29387 (set_value_range_to_nonnegative): Likewise, remove now unused
29388 overflow_infinity arg.
29389 (vrp_operand_equal_p): Adjust.
29390 (update_value_range): Likewise.
29391 (range_int_cst_singleton_p): Likewise.
29392 (operand_less_p): Likewise.
29393 (compare_values_warnv): Likewise.
29394 (extract_range_for_var_from_comparison_expr): Likewise.
29395 (vrp_int_const_binop): Likewise.
29396 (zero_nonzero_bits_from_vr): Likewise.
29397 (extract_range_from_multiplicative_op_1): Likewise.
29398 (extract_range_from_binary_expr_1): Likewise.
29399 (extract_range_from_unary_expr): Likewise.
29400 (extract_range_from_comparison): Likewise.
29401 (extract_range_basic): Likewise.
29402 (adjust_range_with_scev): Likewise.
29403 (compare_ranges): Likewise.
29404 (compare_range_with_value): Likewise.
29405 (dump_value_range): Likewise.
29406 (test_for_singularity): Likewise, remove strict_overflow_p parameter
29408 (simplify_cond_using_ranges): Adjust.
29410 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
29412 * brig-builtins.def: Added a builtin for class_f64.
29413 * builtin-types.def: Added a builtin type needed by class_f64.
29415 2017-05-03 Jason Merrill <jason@redhat.com>
29417 * timevar.def: Add TV_CONSTEXPR.
29419 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29421 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
29423 2017-05-03 Martin Jambor <mjambor@suse.cz>
29425 * ipa-prop.c (ipa_update_after_lto_read): Removed.
29426 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
29427 * ipa-cp.c (ipcp_propagate_stage): Do not call
29428 ipa_update_after_lto_read.
29429 * ipa-inline.c (ipa_inline): Likewise.
29431 2017-05-03 Martin Jambor <mjambor@suse.cz>
29433 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
29434 tag. Added a default constructor and a destructor.
29435 (ipa_edge_args_sum_t): New class;
29436 (ipa_edge_args_sum): Declare.
29437 (ipa_edge_args_vector): Remove declaration.
29438 (IPA_EDGE_REF): Use ipa_edge_args_sum.
29439 (ipa_free_edge_args_substructures): Remove declaration.
29440 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
29441 (ipa_edge_args_info_available_for_edge_p): Likewise.
29442 * ipa-prop.c (ipa_edge_args_vector): Removed.
29443 (edge_removal_hook_holder): Likewise.
29444 (edge_duplication_hook_holder): Likewise.
29445 (ipa_edge_args_sum): New variable.
29446 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
29447 ipa_edge_args_vector.
29448 (ipa_free_edge_args_substructures): Likewise.
29449 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
29450 ipa_edge_args_vector.
29451 (ipa_edge_removal_hook): Turned into method
29452 ipa_edge_args_sum_t::remove.
29453 (ipa_edge_duplication_hook): Turned into method
29454 ipa_edge_args_sum_t::duplicate.
29455 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
29456 registering edge hooks.
29457 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
29458 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
29459 ipa_edge_args_sum instead of ipa_edge_args_vector.
29460 * ipa-profile.c (ipa_profile): Likewise.
29462 2017-05-03 Martin Jambor <mjambor@suse.cz>
29464 * symbol-summary.h (function_summary): New method exists.
29465 (function_summary::symtab_removal): Deallocate through release.
29466 (call_summary): New class.
29467 (gt_ggc_mx): New overload.
29468 (gt_pch_nx): Likewise.
29469 (gt_pch_nx): Likewise.
29471 2017-05-03 Jeff Law <law@redhat.com>
29473 PR tree-optimization/78496
29474 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
29475 from simplify_cond_using_ranges. Split off code to walk
29476 backwards through casts into ...
29477 (simplify_cond_using_ranges_2): New function.
29478 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
29479 (execute_vrp): After identifying jump threads, call
29480 simplify_cond_using_ranges_2.
29482 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
29485 * ipa-inline.h (inline_summary): Add ctor.
29486 (create_ggc): Do not use ggc_cleared_alloc.
29488 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
29489 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29491 * gcc.c (handle_braces): Support escaping in switch matching
29493 * doc/invoke.texi (Spec Files): Document it.
29494 Remove superfluous @code markup in items.
29496 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29498 * diagnostic-show-locus.c (struct column_range): New struct.
29499 (get_affected_columns): New function.
29500 (get_printed_columns): New function.
29501 (struct correction): New struct.
29502 (correction::ensure_capacity): New function.
29503 (correction::ensure_terminated): New function.
29504 (struct line_corrections): New struct.
29505 (line_corrections::~line_corrections): New dtor.
29506 (line_corrections::add_hint): New function.
29507 (layout::print_trailing_fixits): Reimplement in terms of the new
29509 (selftest::test_overlapped_fixit_printing): New function.
29510 (selftest::diagnostic_show_locus_c_tests): Call it.
29512 2017-05-03 Nathan Sidwell <nathan@acm.org>
29514 Canonicalize canonical type hashing
29515 * tree.h (type_hash_canon_hash): Declare.
29516 * tree.c (type_hash_list, attribute_hash_list): Move into
29517 type_hash_canon_hash.
29518 (build_type_attribute_qual_variant): Break out hash code calc into
29519 type_hash_canon_hash.
29520 (type_hash_canon_hash): New. Generic type hash computation.
29521 (build_range_type_1, build_array_type_1, build_function_type,
29522 build_method_type_directly, build_offset_type, build_complex_type,
29523 make_vector_type): Call it.
29525 2017-05-03 Richard Biener <rguenther@suse.de>
29527 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
29528 When all DRs have unknown misaligned do not always peel
29529 when there is a store but apply the same costing model as if
29530 there were only loads.
29532 2017-05-03 Richard Biener <rguenther@suse.de>
29535 PR tree-optimization/80492
29536 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
29537 compare_base_decls returning dont-know properly.
29539 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
29541 * config/arm/iterators.md (CCSI): New mode iterator.
29542 (arch): New mode attribute.
29543 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
29544 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
29545 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
29546 code iterator for success result mode.
29547 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
29548 the corresponding new insn generators.
29550 2017-05-03 Bin Cheng <bin.cheng@arm.com>
29553 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29554 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29556 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
29558 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
29559 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
29560 (DDR_A): Wrap DDR argument in brackets.
29561 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
29562 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
29563 (DDR_REVERSED_P): Likewise.
29565 2017-05-03 Jakub Jelinek <jakub@redhat.com>
29567 PR tree-optimization/79472
29568 * tree-switch-conversion.c (struct switch_conv_info): Add
29569 contiguous_range and default_case_nonstandard fields.
29570 (collect_switch_conv_info): Compute contiguous_range and
29571 default_case_nonstandard fields, don't clear final_bb if
29572 contiguous_range and only the default case doesn't have the required
29574 (check_all_empty_except_final): Set default_case_nonstandard instead
29575 of failing if contiguous_range and the default case doesn't have empty
29577 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
29578 and only the default case doesn't have the required constants. Skip
29580 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
29581 if default_case_nonstandard.
29582 (build_constructors): Build constant 1 just once. Assert that default
29583 values aren't inserted in between cases if contiguous_range. Skip
29585 (build_arrays): Skip virtual phis.
29586 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
29587 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
29588 Handle virtual phis.
29589 (gen_inbound_check): Handle default_case_nonstandard case.
29590 (process_switch): Adjust check_final_bb caller. Call
29591 gather_default_values with the first non-default case instead of
29592 default case if default_case_nonstandard.
29594 2017-05-02 Nathan Sidwell <nathan@acm.org>
29596 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
29597 check. Fix formatting.
29599 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
29601 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
29602 errors when comparing specialized and unspecialized times.
29604 2017-05-02 David Malcolm <dmalcolm@redhat.com>
29606 * diagnostic-show-locus.c
29607 (layout::should_print_annotation_line_p): Make private.
29608 (layout::print_annotation_line): Make private.
29609 (layout::annotation_line_showed_range_p): Make private.
29610 (layout::show_ruler): Make private.
29611 (layout::print_source_line): Make private. Pass in line and
29612 line_width, rather than calling location_get_source_line. Drop
29614 (layout::print_leading_fixits): New method.
29615 (layout::print_any_fixits): Rename to...
29616 (layout::print_trailing_fixits): ...this, and make private.
29617 Don't print newline fixits.
29618 (diagnostic_show_locus): Move logic for printing one row into...
29619 (layout::print_line): ...this new function. Move the
29620 location_get_source_line call and error-handling from
29621 print_source_line to here. Call print_leading_fixits, and rename
29622 print_any_fixits to print_trailing_fixits.
29623 (selftest::test_fixit_insert_containing_newline): Update now that
29624 newlines are partially supported.
29625 (selftest::test_fixit_insert_containing_newline_2): New test.
29626 (selftest::test_fixit_replace_containing_newline): Update comments.
29627 (selftest::diagnostic_show_locus_c_tests): Call the new test.
29628 * edit-context.c (class added_line): New class.
29629 (class edited_line): Describe newline handling in comment.
29630 (edited_line::actually_edited_p): New method.
29631 (edited_line::print_content): Delete redundant decl.
29632 (edited_line::m_predecessors): New field.
29633 (edited_file::print_content): Call edited_line::print_content.
29634 (edited_file::print_diff): Update to support newlines.
29635 (edited_file::print_diff_hunk): Likewise.
29636 (edited_file::print_run_of_changed_lines): New function.
29637 (edited_file::print_diff_line): Convert to...
29638 (print_diff_line): ...this.
29639 (edited_file::get_effective_line_count): New function.
29640 (edited_line::edited_line): Initialize new field m_predecessors.
29641 (edited_line::~edited_line): Clean up m_predecessors.
29642 (edited_line::apply_fixit): Handle newlines.
29643 (edited_line::get_effective_line_count): New function.
29644 (edited_line::print_content): New function.
29645 (edited_line::print_diff_lines): New function.
29646 (selftest::test_applying_fixits_insert_containing_newline): New
29648 (selftest::test_applying_fixits_replace_containing_newline): New
29650 (selftest::insert_line): New function.
29651 (selftest::test_applying_fixits_multiple_lines): Add example of
29653 (selftest::edit_context_c_tests): Call the new tests.
29655 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29657 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
29658 parameter cand. Update dump information.
29659 (get_computation_cost): Update uses.
29661 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29663 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
29664 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
29665 (get_computation_at, rewrite_use_address): Update use of
29666 get_computation_aff.
29668 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29670 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
29671 (get_computation): Delete.
29672 (get_computation_cost): Implement like get_computation_cost_at.
29673 Use get_computation_at.
29674 (get_computation_cost_at): Delete.
29675 (rewrite_use_nonlinear_expr): Use get_computation_at.
29676 (rewrite_use_compare, remove_unused_ivs): Ditto.
29678 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29680 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
29682 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29684 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
29685 (ivopts_global_cost_for_size): Rename parameter and update uses.
29686 (iv_ca_recount_cost): Update uses.
29687 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
29688 candidates seperately in n_invs and n_cands.
29689 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
29691 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29693 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
29694 (find_inv_vars_cb): New.
29695 (find_depends): Renamed to ...
29696 (find_inv_vars): ... this.
29697 (add_candidate_1, force_var_cost): Call find_inv_vars.
29698 (split_address_cost, determine_group_iv_cost_cond): Ditto.
29700 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29702 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
29703 inv_vars. Add inv_exprs.
29704 (struct iv_cand): Rename depends_on to inv_vars.
29705 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
29706 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
29707 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
29708 (dump_cand): Dump inv_vars.
29709 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
29710 (record_invariant, find_depends, add_candidate_1): Ditto.
29711 (set_group_iv_cost, force_var_cost): Ditto.
29712 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
29713 (get_computation_cost_at, get_computation_cost): Ditto.
29714 (determine_group_iv_cost_generic): Ditto.
29715 (determine_group_iv_cost_address): Ditto.
29716 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
29717 (determine_group_iv_costs): Ditto.
29718 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
29719 (iv_ca_set_remove_invariants): Renamed to ...
29720 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
29721 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
29722 (iv_ca_set_add_invariants): Renamed to ...
29723 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
29724 (iv_ca_set_cp): Use iv_ca_set_add_invs.
29725 (iv_ca_has_deps): Support inv_vars and inv_exprs.
29726 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
29727 (create_new_ivs): Remove useless dump.
29729 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29731 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
29733 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
29734 (iv_ca_set_no_cp, create_new_iv): Ditto.
29736 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29738 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29740 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29742 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
29743 function tree_check2.
29745 2017-05-02 Martin Liska <mliska@suse.cz>
29747 * doc/gcov.texi: Add missing preposition.
29748 * gcov.c (function_info::function_info): Properly fill up
29749 all member variables.
29751 2017-05-02 Tamar Christina <tamar.christina@arm.com>
29753 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
29755 2017-05-02 Tamar Christina <tamar.christina@arm.com>
29757 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
29759 2017-05-02 Martin Liska <mliska@suse.cz>
29762 * lto-streamer-in.c (lto_read_tree_1): Remove
29763 LTO_STREAMER_DEBUG.
29764 * lto-streamer.c (struct tree_hash_entry): Likewise.
29765 (struct tree_entry_hasher): Likewise.
29766 (tree_entry_hasher::hash): Likewise.
29767 (tree_entry_hasher::equal): Likewise.
29768 (lto_streamer_init): Likewise.
29769 (lto_orig_address_map): Likewise.
29770 (lto_orig_address_get): Likewise.
29771 (lto_orig_address_remove): Likewise.
29772 * lto-streamer.h: Likewise.
29773 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
29774 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
29776 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
29778 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
29779 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
29780 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
29781 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
29782 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
29783 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
29784 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
29785 (mm_maskz_sub_ss): New intrinsics.
29786 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29787 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
29788 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
29789 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
29790 (__builtin_ia32_subss_mask_round): New builtins.
29791 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29792 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29793 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
29795 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
29796 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
29798 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
29801 2017-05-02 Martin Jambor <mjambor@suse.cz>
29803 PR tree-optimization/78687
29804 * tree-sra.c (access): New field parent.
29805 (process_subtree_disqualification): New function.
29806 (disqualify_candidate): Call it.
29807 (build_accesses_from_assign): Reset write flag if creating an
29809 (build_access_subtree): Fill in parent field and also prpagate
29810 down grp_write flag.
29811 (create_artificial_child_access): New parameter set_grp_write, set
29812 grp_write to its value.
29813 (propagate_subaccesses_across_link): Also propagate grp_write flag
29815 (propagate_all_subaccesses): Push the closest parent back to work
29816 queue if add_access_to_work_queue returned true.
29818 2017-05-02 Richard Biener <rguenther@suse.de>
29820 * common.opt (fstrict-overflow): Alias negative to fwrapv.
29821 * doc/invoke.texi (fstrict-overflow): Remove all traces of
29822 -fstrict-overflow documentation.
29823 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
29824 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
29825 flag_strict_overflow.
29826 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
29827 * lto-opts.c (lto_write_options): Do not stream it.
29828 * lto-wrapper.c (merge_and_complain): Do not handle it.
29829 * opts.c (default_options_table): Do not set -fstrict-overflow.
29830 (finish_options): Likewise do not clear it when sanitizing.
29831 * simplify-rtx.c (simplify_const_relational_operation): Do not
29832 test flag_strict_overflow.
29834 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29836 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
29837 using enabled attribute.
29838 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
29839 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
29840 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
29841 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
29842 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
29843 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
29844 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
29845 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
29846 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
29847 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
29849 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29851 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
29853 2017-05-02 Richard Biener <rguenther@suse.de>
29855 PR tree-optimization/80591
29857 2017-04-10 Richard Biener <rguenther@suse.de>
29859 * tree-ssa-structalias.c (find_func_aliases): Properly handle
29862 2017-05-02 Richard Biener <rguenther@suse.de>
29864 PR tree-optimization/80549
29865 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
29866 (cleanup_tree_cfg_noloop): Create forwarders to known loop
29867 headers if they do not have a preheader.
29869 2017-05-02 Martin Liska <mliska@suse.cz>
29872 * common.opt: Fix typo.
29873 * doc/invoke.texi: Likewise.
29875 2017-05-01 Jan Beulich <jbeulich@suse.com>
29877 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
29878 swapping, add (x,x,m,x,n) alternative.
29880 2017-05-01 Nathan Sidwell <nathan@acm.org>
29882 * calls.c (combine_pending_stack_adjustment_and_call): Remove
29883 unnecessary unadjusted_alignment check.
29885 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29888 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
29890 * gimplify.c (gimplify_cilk_detach): New function.
29891 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
29892 * tree-core.h: Document EXPR_CILK_SPAWN.
29893 * tree.h (EXPR_CILK_SPAWN): Define.
29895 2017-05-01 David Malcolm <dmalcolm@redhat.com>
29897 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
29898 to use new fixit_hint representation, using the "replace" logic.
29899 (get_line_span_for_fixit_hint): Likewise.
29900 (layout::print_any_fixits): Likewise.
29901 (selftest::test_one_liner_many_fixits): Rename to...
29902 (selftest::test_one_liner_many_fixits_1): ...this, and update
29903 comment and expected output to reflect that the multiple fix-it
29904 hints are now consolidated into one insertion.
29905 (selftest::test_one_liner_many_fixits_2): New test.
29906 (selftest::test_diagnostic_show_locus_one_liner): Update for
29908 (selftest::test_fixit_consolidation): Update for fix-it API
29910 * diagnostic.c (print_parseable_fixits): Likewise.
29911 * edit-context.c (edited_line::m_line_events): Convert from
29912 auto_vec <line_event *> to auto_vec <line_event>.
29913 (class line_event): Convert from abstract base class to a concrete
29914 class, taking over the role of replace_event.
29915 (class insert_event): Delete.
29916 (class replace_event): Rename to class line_event. Convert to
29918 (edit_context::add_fixits): Reimplement.
29919 (edit_context::apply_insert): Delete.
29920 (edit_context::apply_replace): Rename to...
29921 (edit_context::apply_fixit): ...this. Convert to half-open range.
29922 (edited_file::apply_insert): Delete.
29923 (edited_file::apply_replace): Rename to...
29924 (edited_file::apply_fixit): ...this.
29925 (edited_line::~edited_line): Drop deletion of events.
29926 (edited_line::apply_insert): Delete.
29927 (edited_line::apply_replace): Rename to...
29928 (edited_line::apply_fixit): ...this. Convert to half-open range.
29929 Update for change to type of m_line_events.
29930 * edit-context.h (edit_context::apply_insert): Delete.
29931 (edit_context::apply_replace): Rename to...
29932 (edit_context::apply_fixit): ...this.
29934 2017-05-01 Martin Sebor <msebor@redhat.com>
29936 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
29939 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
29942 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
29943 __get_cpuid_max returns 0.
29944 (__get_cpuid_count): Ditto.
29946 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
29948 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
29949 replacement expression is another instance of one of its arguments.
29951 2017-05-01 Jakub Jelinek <jakub@redhat.com>
29954 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
29955 check for stack push/pop autoinc.
29956 * config/i386/i386.c (ix86_agi_dependent): Return false
29957 if the only reason why modified_in_p returned true is that
29958 addr is SP based and set_insn is a push or pop.
29960 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
29962 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
29965 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
29968 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
29969 (account_size_time): Use two predicates - exec_pred and
29971 (evaluate_conditions_for_known_args): Compute both clause and
29973 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
29974 (inline_summary_t::duplicate): Update.
29975 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
29977 (compute_inline_parameters): Likewise.
29978 (estimate_edge_size_and_time): Update caluclation of time.
29979 (estimate_node_size_and_time): Compute both time and nonspecialized
29981 (estimate_ipcp_clone_size_and_time): Update.
29982 (inline_merge_summary): Update.
29983 (do_estimate_edge_time): Update.
29984 (do_estimate_edge_size): Update.
29985 (do_estimate_edge_hints): Update.
29986 (inline_read_section, inline_write_summary): Stream both new predicates.
29987 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
29989 (compute_inlined_call_time): Cleanup.
29990 (big_speedup_p): Update.
29991 (edge_badness): Update.
29992 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
29993 (size_time_entry): Replace predicate by exec_predicate and
29994 nonconst_predicate.
29995 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
29996 (estimate_edge_time): Return also nonspec_time.
29997 (reset_edge_growth_cache): Update.
29999 2017-04-29 Jakub Jelinek <jakub@redhat.com>
30001 PR rtl-optimization/80491
30002 * ifcvt.c (noce_process_if_block): When looking for x setter
30003 with missing else_bb, don't check only the insn right before
30004 cond_earliest, but look for the last insn that x is modified in
30005 within the same bb.
30007 PR rtl-optimization/80491
30008 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
30010 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
30012 PR tree-optimization/80487
30013 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
30015 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30017 PR tree-optimization/79697
30018 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
30019 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
30020 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
30022 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
30023 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
30025 2017-04-28 Martin Sebor <msebor@redhat.com>
30027 PR tree-optimization/80523
30028 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
30029 (init_target_to_host_charmap, target_to_host, target_strtol10): New
30031 (maybe_warn, format_directive, parse_directive): Use new functions.
30032 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
30034 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
30036 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
30038 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
30040 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
30041 target_header_dir): Set correctly.
30042 * configure: Regenerated.
30043 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30044 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30045 instead of SYSTEM_HEADER_DIR.
30047 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
30049 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
30050 (estimate_local_effects): Likewise.
30051 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
30052 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
30053 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
30054 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
30055 do_estimate_edge_time, estimate_edge_time): Likewise.
30056 * ipa-inline-analysis.c (estimate_node_size_and_time,
30057 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
30058 (estimate_time_after_inlining): Remove.
30060 2017-04-28 Martin Liska <mliska@suse.cz>
30062 * doc/gcov.texi: Enhance documentation of gcov.
30064 2017-04-28 Martin Liska <mliska@suse.cz>
30066 * doc/gcov.texi: Sort options in alphabetic order.
30067 * doc/gcov-dump.texi: Likewise.
30068 * doc/gcov-tool.texi: Likewise.
30069 * gcov.c (print_usage): Likewise.
30070 * gcov-dump.c (print_usage): Likewise.
30071 * gcov-tool.c (print_merge_usage_message): Likewise.
30072 (print_rewrite_usage_message): Likewise.
30073 (print_overlap_usage_message): Likewise.
30075 2017-04-28 Martin Liska <mliska@suse.cz>
30077 PR gcov-profile/53915
30078 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
30080 2017-04-28 Martin Liska <mliska@suse.cz>
30082 PR gcov-profile/79891
30083 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
30084 is marked by compiler as living on a line.
30085 (get_cycles_count): Remove usage of the union.
30086 (output_intermediate_file): Likewise.
30087 (find_source): Fix GNU coding style.
30088 (accumulate_line_counts): Remove old non-all block mode.
30089 (output_lines): Remove usage of the union.
30090 * profile.c (output_location): Include all BBs, even if
30091 belonging to a same line (and file) as a previous BB.
30093 2017-04-28 Martin Liska <mliska@suse.cz>
30095 * gcov.c (process_args): Handle new argument 'w'.
30096 (read_graph_file): Assign ID to BBs.
30097 (output_branch_count): Display BB # if verbose flag is set.
30098 (output_lines): Likewise for arcs.
30099 (print_usage): Add '--verbose' option help.
30100 * doc/gcov.texi: Document --verbose (-w) option.
30102 2017-04-28 Martin Liska <mliska@suse.cz>
30104 * gcov.c (struct block_location_info): New struct.
30105 (process_file): Fill up the new structure.
30106 (read_graph_file): Replace usage of encoding by the newly added
30108 (add_line_counts): Likewise.
30109 (accumulate_line_counts): Remove usage of the union.
30110 (function_info::function_info): New function.
30111 (function_info::~function_info): Likewise.
30112 (process_file): Call delete instead of release_function.
30113 (release_function): Release the function.
30114 (release_structures): Call delete instead of release_function.
30115 (solve_flow_graph): Replace usage of num_blocks.
30116 (find_exception_blocks): Likewise.
30117 (output_lines): Fix GNU coding style.
30119 2017-04-28 Martin Liska <mliska@suse.cz>
30122 * coverage.c (coverage_remove_note_file): New function.
30123 * coverage.h: Declare the function.
30124 * toplev.c (finalize): Clean if an error has been seen.
30126 2017-04-28 Martin Liska <mliska@suse.cz>
30128 PR gcov-profile/80031
30129 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
30130 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
30131 * gcov.c (read_graph_file): Read just number of blocks.
30132 * profile.c (branch_prob): Do not stream 0 flags per a basic
30135 2017-04-28 Martin Liska <mliska@suse.cz>
30137 * gcov-dump.c (tag_*): Add new argument to declarations.
30138 (dump_gcov_file): Likewise.
30139 (tag_blocks): Add and use new argument depth.
30140 (tag_arcs): Likewise.
30141 (tag_lines): Likewise.
30142 (tag_counters): Likewise.
30143 (tag_summary): Likewise.
30144 (dump_working_sets): Use depth to do a proper indentation.
30146 2017-04-28 Jakub Jelinek <jakub@redhat.com>
30149 * cgraph.h (symtab_node::debug_symtab): No longer inline.
30150 * symtab.c (symtab_node::debug_symtab): Move definition here.
30152 2017-04-28 Richard Biener <rguenther@suse.de>
30154 * lto-streamer.h (LTO_major_version): Bump to 7.
30156 2017-04-28 Richard Biener <rguenther@suse.de>
30158 * tree-vrp.c (assert_info): New struct.
30159 (add_assert_info): New helper.
30160 (register_edge_assert_for_2): Refactor to add asserts to a vector
30162 (register_edge_assert_for_1): Likewise.
30163 (register_edge_assert_for): Likewise.
30164 (finish_register_edge_assert_for): New helper actually registering
30165 asserts where live on edge.
30166 (find_conditional_asserts): Adjust.
30167 (find_switch_asserts): Likewise.
30168 (evrp_dom_walker::try_find_new_range): Generalize.
30169 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
30171 2017-04-27 Marek Polacek <polacek@redhat.com>
30174 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
30175 arg10 and arg11 to itype.
30177 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
30179 * doc/extend.texi (Object Size Checking): Improve grammar.
30181 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
30184 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
30185 that the logic for permitting reciprocal estimates matches that
30188 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30191 * tree.c (type_cache_hasher::equal): Only compare
30192 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
30193 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
30194 non-aggregate element types.
30195 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
30196 about the flag on ARRAY_TYPEs in the comment, formatting fix.
30198 2017-04-27 Richard Biener <rguenther@suse.de>
30200 PR middle-end/80533
30201 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
30202 stripping ARRAY_REFs from MEM_EXPR make sure we're not
30203 keeping a reference to a trailing array.
30205 2017-04-27 Richard Biener <rguenther@suse.de>
30207 PR middle-end/80539
30208 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
30209 being in loop-closed SSA form conservatively.
30210 (chrec_fold_multiply_poly_poly): Likewise.
30212 2017-04-27 Tamar Christina <tamar.christina@arm.com>
30214 PR middle-end/79665
30215 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
30216 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
30218 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30221 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
30222 (aarch64_function_arg_alignment): Return unsigned int again, but still
30223 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
30224 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30225 Don't emit -Wpsabi note.
30226 (aarch64_function_arg_boundary): Likewise.
30227 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30230 2017-04-26 Nathan Sidwell <nathan@acm.org>
30232 * tree.h (crc32_unsigned_n): Declare.
30233 (crc32_unsigned, crc32_unsigned): Make inline.
30234 * tree.c (crc32_unsigned_bits): Replace with ...
30235 (crc32_unsigned_n): ... this.
30236 (crc32_unsigned, crc32_byte): Remove.
30237 (crc32_string): Remove unnecessary braces.
30239 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30241 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
30242 * ipa-inline-analysis.c (MAX_TIME): Remove.
30243 (account_size_time): Use sreal for time.
30244 (dump_inline_summary): Update.
30245 (estimate_function_body_sizes): Update.
30246 (estimate_edge_size_and_time): Update.
30247 (estimate_calls_size_and_time): Update.
30248 (estimate_node_size_and_time): Update.
30249 (inline_merge_summary): Update.
30250 (inline_update_overall_summary): Update.
30251 (estimate_time_after_inlining): Update.
30252 (inline_read_section): Update.
30253 (inline_write_summary): Update.
30254 * ipa-inline.c (compute_uninlined_call_time): Update.
30255 (compute_inlined_call_time): Update.
30256 (recursive_inlining): Update.
30257 (inline_small_functions): Update.
30258 (dump_overall_stats): Update.
30259 * ipa-inline.h: Include sreal.h.
30260 (size_time_entry): Turn time to sreal.
30261 (inline_summary): Turn self_time nad time to sreal.
30263 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30265 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
30267 (sreal::stream_out, sreal::stream_in): New.
30268 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
30270 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30272 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
30275 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
30278 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
30280 (dimode_scalar_chain::compute_convert_gain): Ditto.
30281 (dimode_scalar_chain::make_vector_copies): Ditto.
30282 (dimode_scalar_chain::convert_reg): Ditto.
30283 (dimode_scalar_chain::convert_insn): Ditto.
30284 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
30285 (VI248_AVX512BW_1): New mode iterator.
30286 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
30287 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
30290 2017-04-25 Martin Sebor <msebor@redhat.com>
30292 PR tree-optimization/80497
30293 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
30294 constants are representable in HOST_WIDE_INT.
30295 (parse_directive): Ditto.
30297 2017-04-25 Martin Sebor <msebor@redhat.com>
30300 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
30301 (new_zero_array): Adjust signature.
30302 (dom_info::dom_init): Used unsigned rather that size_t.
30303 (dom_info::dom_info): Same.
30305 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30306 Jakub Jelinek <jakub@redhat.com>
30309 * config/arm/arm.c: Include gimple.h.
30310 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30311 returns negative, increment ncrn only if it returned positive.
30312 (arm_needs_doubleword_align): Return int instead of bool,
30313 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
30314 members, but if there is any such non-FIELD_DECL
30315 > PARM_BOUNDARY aligned decl, return -1 instead of false.
30316 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30317 returns negative, increment nregs only if it returned positive.
30318 (arm_setup_incoming_varargs): Likewise.
30319 (arm_function_arg_boundary): Emit -Wpsabi note if
30320 arm_needs_doubleword_align returns negative, return
30321 DOUBLEWORD_ALIGNMENT only if it returned positive.
30323 2017-04-25 Marek Polacek <polacek@redhat.com>
30326 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30327 first argument to type.
30329 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
30332 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
30333 type checks to test for compatibility instead of equality.
30335 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30336 Jakub Jelinek <jakub@redhat.com>
30339 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
30341 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
30342 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
30343 the alignment computation, but return their maximum in warn_alignment.
30344 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30345 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
30347 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
30348 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30351 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30353 * config/arc/simdext.md (dmpyh): Fix typo.
30355 2017-04-25 Richard Biener <rguenther@suse.de>
30357 PR tree-optimization/80492
30358 * alias.c (compare_base_decls): Handle registers with asm
30359 specification conservatively.
30360 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
30361 compare_base_decls returning dont-know properly.
30363 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30365 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
30366 (legitimate_offset_address_p): New function.
30367 (arc_legitimate_address_p): Use above function.
30369 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30371 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
30373 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30375 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
30376 ACCH registers whenever they are available.
30378 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30380 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
30381 double regs fix when not used.
30383 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30385 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
30387 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
30388 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
30390 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30392 * config/arc/arc.c (arc_output_addsi): Check for h-register class
30393 when emitting short ADD instructions.
30395 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30397 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
30399 (cmpsi_cc_c_insn): Likewise.
30400 (cbranchsi4_scratch): Compute proper instruction length using
30401 compact_hreg_operand.
30402 * config/arc/predicates.md (compact_hreg_operand): New predicate.
30404 2017-04-25 Richard Biener <rguenther@suse.de>
30406 PR middle-end/80509
30407 * passes.c (pass_manager::pass_manager): Initialize
30408 m_name_to_pass_map.
30410 2017-04-25 Richard Biener <rguenther@suse.de>
30412 PR tree-optimization/79201
30413 * tree-ssa-sink.c (statement_sink_location): Handle calls.
30415 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30418 * config/s390/vector.md: Split MEM->GPR vector moves for
30419 non-s_operand addresses.
30421 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30424 * config/s390/predicates.md (reload_const_wide_int_operand): New
30426 * config/s390/s390.md ("movti"): Remove d/P alternative.
30427 ("movti_bigconst"): New pattern definition.
30429 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30432 * s390-protos.h (s390_expand_cs_hqi): Removed.
30433 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
30434 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
30435 modes as well as CCZ1mode and CCZmode.
30436 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
30437 signature of s390_emit_compare_and_swap.
30438 (s390_expand_cs_hqi): Likewise, make static.
30439 (s390_expand_cs_tdsi): Generate an explicit compare before trying
30440 compare-and-swap, in some cases.
30441 (s390_expand_cs): Wrapper function.
30442 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
30444 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
30445 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
30446 patterns for small and large integers. Forbid symref memory operands.
30447 Move expander to s390.c. Require cc register.
30448 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
30449 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
30450 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
30451 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
30452 symref memory operands. Remove CC mode and call s390_match_ccmode
30454 ("atomic_exchange<mode>"): Allow and implement all integer modes.
30456 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30458 * config/s390/s390.md (define_peephole2): New peephole to help
30459 combining the load-and-test pattern with volatile memory.
30461 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30463 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
30464 with CCZmode for TARGET_Z196.
30466 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30468 PR rtl-optimization/80501
30469 * combine.c (make_compound_operation_int): Set subreg_code to SET
30470 even for AND with mask of the sign bit of mode.
30472 PR rtl-optimization/80500
30473 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
30474 sum's initial value.
30476 2017-04-25 Julian Brown <julian@codesourcery.com>
30477 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30479 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
30481 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
30483 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
30485 2017-04-25 Julian Brown <julian@codesourcery.com>
30486 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30488 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
30489 (thunderx2t99_sha): New Reservation.
30491 2017-04-25 Julian Brown <julian@codesourcery.com>
30492 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30494 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
30495 type for 1-element load.
30497 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
30499 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
30501 2017-04-24 Martin Jambor <mjambor@suse.cz>
30503 PR tree-optimization/80293
30504 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
30505 char arrays not totally scalarizable if it is false.
30506 (analyze_all_variable_accesses): Pass correct value in the new
30507 parameter. Add a statistics counter.
30509 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
30511 PR middle-end/79931
30512 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
30514 2017-04-24 Richard Biener <rguenther@suse.de>
30516 PR tree-optimization/80494
30517 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
30518 out for complex types.
30520 2017-04-24 Richard Biener <rguenther@suse.de>
30522 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
30523 * tree-ssa-sccvn.c (print_scc): Print SCC size.
30524 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
30525 (DFS): Adjust and never fail.
30526 (sccvn_dom_walker::fail): Remove.
30527 (sccvn_dom_walker::before_dom_children): Adjust.
30528 (run_scc_vn): Likewise and never fail.
30529 * tree-ssa-pre.c (pass_pre::execute): Adjust.
30530 (pass_fre::execute): Likewise.
30532 2017-04-24 Richard Biener <rguenther@suse.de>
30534 PR tree-optimization/79725
30535 * tree-ssa-sink.c (statement_sink_location): Return whether
30536 failure reason was zero uses. Move that check later.
30537 (sink_code_in_bb): Deal with zero uses by removing the stmt
30540 2017-04-24 Richard Biener <rguenther@suse.de>
30543 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
30544 pointer-based references.
30546 2017-04-24 Richard Biener <rguenther@suse.de>
30549 * pass_manager.h (pass_manager::operator new): Remove.
30550 (pass_manager::operator delete): Likewise.
30551 * passes.c (pass_manager::operator new): Remove.
30552 (pass_manager::operator delete): Likewise.
30553 (pass_manager::pass_manager): Zero individual pass members.
30555 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
30558 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
30559 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
30560 Check "XEXP (src, 1)" operand here.
30561 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
30562 Check "XEXP (src, 1)" operand here.
30563 (dimode_scalar_chain::make_vector_copies): Detect count register
30564 of a shift instruction. Zero extend count register from QImode
30565 to DImode to satisfy vector shift pattern count operand predicate.
30566 Substitute vector shift count operand with a DImode copy.
30567 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
30570 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
30572 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
30573 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
30574 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
30575 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
30576 (UNSPEC_NOREX_MEM): Remove definition.
30578 2017-04-21 Richard Biener <rguenther@suse.de>
30580 PR tree-optimization/79547
30581 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
30582 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
30583 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
30584 without any constraints.
30586 2017-04-21 Richard Biener <rguenther@suse.de>
30588 PR tree-optimization/78847
30589 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
30591 2017-04-21 Richard Biener <rguenther@suse.de>
30593 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
30594 (build_distinct_type_copy): Likewise.
30595 (build_variant_type_copy): Likewise.
30596 * tree.c (build_qualified_type): Pass down mem-stat info.
30597 (build_distinct_type_copy): Likewise.
30598 (build_variant_type_copy): Likewise.
30600 2017-04-21 Richard Biener <rguenther@suse.de>
30602 PR tree-optimization/80237
30603 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
30605 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
30606 for a simplified result.
30608 2016-04-21 Richard Biener <rguenther@suse.de>
30610 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
30611 sth as strict as a simple_iv but a chrec without symbols and an
30612 operand defined in the loop we are peeling (and not some subloop).
30613 (propagate_constants_for_unrolling): Propagate all constants.
30615 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
30618 * config/i386/i386.c (print_reg): Remove assert for disalowed
30619 regno values, call output_operand_lossage instead.
30621 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
30624 * config/i386/constraints.md (Yc): New register constraint.
30625 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
30626 Use Yc constraint for alternative 2 of operand 0. Remove
30627 preferred_for_speed attribute.
30629 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
30631 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
30632 lastprivate clauses in SIMT case.
30634 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
30636 * doc/invoke.texi (-Wextra-semi): Document new warning option.
30638 2017-04-20 Richard Biener <rguenther@suse.de>
30640 PR tree-optimization/57796
30641 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
30642 as N scalar stores.
30643 (vect_model_load_cost): Cost gathers as N scalar loads.
30645 2017-04-20 Richard Biener <rguenther@suse.de>
30647 * ggc-page.c (ggc_allocated_p): Rename to ...
30648 (safe_lookup_page_table_entry): ... this and return the lookup
30650 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
30652 2017-04-20 Richard Biener <rguenther@suse.de>
30654 PR tree-optimization/80453
30655 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
30656 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
30657 from the conditions.
30658 (vn_phi_eq): Pass them down.
30659 (vn_phi_lookup): Record them.
30660 (vn_phi_insert): Likewise.
30662 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
30664 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
30665 uninitialized variable warning to avoid buffer overrun.
30667 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
30670 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
30671 is suppressed for '{ 0 }' in C.
30673 2017-04-20 Jakub Jelinek <jakub@redhat.com>
30675 * BASE-VER: Set to 8.0.0.
30677 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
30679 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
30680 priority .init_array and .fini_array section with SECTION_NOTYPE
30683 2017-04-20 Jakub Jelinek <jakub@redhat.com>
30685 PR middle-end/80423
30686 * tree.h (build_array_type): Add typeless_storage default argument.
30687 * tree.c (type_cache_hasher::equal): Also compare
30688 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
30689 (build_array_type): Add typeless_storage argument, set
30690 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
30692 (build_nonshared_array_type): Adjust build_array_type_1 caller.
30693 (build_array_type): Likewise. Add typeless_storage argument.
30695 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
30696 Jakub Jelinek <jakub@redhat.com>
30698 PR tree-optimization/80426
30699 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
30700 operation on symbolic operands, also compute the overflow for the
30701 invariant part when the operation degenerates into a negation.
30703 2017-04-19 Jakub Jelinek <jakub@redhat.com>
30706 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
30707 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
30710 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
30712 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
30715 * config/avr/avr.c (tree.h): Include it.
30716 (cgraph.h): Include it.
30717 (avr_encode_section_info): Don't warn for uninitialized progmem
30718 variable if it's just an alias.
30720 2017-04-19 Richard Biener <rguenther@suse.de>
30723 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
30724 when needed by AutoPGO.
30726 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
30729 * doc/lto.texi: Remove an extra 'that'.
30731 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
30733 PR rtl-optimization/80429
30734 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
30735 are only used in debug insns.
30737 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
30738 Vladimir Makarov <vmakarov@redhat.com>
30740 * config/sparc/predicates.md (input_operand): Add comment. Return
30741 true for any memory operand when LRA is in progress.
30742 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
30744 2017-04-18 Jeff Law <law@redhat.com>
30747 * mips.md ({return,simple_return}_internal): Do not overwrite
30750 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30752 PR tree-optimization/80443
30753 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
30754 instead of adding 1, subtract -1 and similarly instead of subtracting
30757 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
30759 PR rtl-optimization/80357
30760 * haifa-sched.c (tmp_bitmap): New variable.
30761 (model_recompute): Handle duplicate use records.
30762 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
30763 (free_global_sched_pressure_data): Free it.
30765 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
30768 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
30769 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30770 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30771 instead of SYSTEM_HEADER_DIR.
30773 2017-04-18 Jeff Law <law@redhat.com>
30775 PR middle-end/80422
30776 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
30777 predecessors after walking up the insn chain.
30779 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30782 * dwarf2out.c (modified_type_die): Try harder not to emit internal
30783 sizetype type into debug info.
30785 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30788 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
30789 unneeded test for TARGET_UPPER_REGS_SF.
30790 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
30792 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30795 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
30796 instead of gsi_after_labels.
30798 2017-04-18 Jeff Law <law@redhat.com>
30800 * regcprop.c (maybe_mode_change): Avoid creating copies of the
30804 2017-04-13 Jeff Law <law@redhat.com>
30805 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30806 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30808 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
30811 * config/avr/avr.c (intl.h): Include it.
30812 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
30814 2017-04-18 Martin Liska <mliska@suse.cz>
30816 PR gcov-profile/78783
30817 * gcov-tool.c (gcov_output_files): Validate that destination
30818 file is either removed by the tool or by a user.
30820 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30821 Guy Benyei <guybe@mellanox.com>
30823 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
30824 block, and do not negate it, the stored id is already negative.
30826 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30828 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
30830 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
30833 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
30834 masks of options that should be turned off if the VSX vector
30835 options are turned off.
30836 (OTHER_P8_VECTOR_MASKS): Likewise.
30837 (OTHER_VSX_VECTOR_MASKS): Likewise.
30838 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
30839 rs6000_disable_incompatible_switches to validate no type switches
30841 (rs6000_incompatible_switch): New function to disallow turning on
30842 other vector options if -mno-vsx, -mno-power8-vector, or
30843 -mno-power9-vector are specified.
30845 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30847 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
30849 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30851 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
30852 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
30853 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
30854 (ARG_POINTER_CFA_OFFSET): Likewise.
30856 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30858 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
30859 conditions to take advantage of various optimizations.
30861 2017-04-13 Jeff Law <law@redhat.com>
30863 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30864 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30865 (zero_extendsidi2_dext): Likewise.
30867 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30870 * fold-const.c (fold_ternary_loc): Revert
30871 use op0 instead of fold_convert_loc (loc, type, arg0) part of
30874 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
30876 PR rtl-optimization/80343
30877 * lra-remat.c (update_scratch_ops): Assign original hard reg to
30878 new scratch pseudo.
30880 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
30883 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
30884 to ubsan_encode_value.
30886 2017-04-13 Jeff Law <law@redhat.com>
30888 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
30889 appearing in DEBUG_INSNs.
30891 2017-04-13 Martin Liska <mliska@suse.cz>
30893 PR gcov-profile/80413
30894 * gcov-io.c (gcov_write_string): Copy to buffer just when
30895 allocated size is greater than zero.
30897 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30900 * dwarf2out.c (decls_for_scope): Ignore declarations of
30901 current_function_decl in BLOCK_NONLOCALIZED_VARS.
30903 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
30906 * ipa-visibility.c (non_local_p): Fix typos.
30907 (localize_node): When localizing symbol in same comdat group,
30908 dissolve the group only when we know external symbols are going
30910 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
30912 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30914 PR tree-optimization/79390
30915 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
30916 order does not result in usable sequence, retry with reversed operand
30922 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
30923 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
30924 op0 instead of fold_convert_loc (loc, type, arg0).
30926 2017-04-12 Jeff Law <law@redhat.com>
30928 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
30929 has a delay slot in the generated code.
30931 * config/cris/cris.md (cris_preferred_reload_class): Return
30932 GENNONACR_REGS rather than GENERAL_REGS.
30934 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30937 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
30938 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
30939 signedness of the result type.
30941 2017-04-12 Richard Biener <rguenther@suse.de>
30942 Jeff Law <law@redhat.com>
30944 PR tree-optimization/80359
30945 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
30946 trim stores to TARGET_MEM_REFs.
30948 2017-04-12 Richard Biener <rguenther@suse.de>
30950 PR tree-optimization/79390
30951 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
30952 threading case even more.
30954 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
30957 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
30958 for quad_address_p for TImode, instead of just not indexed_address.
30960 2017-04-12 Richard Biener <rguenther@suse.de>
30961 Bernd Edlinger <bernd.edlinger@hotmail.de>
30963 PR middle-end/79671
30964 * alias.c (component_uses_parent_alias_set_from): Handle
30965 TYPE_TYPELESS_STORAGE.
30966 (get_alias_set): Likewise.
30967 * tree-core.h (tree_type_common): Add typeless_storage flag.
30968 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
30969 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
30970 for types containing members with TYPE_TYPELESS_STORAGE.
30971 (place_field): Likewise.
30972 (layout_type): Likewise for ARRAY_TYPE.
30973 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
30974 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
30975 TYPE_TYPELESS_STORAGE.
30976 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
30978 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30981 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30982 first argument to type.
30984 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30988 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
30989 CONST0_RTX (mode) rather than const0_rtx where appropriate.
30990 (rs6000_expand_binop_builtin): Likewise.
30991 (rs6000_expand_ternop_builtin): Likewise; also add missing
30992 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
30993 vshasigma built-ins.
30994 * doc/extend.texi: Document that vec_xxpermdi's third argument
30995 must be a constant.
30997 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
30999 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
31000 Use shift_const cost parameter when calculating gain of STV shifts.
31002 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
31004 PR rtl-optimization/70478
31005 * lra-constraints.c (process_alt_operands): Check memory for
31006 disfavoring memory insn operand.
31008 2017-04-11 Jakub Jelinek <jakub@redhat.com>
31010 PR middle-end/80100
31011 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
31012 left shift in unsigned HOST_WIDE_INT type.
31014 PR rtl-optimization/80385
31015 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
31016 (not (neg X)) into (plus X -1) for complex or non-integral modes.
31019 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
31020 if they have any depend clauses.
31022 2017-04-11 Martin Liska <mliska@suse.cz>
31025 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
31026 * ipa-split.c (split_function): Create a local comdat symbol
31027 if caller is in a comdat group.
31029 2017-04-11 Martin Liska <mliska@suse.cz>
31032 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
31035 2017-04-11 Martin Sebor <msebor@redhat.com>
31037 PR middle-end/80364
31038 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
31039 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
31041 (directive::set_width, directive::set_precision, format_character):
31043 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
31046 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
31049 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
31050 conflict, set target->arch_name instead of target->cpu_name.
31052 2017-04-11 Richard Biener <rguenther@suse.de>
31054 PR tree-optimization/80374
31055 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
31056 build_zero_cst, remove fold_convertible_p check again.
31058 2017-04-11 Martin Liska <mliska@suse.cz>
31061 * ubsan.c (instrument_object_size): Do not instrument register
31064 2017-04-11 Jakub Jelinek <jakub@redhat.com>
31067 * config/i386/i386-builtin-types.def
31068 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
31069 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
31070 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
31071 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
31072 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
31073 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
31074 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
31075 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
31076 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
31077 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
31078 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
31079 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
31080 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
31081 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
31082 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
31083 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
31084 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
31085 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
31086 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
31087 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
31088 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
31089 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
31090 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
31091 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
31092 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
31093 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
31094 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
31095 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
31096 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
31097 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
31098 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
31099 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
31100 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
31101 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
31102 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
31103 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
31104 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
31105 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
31106 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
31107 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
31108 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
31109 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
31110 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
31111 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
31112 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
31113 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
31115 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
31116 flag to second_arg_count, handle 4 argument function type _COUNT
31117 aliases, handle second_arg_count on second argument rather than last.
31119 2017-04-10 Jeff Law <law@redhat.com>
31121 PR tree-optimization/80374
31122 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
31123 record anything if we can not convert integer_zero_node to the
31126 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
31129 * config/rs6000/rs6000.c (rs6000_option_override_internal):
31130 Enhance special handling given to the TARGET_P9_MINMAX option in
31131 relation to certain other options.
31133 2017-04-10 Bin Cheng <bin.cheng@arm.com>
31135 PR tree-optimization/80153
31136 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
31137 remove POINTER_PLUS_EXPR's base part directly, rather than through
31140 2017-04-10 Richard Biener <rguenther@suse.de>
31141 Bin Cheng <bin.cheng@arm.com>
31143 PR tree-optimization/80153
31144 * tree-affine.c (aff_combination_to_tree): Get base pointer from
31145 the first element of pointer type aff_tree. Build result expr in
31147 (add_elt_to_tree): Convert to type unconditionally. Remove other
31148 fold_convert calls.
31149 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
31150 (rewrite_use_nonlinear_expr): Check invariant using iv information.
31152 2017-04-10 Richard Biener <rguenther@suse.de>
31154 * tree-ssa-structalias.c (find_func_aliases): Properly handle
31157 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
31159 PR rtl-optimization/70478
31160 * lra-constraints.c (curr_small_class_check): New.
31161 (update_and_check_small_class_inputs): New.
31162 (process_alt_operands): Update curr_small_class_check. Disfavor
31163 alternative insn memory operands. Check available regs for small
31166 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31169 * config/mips/mips.opt (-mvirt): Update description.
31170 * doc/invoke.texi (-mvirt): Likewise.
31172 2017-04-10 Richard Biener <rguenther@suse.de>
31174 PR middle-end/80362
31175 * fold-const.c (fold_binary_loc): Look at unstripped ops when
31176 looking for NEGATE_EXPR in -A / -B to A / B folding.
31178 2017-04-10 Martin Liska <mliska@suse.cz>
31180 PR gcov-profile/80224
31181 * gcov.c (print_usage): Fix usage string.
31182 (get_gcov_intermediate_filename): Remove.
31183 (output_gcov_file): Use both for normal and intermediate format.
31184 (generate_results): Do not initialize special file for
31185 intermediate format.
31187 2017-04-10 Richard Biener <rguenther@suse.de>
31189 PR tree-optimization/80304
31190 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
31193 2017-04-10 Nathan Sidwell <nathan@acm.org>
31196 * config/rs6000/rs6000.c (rs6000_vector_type): New.
31197 (rs6000_init_builtins): Use it.
31199 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31201 * config/arm/arm.md (<mrc>): Add mode to SET source.
31202 (<mrrc>): Likewise.
31204 2017-04-10 Richard Biener <rguenther@suse.de>
31206 PR middle-end/80344
31207 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
31209 2017-04-10 Jakub Jelinek <jakub@redhat.com>
31212 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
31213 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
31214 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
31215 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
31216 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
31217 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
31218 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
31219 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
31220 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
31221 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
31222 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
31223 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
31224 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
31225 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
31226 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
31227 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
31228 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
31229 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
31230 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
31231 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
31232 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
31233 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
31234 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
31236 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
31238 PR rtl-optimization/70478
31239 * lra-constraints.c: Reverse the last patch.
31241 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
31243 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
31244 Add comment for WCHAR_T.
31246 2017-04-08 Martin Liska <mliska@suse.cz>
31249 2017-04-07 Martin Liska <mliska@suse.cz>
31252 * ipa-split.c (split_function): Add function part to a same comdat
31255 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31258 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
31260 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
31262 * rs6000/rs6000.c (vec_load_pendulum): Rename...
31263 (vec_pairing): ...to this.
31264 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
31265 (rs6000_sched_init): Adjust for name change.
31266 (struct rs6000_sched_context): Likewise.
31267 (rs6000_init_sched_context): Likewise.
31268 (rs6000_set_sched_context): Likewise.
31270 2017-04-07 Jakub Jelinek <jakub@redhat.com>
31276 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
31278 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
31279 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
31280 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
31282 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
31284 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
31286 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31288 PR rtl-optimization/70703
31289 * ira-color.c (update_conflict_hard_regno_costs): Use
31290 int64_t instead of HOST_WIDE_INT.
31292 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31294 PR rtl-optimization/70478
31295 * lra-constraints.c (process_alt_operands): Disfavor alternative
31296 insn memory operands.
31298 2017-04-07 Jeff Law <law@redhat.com>
31300 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
31301 CALL and NOTE_INSN_CALL_ARG_LOCATION.
31303 2017-04-07 Martin Liska <mliska@suse.cz>
31306 * config/aarch64/aarch64.c (aarch64_process_target_attr):
31307 Show error message instead of an ICE.
31309 2017-04-07 Martin Liska <mliska@suse.cz>
31312 * ipa-split.c (split_function): Add function part to a same comdat
31315 2017-04-07 Richard Biener <rguenther@suse.de>
31317 PR middle-end/80341
31318 * tree.c (get_unwidened): Also handle ! for_type case for
31320 * convert.c (do_narrow): Split out from ...
31321 (convert_to_integer_1): ... here. Do not pass final truncation
31322 type to get_unwidened for TRUNC_DIV_EXPR.
31324 2017-04-07 Richard Biener <rguenther@suse.de>
31326 * tree-affine.c (wide_int_ext_for_comb): Take type rather
31328 (aff_combination_const): Adjust.
31329 (aff_combination_scale): Likewise.
31330 (aff_combination_add_elt): Likewise.
31331 (aff_combination_add_cst): Likewise.
31332 (aff_combination_convert): Likewise.
31333 (add_elt_to_tree): Likewise. Remove unused argument.
31334 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
31336 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
31338 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
31340 * config/arm/arm.c (arm_default_short_enums): Use
31341 ARM_DEFAULT_SHORT_ENUMS.
31342 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
31344 2017-04-06 Jakub Jelinek <jakub@redhat.com>
31347 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
31348 members with redundant out-of-class redeclaration.
31350 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31353 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
31354 * config/i386/i386.md (*zero_extendsidi2):
31355 Add (?*x,*x) and (?*v,*v) alternatives.
31357 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31360 * config/i386/i386.c (ix86_expand_builtin)
31361 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
31362 mode from insn data. Convert operands to insn operand mode.
31363 Copy operands that don't satisfy insn predicate to a register.
31365 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
31367 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
31370 2017-04-06 Richard Biener <rguenther@suse.de>
31372 PR tree-optimization/80334
31373 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
31374 preserve alignment of accesses.
31376 2017-04-06 Richard Biener <rguenther@suse.de>
31378 PR tree-optimization/80262
31379 * tree-sra.c (build_ref_for_offset): Preserve address-space
31381 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
31382 Drop useless address-space information on MEM_REF offsets.
31384 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
31386 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
31388 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31390 PR rtl-optimization/70703
31391 * ira-color.c (update_conflict_hard_regno_costs): Use
31392 HOST_WIDE_INT instead of long.
31394 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
31397 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
31398 not defined for x86_64 target. Add -mmmx target option when __SSE2__
31400 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
31401 for x86_64 target. Handle -m3dnowa option.
31403 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31405 PR rtl-optimization/70703
31406 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
31407 (update_conflict_hard_regno_costs): Use long instead of unsigned
31408 arithmetic for cost calculation.
31410 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31411 Bernd Edlinger <bernd.edlinger@hotmail.de>
31414 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
31417 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
31420 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
31421 ptr_mode with Pmode throughout.
31422 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
31423 into probe_stack_range and use DImode.
31425 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
31428 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
31429 call_eh_return is true.
31431 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31433 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
31434 Initialize last_match_fntype_index.
31436 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31439 * tree-nvr.c: Include internal-fn.h.
31440 (pass_return_slot::execute): Ignore internal calls without
31443 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31444 Richard Biener <rguenther@suse.de>
31447 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
31448 captures used multiple times, except for the last use.
31449 * generic-match-head.c: Include gimplify.h.
31451 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31453 PR tree-optimization/79390
31454 * target.h (struct noce_if_info): Declare.
31455 * targhooks.h (default_noce_conversion_profitable_p): Declare.
31456 * target.def (noce_conversion_profitable_p): New target hook.
31457 * ifcvt.h (struct noce_if_info): New type, moved from ...
31458 * ifcvt.c (struct noce_if_info): ... here.
31459 (noce_conversion_profitable_p): Renamed to ...
31460 (default_noce_conversion_profitable_p): ... this. No longer
31462 (noce_try_store_flag_constants, noce_try_addcc,
31463 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
31464 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
31465 instead of noce_conversion_profitable_p.
31466 * config/i386/i386.c: Include ifcvt.h.
31467 (ix86_option_override_internal): Don't override
31468 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
31469 (ix86_noce_conversion_profitable_p): New function.
31470 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
31471 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
31472 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
31473 * doc/tm.texi: Regenerated.
31475 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31477 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
31480 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31483 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
31484 instructions for small multiply cores.
31486 2017-04-04 Jeff Law <law@redhat.com>
31488 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
31490 (mips_expand_vec_perm_const): Initialize elements in orig_perm
31491 that are not set by the loop over the elements.
31493 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31496 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
31497 int mode, convert_modes it to mode as unsigned, otherwise use
31498 lowpart_subreg to mode rather than SImode.
31499 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
31500 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
31501 Use DImode instead of SImode for the shift count operand.
31502 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
31505 2017-04-04 Richard Biener <rguenther@suse.de>
31507 PR middle-end/80281
31508 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
31509 arithmetic done for the negate or the plus. Simplify.
31510 (A - (-B) -> A + B): Likewise.
31511 * fold-const.c (split_tree): Make sure to not negate pointers.
31513 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
31515 PR rtl-optimization/60818
31516 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
31517 a compare of comparisons with the thing compared if this results
31518 in a different machine mode.
31520 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
31522 * alias.c (base_alias_check): Fix typo in comment.
31523 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
31524 * cgraphunit.c (symbol_table::compile): Likewise.
31525 * collect2.c (maybe_run_lto_and_relink): Likewise.
31526 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
31527 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
31528 * config/avr/avr.c (avr_map_op_t): Likewise.
31529 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
31530 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
31531 * config/epiphany/epiphany.md (movcc): Likewise.
31532 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
31533 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
31535 * config/mips/mips.c (mips_save_restore_reg): Likewise.
31536 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
31537 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
31538 * config/sh/sh.c (sh_rtx_costs): Likewise.
31539 * fold-const.c (fold_truth_andor): Likewise.
31540 * genautomata.c (collapse_flag): Likewise.
31541 * gengtype.h (struct type::u::s): Likewise.
31542 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
31543 * input.c (FORMAT_AMOUNT): Likewise.
31544 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
31545 (known_aggs_to_agg_replacement_list): Likewise.
31546 * ipa-inline-analysis.c: Likewise.
31547 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
31548 * ipa-polymorphic-call.c
31549 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
31550 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
31551 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
31553 * modulo-sched.c (apply_reg_moves): Likewise.
31554 * omp-expand.c (build_omp_regions_1): Likewise.
31555 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
31556 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
31557 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
31558 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
31559 * value-prof.c: Likewise.
31560 * var-tracking.c (val_reset): Likewise.
31562 2017-04-03 Richard Biener <rguenther@suse.de>
31564 PR tree-optimization/80275
31565 * fold-const.c (split_address_to_core_and_offset): Handle
31568 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
31570 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
31571 descriptors is at least equal to that of functions.
31573 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
31575 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
31577 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
31580 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
31581 (mov<IMOD4:mode>): New expander.
31582 (*mov<IMOD4:mode>_internal): New insn and split pattern.
31584 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
31586 PR rtl-optimization/79405
31587 * fwprop.c (propagations_left): New variable.
31588 (forward_propagate_into): Decrement it.
31589 (fwprop_init): Initialize it.
31590 (fw_prop): If the variable has reached zero, stop propagating.
31591 (fwprop_addr): Ditto.
31593 2017-03-31 Jakub Jelinek <jakub@redhat.com>
31596 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
31597 a FUNCTION_DECL, pass it as decl instead of origin to
31600 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
31602 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
31605 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
31608 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
31609 TARGET_VSX_SMALL_INTEGER.
31611 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31613 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
31614 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
31616 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31618 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
31619 extraction from odd-numbered MSA register.
31621 2017-03-31 Jakub Jelinek <jakub@redhat.com>
31623 PR middle-end/80173
31624 * expmed.c (store_bit_field_1): Don't attempt to create
31625 a word subreg out of hard registers wider than word if they
31626 have HARD_REGNO_NREGS of 1 for their mode.
31628 PR middle-end/80163
31629 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
31630 conversions to integer types wider than word and pointer.
31633 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
31634 (rtx_equal_for_cselib_p): Pass 0 to it.
31635 * cselib.c (cselib_hasher::equal): Likewise.
31636 (rtx_equal_for_cselib_1): Add depth argument. If depth
31637 is 128, don't look up VALUE locs and punt. Increment
31638 depth in recursive calls when walking VALUE locs.
31640 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
31642 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
31643 (make_gcov_file_name): Use the canonical path name for generating
31645 (read_line): Fix handling of files with ascii null bytes.
31647 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
31649 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
31650 to initialise a vector register instead
31651 of using a const_int.
31653 2017-03-30 Jakub Jelinek <jakub@redhat.com>
31655 PR translation/80189
31656 * gimplify.c (omp_default_clause): Use %qs instead of %s in
31657 diagnostic messages.
31659 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
31662 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
31663 (dfp_diex_<mode>): Update mode of operand 1.
31664 * doc/extend.texi (dxex, dxexq): Document change to return type.
31665 (diex, diexq): Document change to argument type.
31667 2017-03-30 Martin Jambor <mjambor@suse.cz>
31670 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
31671 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
31672 it reflects the signature changes performed at the callee side.
31673 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
31674 to cgraph_build_function_type_skip_args.
31675 (build_function_decl_skip_args): Adjust call to the above function.
31677 2017-03-30 Jakub Jelinek <jakub@redhat.com>
31680 * config/i386/sse.md
31681 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
31682 register as dest whenever it is a MEM not rtx_equal_p to the
31683 corresponding dup operand, and when forcing into reg move the
31684 reg into the memory afterwards.
31685 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
31686 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
31687 for the force_reg mode.
31688 (avx512vl_vextractf128<mode>): Use register as dest either
31689 always when a MEM, or when it is a MEM not rtx_equal_p to the
31690 corresponding dup operand, or even not when it is a CONST_VECTOR
31691 depending on the mode and lo vs. hi.
31692 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
31694 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
31695 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
31696 Likewise. Require that operands[2] is even.
31697 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
31698 Remove extraneous parens. Require that operands[2] is a multiple
31700 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
31701 operands[0] is a MEM if <mask_applied>, the predicates/constraints
31702 disallow memory then.
31704 2017-03-30 Richard Biener <rguenther@suse.de>
31706 PR tree-optimization/77498
31707 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
31708 to non-constants over backedges.
31710 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
31712 PR rtl-optimization/80233
31713 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
31714 as last_combined_insn. Do not test for BARRIER_P separately.
31716 2017-03-29 Andreas Schwab <schwab@suse.de>
31719 * calls.c (prepare_call_address): Convert funexp to Pmode before
31720 copying to temp reg.
31722 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31724 PR tree-optimization/80158
31725 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
31726 Handle possible future case of more than one alternate
31728 (replace_rhs_if_not_dup): Likewise.
31729 (replace_one_candidate): Likewise.
31731 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
31733 PR rtl-optimization/80193
31734 * ira.c (ira): Do not check allocation for LRA.
31736 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
31738 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
31739 (nvptx_output_simt_exit): Declare.
31740 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
31741 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
31742 (init_softstack_frame): Move initialization of crtl->is_leaf to...
31743 (nvptx_declare_function_name): ...here. Emit declaration of local
31744 memory space buffer for omp_simt_enter insn.
31745 (nvptx_output_unisimt_switch): New.
31746 (nvptx_output_softstack_switch): New.
31747 (nvptx_output_simt_enter): New.
31748 (nvptx_output_simt_exit): New.
31749 * config/nvptx/nvptx.h (struct machine_function): New fields
31750 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
31751 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
31752 (UNSPECV_SIMT_EXIT): Ditto.
31753 (omp_simt_enter_insn): New insn.
31754 (omp_simt_enter): New expansion.
31755 (omp_simt_exit): New insn.
31756 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
31758 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
31759 (expand_GOMP_SIMT_ENTER_ALLOC): New.
31760 (expand_GOMP_SIMT_EXIT): New.
31761 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
31762 (GOMP_SIMT_ENTER_ALLOC): Ditto.
31763 (GOMP_SIMT_EXIT): Ditto.
31764 * target-insns.def (omp_simt_enter): New insn.
31765 (omp_simt_exit): Ditto.
31766 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
31768 (lower_rec_simd_input_clauses): Implement SIMT privatization.
31769 (lower_rec_input_clauses): Likewise.
31770 (lower_lastprivate_clauses): Handle SIMT privatization.
31772 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
31773 (ompdevlow_adjust_simt_enter): New.
31774 (find_simtpriv_var_op): New.
31775 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
31776 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
31778 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
31779 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
31780 (copy_decl_for_dup_finish): Ditto.
31782 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
31784 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
31787 * config/i386/i386.c (ix86_option_override_internal): Always
31788 allow -mpreferred-stack-boundary=3 for 64-bit targets.
31790 2017-03-28 Bin Cheng <bin.cheng@arm.com>
31792 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
31794 2017-03-28 Bin Cheng <bin.cheng@arm.com>
31796 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
31797 mark new edge's irreducible flag accordign to it.
31798 (vect_do_peeling): Check loop preheader edge's irreducible flag
31799 and pass it to function slpeel_add_loop_guard.
31801 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
31803 PR tree-optimization/80218
31804 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
31805 Update block frequencies and counts.
31807 2017-03-28 Richard Biener <rguenther@suse.de>
31809 PR tree-optimization/78644
31810 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
31811 of a simplification result we may not use it at all.
31813 2017-03-28 Richard Biener <rguenther@suse.de>
31816 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
31817 without arguments, generate default definition of a SSA name.
31819 2017-03-28 Richard Biener <rguenther@suse.de>
31821 PR middle-end/80222
31822 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
31823 TYPE_REF_CAN_ALIAS_ALL references.
31824 * fold-const.c (fold_indirect_ref_1): Likewise.
31826 2017-03-28 Martin Liska <mliska@suse.cz>
31829 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
31830 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
31832 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31833 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
31835 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
31836 (EXTRA_SPECS): Define.
31837 (SUBTARGET_EXTRA_SPECS): Likewise.
31838 (SUBTARGET_CPP_SPEC): Likewise.
31839 * config/arc/elf.h (EXTRA_SPECS): Renamed to
31840 SUBTARGET_EXTRA_SPECS.
31841 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
31843 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31845 * config/arc/simdext.md (vst64_insn): Update pattern.
31846 (vld32wh_insn): Likewise.
31847 (vld32wl_insn): Likewise.
31848 (vld64_insn): Likewise.
31849 (vld32_insn): Likewise.
31851 2017-03-28 Marek Polacek <polacek@redhat.com>
31854 * fold-const.c (fold_comparison): Use protected_set_expr_location
31855 instead of SET_EXPR_LOCATION.
31857 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
31859 * tree.c (add_expr): Avoid name lookup warning.
31861 2017-03-27 Jeff Law <law@redhat.com>
31863 PR tree-optimization/80216
31864 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
31865 function name. Limit recursion depth.
31866 (record_temporary_equivalences): Corresponding changes.
31868 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
31870 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
31873 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31876 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
31878 * cfgcleanup.c (reg_note_cfa_p): New array.
31879 (insns_have_identical_cfa_notes): New function.
31880 (old_insns_match_p): Don't cross-jump in between /f
31881 and non-/f instructions. If both i1 and i2 are frame related,
31882 verify all CFA notes, their order and content.
31884 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
31887 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
31888 HImode and SImode with zero extend to DImode to one insn.
31889 (bswap<mode>2_extenddi): Likewise.
31890 (bswapsi2_extenddi): Likewise.
31891 (bswaphi2_extendsi): Likewise.
31892 (bswaphi2): Combine bswap HImode and SImode into one insn.
31893 Separate memory insns from swapping register.
31894 (bswapsi2): Likewise.
31895 (bswap<mode>2): Likewise.
31896 (bswaphi2_internal): Delete, no longer used.
31897 (bswapsi2_internal): Likewise.
31898 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
31899 store, and gpr<-gpr swap insns.
31900 (bswap<mode>2_store): Likewise.
31901 (bswaphi2_reg): Register only splitter, combine with the splitter.
31902 (bswaphi2 splitter): Likewise.
31903 (bswapsi2_reg): Likewise.
31904 (bswapsi2 splitter): Likewise.
31905 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
31906 the insns into load, store, and register/register insns.
31907 (bswapdi2_ldbrx): Likewise.
31908 (bswapdi2_load): Likewise.
31909 (bswapdi2_store): Likewise.
31910 (bswapdi2_reg): Likewise.
31912 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
31914 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
31915 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
31917 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
31920 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
31922 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31923 special handling for target option conflicts between dform
31924 options (-mpower9-dform, -mpower9-dform-vector,
31925 -mpower9-dform-scalar) and -mno-direct-move.
31927 2017-03-27 Richard Biener <rguenther@suse.de>
31929 PR tree-optimization/80181
31930 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
31932 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31934 * config/arc/predicates.md (move_double_src_operand): Replace the
31935 call to move_double_src_operand with a call to address_operand.
31937 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31939 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
31940 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
31941 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
31943 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31945 * config/arc/predicates.md (long_immediate_loadstore_operand):
31946 Consider scaled addresses cases.
31948 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31950 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
31951 restored when in interrupt.
31952 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
31953 doesn't have delay slot.
31955 2017-03-27 Richard Biener <rguenther@suse.de>
31958 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
31959 inlined thunk clones.
31961 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31964 * asan.c (instrument_derefs): Copy over last operand from
31965 original COMPONENT_REF to the new COMPONENT_REF with
31966 DECL_BIT_FIELD_REPRESENTATIVE.
31967 * ubsan.c (instrument_object_size): Likewise.
31969 2017-03-27 Richard Biener <rguenther@suse.de>
31971 PR tree-optimization/80170
31972 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
31973 sure DR/SCEV didnt fold in constants we do not see when looking
31974 at the reference base alignment.
31976 2017-03-27 Richard Biener <rguenther@suse.de>
31978 PR middle-end/80171
31979 * gimple-fold.c (fold_ctor_reference): Properly guard against
31980 NULL return value from canonicalize_constructor_val.
31982 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
31985 * config/i386/i386.c (ix86_expand_builtin)
31986 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
31987 flags reg setting and flags reg using instructions.
31988 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
31989 clobbering instructions to zero extend op2.
31991 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
31993 * doc/install.texi (Configuration) <--with-aix-soname>:
31994 Update link to AIX ld.
31996 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
31998 PR rtl-optimization/80160
31999 PR rtl-optimization/80159
32000 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
32001 reg_alternate_class into account.
32003 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
32006 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
32007 to consider in curr_insn_transform.
32009 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32011 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
32012 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
32013 and emit_mode_inner.
32015 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32017 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
32018 argument to the overloaded builtin variants. Use the new flag to
32019 deprecate certain builtin variants.
32020 * config/s390/s390-builtin-types.def: Add new builtin types.
32021 * config/s390/s390-builtins.h: Support new flags field for
32022 overloaded builtins.
32023 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
32024 (s390_macro_to_expand): Enable vector float data type.
32025 (s390_cpu_cpp_builtins_internal): Indicate support of the new
32026 builtins by incrementing the __VEC__ version number.
32027 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
32029 (s390_resolve_overloaded_builtin): Emit error messages depending
32030 on the builtin flags.
32031 * config/s390/s390.c (s390_expand_builtin): Support additional
32032 flags argument. Change error message to match the messages
32033 emitted in s390-c.c.
32034 * config/s390/s390.md: New UNSPEC_* constants.
32035 (op_type): Add new instruction types.
32036 * config/s390/vecintrin.h: Add new builtins and test data class
32038 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
32039 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
32040 (VEC_INEXACT, VEC_NOINEXACT): New constants.
32041 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
32042 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
32043 ("vec_mergel<mode>"): V_HW -> VEC_HW.
32045 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
32046 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
32047 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
32048 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
32050 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
32051 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
32052 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
32053 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
32055 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
32056 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
32057 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
32058 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
32059 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
32060 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
32061 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
32063 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
32064 ("vec_scatter_element<V_HW_4:mode>_DI")
32065 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
32066 ("vec_fpint<mode>", "vflls")
32067 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
32068 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
32069 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
32070 ("*vec_cmphe<mode>_cc"): ... these.
32072 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
32073 mode constant instead of magic value.
32075 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32077 * config/s390/s390.c (s390_expand_vec_compare): Support other
32078 vector floating point modes than just V2DF.
32079 (s390_expand_vcond): Likewise.
32080 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
32081 (s390_cannot_change_mode_class): Prevent mode changes between TF
32082 and V1TF in vector registers.
32083 * config/s390/s390.md (DF, SF): New mode attributes.
32084 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
32085 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
32086 SFmode support for VRs.
32087 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
32089 (VFT, VF_HW): New mode iterators.
32090 (vw, sdx): New mode attributes.
32091 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
32092 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
32093 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
32094 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
32095 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
32096 also the new vector floating point modes. Renaming to ...
32098 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
32099 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
32100 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
32101 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
32102 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
32103 ("vec_unordered<mode>"): ... these.
32105 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
32106 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
32107 ("*vec_extendv2df"): New insn definitions.
32109 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32111 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
32112 ("mulditi3_2", "*muldi3_sign"): New patterns.
32113 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
32114 rename the pattern definition.
32116 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32118 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
32120 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
32122 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32124 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
32125 instruction if possible.
32126 * config/s390/vector.md (vec_halfnumelts): New mode
32128 ("*vec_vllezlf<mode>"): New pattern.
32130 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32132 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
32133 ("popcountv4si2", "popcountv2di2"): Rename to ...
32134 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
32135 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
32137 ("popcount<mode>2_vxe"): New pattern.
32139 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32141 * common/config/s390/s390-common.c (processor_flags_table): Add
32143 * config.gcc: Add arch12.
32144 * config/s390/driver-native.c (s390_host_detect_local_cpu):
32145 Default to arch12 for unknown CPU model numbers.
32146 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
32147 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
32148 PROCESSOR_max sanity check.
32149 * config/s390/s390-opts.h (enum processor_type): Add
32151 * config/s390/s390.c (processor_table): Add arch12.
32152 (s390_expand_builtin): Add check for B_VXE flag.
32153 (s390_issue_rate): Add PROCESSOR_ARCH12.
32154 (s390_get_sched_attrmask): Likewise.
32155 (s390_get_unit_mask): Likewise.
32156 (s390_sched_score): Enable z13 scheduling for arch12.
32157 (s390_sched_reorder): Likewise.
32158 (s390_sched_variable_issue): Likewise.
32159 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
32161 (s390_tune_attr): Use z13 scheduling also for arch12.
32162 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
32163 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
32164 (TARGET_VXE_P): New macros.
32165 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
32166 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
32167 * config/s390/s390.opt: Add arch12 as processor_type.
32169 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32171 * config/s390/s390.md
32172 ("fixuns_truncdddi2", "fixuns_trunctddi2")
32173 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
32174 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
32176 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
32177 Rename expanders to ...
32179 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
32180 ("fixuns_truncdddi2_emu"): ... these.
32182 ("fixuns_trunc<mode>si2_emu"): New expander.
32184 ("*fixuns_truncdfdi2_z13"): Rename to ...
32185 ("*fixuns_truncdfdi2_vx"): ... this.
32187 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32189 * config/s390/2964.md: Remove the single element vector compare
32190 instructions which are no longer used.
32191 * config/s390/s390.c (s390_select_ccmode): Remove handling of
32193 (s390_canonicalize_comparison): Remove handling of DFmode
32195 (s390_expand_vec_compare_scalar): Remove function.
32196 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
32197 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
32199 ("*cmp<mode>_ccs"): Add wfcdb instruction.
32201 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32203 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
32205 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
32206 will anyway by matched by mov<mode>_64dfp.
32208 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32210 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
32211 vlef/vstef. Add missing operand to vleif.
32213 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32215 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
32216 pair for all vector types with 64 bit elements.
32217 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
32218 * config/s390/vector.md (V_HW_64): ... here.
32219 (V_128_NOSINGLE): New mode iterator.
32220 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
32221 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
32222 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
32223 ("*vec_load_pairv2di"): Change to ...
32224 ("*vec_load_pair<mode>"): ... this one.
32226 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32228 * config/s390/constraints.md: Add comments.
32229 (jKK): Reject element sizes > 8 bytes.
32230 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
32232 * config/s390/s390.md: Add the s_operand checks formerly in
32233 s390_split_ok_p to various splitters where they are still
32235 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
32236 for 128 bit vectors. Plus two splitters.
32238 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32240 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
32243 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32246 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
32247 error if the boundary argument is not constant.
32249 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32251 PR rtl-optimization/80112
32252 * loop-doloop.c (doloop_condition_get): Don't check condition
32253 if cmp isn't SET with IF_THEN_ELSE src.
32255 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32257 PR tree-optimization/80158
32258 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
32259 replacing a candidate statement, also replace it for the
32260 candidate's alternate interpretation.
32261 (replace_rhs_if_not_dup): Likewise.
32262 (replace_one_candidate): Likewise.
32264 2017-03-24 Richard Biener <rguenther@suse.de>
32266 PR tree-optimization/80167
32267 * graphite-isl-ast-to-gimple.c
32268 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
32270 (translate_isl_ast_to_gimple::get_rename): Likewise.
32272 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32274 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
32275 handling of certain combinations of target options, including the
32276 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
32277 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
32279 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32282 * config/arm/arm.md (*load_multiple): Add reload_completed to
32283 matching condition.
32285 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32286 Richard Biener <rguenth@suse.de>
32288 PR tree-optimization/79908
32289 PR tree-optimization/80136
32290 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32291 been cast away, gimplify_and_add suffices.
32293 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
32295 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
32297 2017-03-23 Richard Biener <rguenther@suse.de>
32299 PR tree-optimization/80032
32300 * gimplify.c (gimple_push_cleanup): Forced unconditional
32301 cleanups still have to go to the conditional_cleanups
32304 2017-03-22 Jakub Jelinek <jakub@redhat.com>
32306 PR tree-optimization/80072
32307 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
32309 (next_operand_entry_id): Change type to unsigned int.
32310 (sort_by_operand_rank): Make sure to return the right return value
32311 even if unsigned fields are bigger than INT_MAX.
32312 (struct oecount): Change cnt and id type to unsigned int.
32313 (oecount_hasher::equal): Formatting fix.
32314 (oecount_cmp): Make sure to return the right return value
32315 even if unsigned fields are bigger than INT_MAX.
32316 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
32319 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
32320 TREE_READONLY on result if writing it more than once.
32323 * doc/invoke.texi (-fsanitize=thread): Document that with
32324 -fnon-call-exceptions atomics are not able to throw
32328 * tsan.c: Include tree-eh.h.
32329 (instrument_builtin_call): Call maybe_clean_eh_stmt or
32330 maybe_clean_or_replace_eh_stmt where needed.
32331 (instrument_memory_accesses): Add cfg_changed argument.
32332 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
32333 if it returned true.
32334 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
32336 PR rtl-optimization/63191
32337 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
32338 wrapper function, moved the whole old content into ...
32339 (ix86_delegitimize_address_1): ... this. New inline function.
32340 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
32341 true as last argument instead of ix86_delegitimize_address.
32343 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32345 * config/aarch64/aarch64.c (generic_branch_cost): Copy
32346 cortexa57_branch_cost.
32348 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32350 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
32352 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32355 * doc/md.texi (Constraints): Document wA constraint.
32356 * config/rs6000/constraints.md (wA): New.
32357 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
32358 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
32359 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
32360 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
32362 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
32365 * gimplify.c (is_oacc_declared): New function.
32366 (oacc_default_clause): Use it to set default flags for acc declared
32367 variables inside parallel regions.
32368 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
32369 declared variables.
32370 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
32371 declare attribute to any decl as necessary.
32373 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
32376 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
32377 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
32378 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
32379 (arm_arch_lpae): This.
32380 * config/arm/arm.c (arm_arch7ve): Rename into ...
32381 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
32382 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
32385 2017-03-22 Martin Liska <mliska@suse.cz>
32388 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
32389 error message instead of an ICE.
32391 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32393 * doc/extend.texi (6.11 Additional Floating Types): Revise.
32395 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
32397 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
32399 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32402 2017-03-21 Martin Sebor <msebor@redhat.com>
32404 * doc/extend.texi: Use "cannot" instead of "can't."
32405 * doc/hostconfig.texi: Same.
32406 * doc/install.texi: Same.
32407 * doc/invoke.texi: Same.
32408 * doc/loop.texi: Same.
32409 * doc/md.texi: Same.
32410 * doc/objc.texi: Same.
32411 * doc/rtl.texi: Same.
32412 * doc/tm.texi: Same.
32413 * doc/tm.texi.in: Same.
32414 * doc/trouble.texi: Same.
32416 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
32419 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
32420 (collect_checksum_attributes): Set it.
32421 (die_checksum_ordered): Use it.
32423 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32425 PR tree-optimization/79908
32426 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
32427 change: For a VA_ARG whose LHS has been cast away, use
32428 force_gimple_operand to construct the side effects.
32430 2017-03-21 David Malcolm <dmalcolm@redhat.com>
32432 PR translation/80001
32433 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
32434 more amenable to translation.
32435 (oacc_loop_auto_partitions): Likewise.
32437 2017-03-21 Marek Polacek <polacek@redhat.com>
32438 Martin Sebor <msebor@redhat.com>
32440 PR tree-optimization/80109
32441 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
32442 on INTEGRAL_TYPE_P.
32444 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32445 Segher Boessenkool <segher@kernel.crashing.org>
32448 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
32449 check reg_used_between_p between insn and one of succ or succ2
32450 depending on if succ is artificial insn not inserted into insn
32453 2017-03-21 Martin Liska <mliska@suse.cz>
32455 PR gcov-profile/80081
32456 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
32457 * doc/gcc.texi: Include gcov-dump stuff.
32458 * doc/gcov-dump.texi: New file.
32460 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
32462 PR rtl-optimization/79150
32463 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
32464 conditional jump, if the jump is the last insn of the loop.
32466 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32467 Richard Biener <rguenth@suse.de>
32469 PR tree-optimization/79908
32470 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32471 been cast away, use force_gimple_operand to construct the side
32474 2017-03-21 Martin Liska <mliska@suse.cz>
32476 PR libfortran/79956
32477 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
32480 2017-03-21 Brad Spengler <spender@grsecurity.net>
32483 * plugin.c (htab_hash_plugin): New function.
32484 (add_new_plugin): Use it and adjust.
32485 (parse_plugin_arg_opt): Adjust.
32486 (init_one_plugin): Likewise.
32488 2017-03-21 Richard Biener <rguenther@suse.de>
32490 PR tree-optimization/80032
32491 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
32492 if set force the cleanup to happen unconditionally.
32493 (gimplify_target_expr): Push inserted clobbers with force_uncond
32494 to avoid them being removed by control-dependent DCE.
32496 2017-03-21 Richard Biener <rguenther@suse.de>
32498 PR tree-optimization/80122
32499 * tree-inline.c (copy_bb): Do not expans va-arg packs or
32500 va_arg_pack_len when the inlined call stmt requires pack
32502 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
32504 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32507 * tsan.c (instrument_builtin_call): If the memory model argument
32508 is not a constant, assume it is valid.
32511 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
32514 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
32516 PR rtl-optimization/79910
32517 * combine.c (can_combine_p): Do not allow combining an I0 or I1
32518 if its dest is used by an insn before I2 (other than the combined
32519 insns themselves, which are properly handled already).
32521 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
32524 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
32526 * combine.c (record_used_regs): New static function.
32527 (try_combine): Handle situations where there is an additional
32528 instruction between I2 and I3 which needs to have a LOG_LINK
32532 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
32534 * combine.c (try_combine): Delete redundant i1 test. Call
32535 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
32537 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32540 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
32541 alternatives 13/14.
32543 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32545 PR tree-optimization/80054
32546 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
32547 the optimization if a PHI or any of its arguments is not dominated
32548 by the candidate's basis. Use gphi* rather than gimple* as
32550 (replace_profitable_candidates): Clean up a gimple* variable that
32551 should be a gphi* variable.
32553 2017-03-20 Martin Sebor <msebor@redhat.com>
32556 * doc/extend.texi (attribute constructor): Document present limitation.
32558 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
32561 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
32562 __POWER9_VECTOR__ #ifdef control, change template definition to
32563 use Power9-specific built-in function.
32564 (vec_any_eq): Likewise.
32565 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
32566 to control outcomes from this test.
32567 (vector_ae_<mode>p): For VEC_F modes, likewise.
32569 2017-03-20 Ian Lance Taylor <iant@google.com>
32571 * config/i386/i386.c (ix86_function_regparm): Save an extra
32572 register for -fsplit-stack with DECL_STATIC_CHAIN.
32574 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
32577 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
32578 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
32580 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
32582 * config/riscv/riscv.c (riscv_print_operand): Use "fence
32584 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
32587 2017-03-20 Marek Polacek <polacek@redhat.com>
32590 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
32592 2017-03-20 Richard Biener <rguenther@suse.de>
32594 PR tree-optimization/80113
32595 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
32596 allocate extra SSA name for PHI def.
32597 (add_close_phis_to_outer_loops): Likewise.
32598 (add_close_phis_to_merge_points): Likewise.
32599 (copy_loop_close_phi_args): Likewise.
32600 (copy_cond_phi_nodes): Likewise.
32602 2017-03-20 Martin Liska <mliska@suse.cz>
32604 PR middle-end/79753
32605 * tree-chkp.c (chkp_build_returned_bound): Do not build
32606 returned bounds for a LHS that's not a BOUNDED_P type.
32608 2017-03-20 Martin Liska <mliska@suse.cz>
32612 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
32613 COMPLEX_CST and VECTOR_CST.
32615 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32618 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
32619 target operand. A new splitter adds the clobber statement in case
32620 the target operand is dead anyway.
32622 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
32624 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
32625 to age-old versions of binutils and glibc.
32627 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
32629 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
32631 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
32633 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
32635 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
32637 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
32638 requirement for binutils 2.13.
32640 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
32642 * combine.c (try_combine): Delete redundant i1 test. Call
32643 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
32645 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
32647 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
32648 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
32650 <riscv64-*-elf>: Re-arrange section
32651 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
32652 <riscv32-*-linux>: Likewise.
32653 <riscv64-*-elf>: Likewise
32654 <riscv64-*-linux>: Likewise.
32656 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
32659 * aarch64.opt(verbose-cost-dump): Fix typo.
32661 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
32664 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
32665 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
32667 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
32669 * reload.c (find_reloads): When reloading a nonoffsettable address,
32670 use RELOAD_OTHER for it and its address reloads.
32672 PR rtl-optimization/79910
32673 * combine.c (record_used_regs): New static function.
32674 (try_combine): Handle situations where there is an additional
32675 instruction between I2 and I3 which needs to have a LOG_LINK
32678 2017-03-17 Jeff Law <law@redhat.com>
32680 PR tree-optimization/71437
32681 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
32682 conditional in the hash table first.
32683 (vrp_dom_walker::before_dom_children): Extract condition from
32684 ASSERT_EXPR. Record condition, its inverion and any implied
32685 conditions as well.
32687 2017-03-17 Marek Polacek <polacek@redhat.com>
32688 Markus Trippelsdorf <markus@trippelsdorf.de>
32690 PR tree-optimization/80079
32691 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
32694 2017-03-17 Richard Biener <rguenther@suse.de>
32696 PR middle-end/80075
32697 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
32698 Properly verify the LHS before the RHS possibly claims to be
32700 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
32703 2017-03-17 Martin Jambor <mjambor@suse.cz>
32705 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
32706 (List of -O2 options): Likewise.
32707 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
32710 2017-03-17 Tom de Vries <tom@codesourcery.com>
32712 * gcov-dump.c (print_usage): Print bug_report_url.
32714 2017-03-17 Richard Biener <rguenther@suse.de>
32716 PR middle-end/80050
32717 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
32718 (parser::peek): Likewise.
32720 2017-03-17 Richard Biener <rguenther@suse.de>
32722 PR tree-optimization/80048
32723 * sese.c (free_sese_info): Properly release rename_map and
32724 copied_bb_map elements.
32726 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
32728 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
32729 Add linked-list forward and backlinks. Insert on
32730 construction, remove on destruction.
32731 (class pass_store_merging): Add m_stores_head field.
32732 (pass_store_merging::terminate_and_process_all_chains):
32733 Iterate over m_stores_head list.
32734 (pass_store_merging::terminate_all_aliasing_chains):
32736 (pass_store_merging::execute): Check for debug stmts first.
32737 Push new chains onto the m_stores_head stack.
32739 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
32742 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
32743 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
32744 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
32746 2017-03-16 Jeff Law <law@redhat.com>
32748 PR tree-optimization/71437
32749 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
32750 member function. Implementation moved into after_dom_children
32751 member function and into the threader's thread_outgoing_edges
32753 (dom_opt_dom_walker::after_dom_children): Simplify by moving
32754 some code into new thread_outgoing_edges.
32755 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
32756 definition. Simplify marker handling (do it here). Assume we always
32757 have the available expression and the const/copies tables.
32758 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
32760 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
32761 * tree-vrp.c (equiv_stack): No longer file scoped.
32762 (vrp_dom_walker): New class.
32763 (vrp_dom_walker::before_dom_children): New member function.
32764 (vrp_dom_walker::after_dom_children): Likewise.
32765 (identify_jump_threads): Setup domwalker. Use it rather than
32766 walking edges in a random order by hand. Simplify setup/finalization.
32767 (finalize_jump_threads): Remove.
32768 (vrp_finalize): Do not call identify_jump_threads here.
32769 (execute_vrp): Do it here instead and call thread_through_all_blocks
32772 PR tree-optimization/71437
32773 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
32775 (simplify_stmt_for_jump_threading): Add basic_block argument. All
32777 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
32778 (dom_opt_dom_walker::thread_across_edge): Remove
32779 handle_dominating_asserts argument. All callers changed.
32780 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
32781 changes. Remove calls to lhs_of_dominating_assert. Other
32782 uses of handle_dominating_asserts turn into unconditional code
32783 (simplify_control_stmt_condition_1): Likewise.
32784 (simplify_control_stmt_condition): Likewise.
32785 (thread_through_normal_block, thread_across_edge): Likewise.
32786 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
32787 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
32788 object if it is not an SSA_NAME.
32789 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
32790 before calling into the VRP specific simplifiers.
32791 (identify_jump_threads): Remove handle_dominating_asserts
32794 2017-03-16 Jakub Jelinek <jakub@redhat.com>
32797 * tree-diagnostic.c (default_tree_printer): No longer static.
32798 * tree-diagnostic.h (default_tree_printer): New prototype.
32800 2017-03-16 Tamar Christina <tamar.christina@arm.com>
32802 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
32803 Change ins into fmov.
32805 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32807 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
32808 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
32809 Use h_con constraint for operand 1.
32810 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
32811 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
32813 2017-03-15 Jeff Law <law@redhat.com>
32815 PR tree-optimization/71437
32816 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
32817 (record_temporary_equivalences): Use it.
32819 PR tree-optimization/71437
32820 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
32821 tree-ssa-scopedtables.
32822 (lookup_avail_expr, build_and_record_new_cond): Likewise.
32823 (record_conditions, record_cond, vuse_eq): Likewise.
32824 (record_edge_info): Adjust to API tweak of record_conditions.
32825 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
32826 (record_temporary_equivalences, optimize_stmt): Likewise.
32827 (eliminate_redundant_computations): Likewise.
32828 (record_equivalences_from_stmt): Likewise.
32829 * tree-ssa-scopedtables.c: Include options.h and params.h.
32830 (vuse_eq): New function, moved from tree-ssa-dom.c
32831 (build_and_record_new_cond): Likewise.
32832 (record_conditions): Likewise. Accept vector of conditions rather
32833 than edge_equivalence structure for first argument.
32834 for the first argument.
32835 (avail_exprs_stack::lookup_avail_expr): New member function, moved
32836 from tree-ssa-dom.c.
32837 (avail_exprs_stack::record_cond): Likewise.
32838 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
32839 from tree-ssa-dom.c.
32840 (avail_exprs_stack): Add new member functions lookup_avail_expr
32842 (record_conditions): Declare.
32844 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
32847 * lra-constraints.c (process_alt_operands): Increase reject for
32848 reloading an input/output operand.
32850 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
32853 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
32854 insns to convert from signed/unsigned char/short to IEEE 128-bit
32856 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
32858 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
32861 * config/i386/i386.c (ix86_vector_duplicate_value): Create
32862 subreg of inner mode for values already in registers.
32864 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
32866 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
32867 iteration reg is used after the loop.
32869 2017-03-14 Martin Sebor <msebor@redhat.com>
32871 PR tree-optimization/79800
32872 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
32873 precision in negative-positive range.
32874 (format_floating): Call non-const overload with adjusted precision.
32876 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
32879 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
32882 2017-03-14 Martin Sebor <msebor@redhat.com>
32884 PR middle-end/80020
32885 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
32886 * builtins.def (aligned_alloc): Use it.
32889 * Makefile.in (GTFILES): Add calls.c.
32890 * calls.c: Include "gt-calls.h".
32892 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
32894 PR rtl-optimization/79728
32895 * regs.h (struct target_regs): New field
32896 x_contains_allocatable_regs_of_mode.
32897 (contains_allocatable_regs_of_mode): New macro.
32898 * reginfo.c (init_reg_sets_1): Initialize it, and change
32899 contains_reg_of_mode so it includes global regs as well.
32900 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
32901 rather than contains_regs_of_mode.
32903 2017-03-14 Martin Liska <mliska@suse.cz>
32905 * doc/invoke.texi: Document options that can't be combined with
32906 -fcheck-pointer-bounds.
32908 2017-03-14 Martin Liska <mliska@suse.cz>
32910 PR middle-end/79831
32911 * doc/invoke.texi (-Wchkp): Document the option.
32913 2017-03-14 Martin Liska <mliska@suse.cz>
32915 * Makefile.in: Install gcov-dump.
32917 2017-03-14 Martin Liska <mliska@suse.cz>
32919 * multiple_target.c (expand_target_clones): Bail out for
32920 an invalid attribute.
32922 2017-03-14 Richard Biener <rguenther@suse.de>
32924 * alias.c (struct alias_set_entry): Pack properly.
32925 * cfgloop.h (struct loop): Likewise.
32926 * cse.c (struct set): Likewise.
32927 * ipa-utils.c (struct searchc_env): Likewise.
32928 * loop-invariant.c (struct invariant): Likewise.
32929 * lra-remat.c (struct cand): Likewise.
32930 * recog.c (struct change_t): Likewise.
32931 * rtl.h (struct address_info): Likewise.
32932 * symbol-summary.h (function_summary): Likewise.
32933 * tree-loop-distribution.c (struct partition): Likewise.
32934 * tree-object-size.c (struct object_size_info): Likewise.
32935 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
32936 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
32937 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
32938 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
32939 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
32940 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
32941 (struct _stmt_vec_info): Likewise.
32943 2017-03-14 Martin Liska <mliska@suse.cz>
32946 * multiple_target.c (create_dispatcher_calls): Check that
32947 a target can create a function dispatcher.
32949 2017-03-14 Martin Liska <mliska@suse.cz>
32952 * multiple_target.c (expand_target_clones): Drop local.local
32953 flag for default implementation.
32955 2017-03-14 Richard Biener <rguenther@suse.de>
32957 PR tree-optimization/80030
32958 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
32960 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
32962 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
32963 gcc_fallthrough() instead of __attribute__((fallthrough));
32965 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
32967 * doc/gcc.texi: Remove "up" link to (DIR).
32968 * doc/gccint.texi: Ditto.
32970 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
32972 * doc/install.texi (Specific) <avr>: Remove reference to
32975 2017-03-13 Jeff Law <law@redhat.com>
32977 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
32978 attribute rather than comments.
32980 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
32981 match_scratch operand is highest.
32983 2017-03-13 Martin Liska <mliska@suse.cz>
32985 PR middle-end/78339
32986 * ipa-pure-const.c (warn_function_noreturn): If the declarations
32987 is a CHKP clone, use original declaration.
32989 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32991 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
32992 (arc_conditional_register_usage): Use a different allocation order
32993 when optimizing for size.
32994 * common/config/arc/arc-common.c (arc_option_optimization_table):
32995 Section anchors default on when optimizing for size.
32997 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32999 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
33001 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33003 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
33004 * config/arc/arc.md (cpu_facility): Add cd variant.
33005 (*movqi_insn): Add code density variant.
33006 (*movhi_insn): Likewise.
33007 (*movqi_insn): Likewise.
33008 (*addsi3_mixed): Likewise.
33009 (subsi3_insn): Likewise.
33011 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33013 * config/arc/arc.md (movsi_cond_exec): Update constraint.
33015 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33017 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
33018 expressions with MINUS and UNARY ops.
33020 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33023 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
33025 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
33026 between vec_select and vector argument.
33027 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
33028 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
33029 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
33030 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
33031 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
33032 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
33034 2017-03-13 Richard Biener <rguenther@suse.de>
33037 * params.def (vect-max-peeling-for-alignment): Fix typo.
33039 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33041 * doc/install.texi (Specific) <mips-*-*>: Remove description of
33042 issue that only occurred with binutils below 2.18.
33044 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33046 * doc/install.texi (Specific) <cris-axis-elf>: No longer
33047 refer to binutils 2.11/2.12 minimum.
33049 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33051 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
33052 ftp.kernel.org and simplify binutils requirement.
33054 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
33056 * doc/invoke.texi (Warning Options): Fix spelling of link-time
33058 (Optimize Options): Ditto. Also remove redundancy.
33060 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33062 PR translation/79848
33063 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
33065 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
33066 to G_ to avoid double translation.
33068 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33070 PR translation/79923
33071 * auto-profile.c (get_combined_location): Convert leading
33072 character of diagnostics to lower case and remove trailing period.
33073 (read_profile): Likewise for various diagnostics.
33074 * config/arm/arm.c (arm_option_override): Remove trailing period
33075 from various diagnostics.
33076 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
33077 (msp430_expand_delay_cycles): Likewise.
33079 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33082 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
33083 full command-line argument, rather than just "str".
33084 (aarch64_validate_march): Likewise.
33085 (aarch64_validate_mtune): Likewise.
33087 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
33089 PR rtl-optimization/78911
33090 * lra-assigns.c (must_not_spill_p): New function.
33091 (spill_for): Use it.
33093 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33095 PR tree-optimization/79981
33096 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
33097 ATOMIC_COMPARE_EXCHANGE ifn result.
33098 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
33099 IFN_ATOMIC_COMPARE_EXCHANGE.
33101 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33104 * opts.c (parse_sanitizer_options): Add missing question mark to
33105 "did you mean" message.
33107 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33109 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
33111 (VMULEUH_UNS): Likewise.
33112 (VMULOUB_UNS): Likewise.
33113 (VMULOUH_UNS): Likewise.
33114 * config/rs6000/rs6000.c (builtin_function_type): Remove
33115 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
33117 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33120 * read-rtl-function.c (function_reader::read_rtx_operand): Update
33121 x with result of extra_parsing_for_operand_code_0.
33122 (function_reader::extra_parsing_for_operand_code_0): Convert
33123 return type from void to rtx, returning x. When reading
33124 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
33125 larger size containing struct block_symbol.
33127 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
33129 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
33130 -mfloat128-hardware without -m64.
33132 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
33135 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
33136 entries to the case statement that marks unsigned arguments to
33137 overloaded functions.
33139 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33141 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
33142 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
33144 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
33147 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
33148 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
33150 2017-03-10 Martin Liska <mliska@suse.cz>
33154 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
33155 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
33158 2017-03-10 Olivier Hainque <hainque@adacore.com>
33160 * tree-switch-conversion (array_value_type): Start by resetting
33161 candidate type to it's main variant.
33163 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33165 PR rtl-optimization/79909
33166 * combine.c (try_combine): Use simplify_replace_rtx on individual
33167 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
33168 of the whole CALL_INSN_FUNCTION_USAGE.
33170 PR tree-optimization/79972
33171 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
33172 get_range_info on SSA_NAMEs. Formatting fixes.
33174 2017-03-10 Richard Biener <rguenther@suse.de>
33175 Jakub Jelinek <jakub@redhat.com>
33177 PR tree-optimization/77975
33178 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
33179 edge to be constant.
33180 (get_val_for): For constant x return it. Formatting fix.
33181 (loop_niter_by_eval): Avoid pointless looping if the next iteration
33182 would use the same bases as the current one.
33184 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33186 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
33187 instead of vec_select for V1TImode.
33188 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
33190 (VSX_LE_128): Add V1TI to this mode iterator.
33191 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
33192 (*vsx_le_perm_store_<mode>): Likewise.
33193 (pre-reload splitter for VSX stores): Likewise.
33194 (post-reload splitter for VSX stores): Likewise.
33195 (*vsx_xxpermdi2_le_<mode>): Likewise.
33196 (*vsx_lxvd2x2_le_<mode>): Likewise.
33197 (*vsx_stxvd2x2_le_<mode>): Likewise.
33199 2017-03-09 Michael Eager <eager@eagercon.com>
33201 Correct failures with --enable-checking=yes,rtl.
33203 * config/microblaze/microblaze.c (microblaze_expand_shift):
33204 Replace GET_CODE test with CONST_INT_P and INTVAL test with
33205 test for const0_rtx.
33206 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
33207 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
33209 2017-03-09 Richard Biener <rguenther@suse.de>
33211 PR tree-optimization/79977
33212 * graphite-scop-detection.c (scop_detection::merge_sese):
33213 Handle the case of extra exits to blocks dominating the entry.
33215 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
33217 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
33220 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
33222 PR rtl-optimization/79949
33223 * lra-constraints.c (process_alt_operands): Check memory when
33224 trying to predict a cycle. Print about the overall increase.
33226 2017-03-09 Richard Biener <rguenther@suse.de>
33228 PR middle-end/79971
33229 * gimple-expr.c (useless_type_conversion_p): Preserve
33230 TYPE_SATURATING for fixed-point types.
33232 2017-03-09 Richard Biener <rguenther@suse.de>
33235 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
33236 alignment of BLKmode params.
33238 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33241 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
33242 (VALL_NO_V2Q): Likewise.
33244 * config/aarch64/aarch64-simd.md
33245 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
33247 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
33248 VALL_NO_V2Q mode iterator.
33249 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
33251 2017-03-09 Martin Liska <mliska@suse.cz>
33253 PR tree-optimization/79631
33254 * tree-chkp-opt.c (chkp_is_constant_addr): Call
33255 tree_int_cst_sign_bit just for INTEGER constants.
33257 2017-03-09 Martin Liska <mliska@suse.cz>
33261 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
33264 2017-03-09 Marek Polacek <polacek@redhat.com>
33267 * tree.c (inchash::add_expr): Handle TREE_VEC.
33269 2017-03-09 Martin Liska <mliska@suse.cz>
33272 (chkp_narrow_size_and_offset): New function.
33273 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
33274 (void chkp_parse_bit_field_ref): New function.
33275 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
33276 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
33278 2017-03-09 Martin Liska <mliska@suse.cz>
33281 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
33282 (chkp_find_bounds_1): Remove gcc_unreachable.
33284 2017-03-09 Jakub Jelinek <jakub@redhat.com>
33287 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
33288 BUILT_IN_SYNC*, determine the access type from the size suffix and
33289 always build a MEM_REF with that type. Handle forgotten
33290 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
33293 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
33294 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
33295 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
33296 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
33297 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
33298 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
33299 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
33300 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
33301 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
33302 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
33303 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
33304 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
33305 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
33306 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
33307 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
33308 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
33309 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
33310 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
33311 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
33312 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
33313 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
33314 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
33315 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
33316 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
33317 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
33318 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
33319 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
33320 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
33321 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
33322 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
33323 definitions outside of __OPTIMIZE__ guarded section.
33326 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
33327 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
33328 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
33329 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
33332 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33334 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
33335 ("vfenez<mode>"): Add missing constraints.
33337 2017-03-08 Martin Sebor <msebor@redhat.com>
33340 * config/nds32/nds32.c (nds32_option_override):
33341 Fix misspelled diagnostic.
33343 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33346 * gimplify.c (gimplify_omp_for): Replace index var in outer
33347 taskloop statement with an artificial variable and add
33348 OMP_CLAUSE_PRIVATE clause for it.
33350 2017-03-08 Richard Biener <rguenther@suse.de>
33352 PR tree-optimization/79955
33353 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
33354 for accesses that are completely outside of the variable.
33356 2017-03-08 Andrew Haley <aph@redhat.com>
33358 PR tree-optimization/79943
33359 * tree-ssa-loop-split.c (compute_new_first_bound): When
33360 calculating the new upper bound, (END-BEG) should be added, not
33363 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33365 * config/avr/avr.md (setmemhi): Make sure match_dup
33366 operand number comes before match_scratch.
33368 2017-03-08 Richard Biener <rguenther@suse.de>
33370 PR tree-optimization/79920
33371 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
33372 with ncopies == 1 to ...
33373 (vect_transform_slp_perm_load): ... here. Properly compute
33374 all element loads by iterating VF times over the group. Do
33375 not handle ncopies (computed in a broken way) in
33376 vect_create_mask_and_perm.
33378 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33381 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
33382 is a uniform vector, use uniform_vector_p return value instead of
33383 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
33385 2017-03-07 Marek Polacek <polacek@redhat.com>
33387 PR middle-end/79809
33388 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
33389 (alloca_call_type): Likewise.
33391 2017-03-07 Martin Liska <mliska@suse.cz>
33393 * gcov.c (process_args): Put comment to correct location.
33395 2017-03-07 Martin Liska <mliska@suse.cz>
33397 PR middle-end/68270
33398 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
33399 Use array_at_struct_end_p instead of DECL_CHAIN (field).
33400 (chkp_narrow_bounds_for_field): Likewise.
33401 (chkp_parse_array_and_component_ref): Pass one more argument to
33404 2017-03-07 Richard Biener <rguenther@suse.de>
33406 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
33409 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
33411 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
33412 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
33414 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33417 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
33418 to end of description.
33419 (PARAM_MAX_STORES_TO_MERGE): Likewise.
33421 2017-03-07 Jakub Jelinek <jakub@redhat.com>
33423 PR rtl-optimization/79901
33424 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
33426 (*avx512f_<code><mode>3<mask_name>): ... this.
33427 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
33428 iterator instead of VI8_AVX2_AVX512BW.
33430 PR rtl-optimization/79901
33431 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
33432 min/max expander, expand it using expand_vec_cond_expr.
33435 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
33438 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33441 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
33442 to void * for PCH reasons.
33443 * dwarf2out.c (output_loc_operands, output_die): Cast
33444 v.val_vec.array to unsigned char *.
33446 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
33449 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
33452 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
33454 PR rtl-optimization/79571
33455 * lra-constraints.c (process_alt_operands): Calculate static
33456 reject and subtract it from overall when only addresses will be
33459 2017-03-06 Julia Koval <julia.koval@intel.com>
33462 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
33463 incoming stack boundary to 128 for 64-bit targets.
33465 2017-03-06 Richard Biener <rguenther@suse.de>
33467 PR tree-optimization/79894
33468 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
33469 to NULL after folding it.
33471 2017-03-06 Richard Biener <rguenther@suse.de>
33473 PR tree-optimization/79824
33474 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
33475 check disabling peeling for gaps.
33477 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
33479 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
33480 attributes): Document gettimeofday.
33482 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33484 * config/s390/s390.c (s390_option_override_internal): Set
33485 PARAM_MIN_VECT_LOOP_BOUND
33487 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33489 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
33490 * config/s390/s390.md: Likewise.
33492 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33495 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
33496 (<avx2_avx512>_perm<mode>): Rename to ...
33497 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
33499 (<avx512>_perm<mode>_mask): Rename to ...
33500 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
33502 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
33503 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
33504 instead of VI8F_256_512.
33505 (avx512f_perm<mode>): New define_expand.
33506 (avx512f_perm<mode>_mask): Likewise.
33507 (avx512f_perm<mode>_1<mask_name>): New define_insn.
33508 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
33510 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33512 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
33513 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
33515 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
33517 2017-03-06 Martin Liska <mliska@suse.cz>
33520 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
33521 when having a SSA NAME w/o VAR_DECL assigned to it.
33523 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33525 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
33526 msa_dpsub_<su>_d): Fix MODE for vec_select.
33528 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33530 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
33532 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
33534 2017-03-06 Richard Biener <rguenther@suse.de>
33536 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
33537 * plugin.c (register_plugin_info): Likewise.
33538 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
33540 2017-03-05 Jakub Jelinek <jakub@redhat.com>
33542 * config/i386/sse.md (sse_storehps, sse_storelps,
33543 avx_<castmode><avxsizesuffix>_<castmode>,
33544 avx512f_<castmode><avxsizesuffix>_<castmode>,
33545 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
33546 in condition that at least one operand is not a MEM.
33548 2017-03-03 Jakub Jelinek <jakub@redhat.com>
33550 PR middle-end/79805
33551 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
33552 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
33554 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
33555 gimple_call_nothrow_p flag based on whether original builtin can throw.
33556 If it can, emit following stmts on the fallthrough edge.
33557 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
33558 don't create new bb if inserting just debug stmts on the edge, try to
33559 insert them on the fallthru bb or just reset debug stmts.
33561 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
33564 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
33565 restore recog_data (including the operand rtxes inside it) around
33566 the call to get_insn_template.
33568 2017-03-03 Martin Sebor <msebor@redhat.com>
33570 PR tree-optimization/79699
33571 * context.c (context::~context): Free MPFR caches to avoid
33572 a memory leak on program exit.
33574 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33576 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
33577 Use wide_int::ulow () instead of .elt (0).
33579 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
33581 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
33582 (*pushxf): Limit oF constraint to 32bit targets and add oC
33583 constraint for 64bit targets.
33584 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
33585 (*pushdf): Change rmF constraint to rmC.
33587 2017-03-03 Martin Liska <mliska@suse.cz>
33589 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
33590 Remove unused variable.
33592 2017-03-03 Jakub Jelinek <jakub@redhat.com>
33595 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
33596 is a memory operand, increase num_memory.
33597 (ix86_expand_args_builtin): Likewise.
33599 2017-03-03 Jan Hubicka <jh@suse.cz>
33602 * ipa-devirt.c (maybe_record_node): Properly handle
33603 __cxa_pure_virtual visibility.
33605 2017-03-03 Martin Liska <mliska@suse.cz>
33607 PR tree-optimization/79803
33608 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
33610 (pass_loop_prefetch::execute): Disabled optimization if an
33611 assumption about L1 cache size is not met.
33613 2017-03-03 Martin Liska <mliska@suse.cz>
33615 PR rtl-optimization/79574
33616 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
33617 (hash_scan_set): Likewise.
33618 (dump_hash_table): Likewise.
33619 (hoist_code): Likewise.
33621 2017-03-03 Richard Biener <rguenther@suse.de>
33623 * fixed-value.c (fixed_from_string): Restore use of elt (1)
33624 in place of uhigh ().
33625 (fixed_convert_from_real): Likewise.
33627 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
33630 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
33632 2017-03-03 Richard Biener <rguenther@suse.de>
33634 PR middle-end/79818
33635 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
33636 TYPE_OVERFLOW_UNDEFINED check.
33638 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33640 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
33642 (vector_ae_<mode>_p): Likewise.
33643 (vector_nez_<mode>_p): Likewise.
33644 (vector_ne_v2di_p): Likewise.
33645 (vector_ae_v2di_p): Likewise.
33646 (vector_ne_<mode>_p): Likewise.
33647 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
33649 (vsx_tsqrt<mode>2_fe): Likewise.
33651 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
33654 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
33656 2017-03-02 Jakub Jelinek <jakub@redhat.com>
33658 PR rtl-optimization/79780
33659 * cprop.c (one_cprop_pass): When second and further conditional trap
33660 in a single basic block is turned into an unconditional trap, turn it
33661 into a deleted note to avoid RTL verification failures.
33663 2017-03-02 Richard Biener <rguenther@suse.de>
33665 * fold-const.c (const_binop): Use ulow () instead of elt (0).
33667 2017-03-02 Richard Biener <rguenther@suse.de>
33669 PR tree-optimization/79345
33671 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
33672 param and abort the walk, returning -1 if it is hit.
33673 (walk_aliased_vdefs): Take a limit param and pass it on.
33674 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
33675 defaulting to 0 and return a signed int.
33676 * tree-ssa-uninit.c (struct check_defs_data): New struct.
33677 (check_defs): New helper.
33678 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
33679 about uninitialized memory.
33680 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
33681 bogus uninitialized warning.
33682 (fixed_convert_from_real): Likewise.
33684 2017-03-02 Bin Cheng <bin.cheng@arm.com>
33686 PR tree-optimization/66768
33687 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
33688 iv_use if base object can't be determined.
33690 2017-03-02 Jakub Jelinek <jakub@redhat.com>
33692 PR tree-optimization/79345
33693 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
33694 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
33695 (get_pattern_stats): Initialize it.
33696 * genemit.c (gen_expand): Verify match_scratch numbers come after
33697 match_operand/match_dup numbers.
33698 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
33699 match_scratch numbers.
33700 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
33702 * config/s390/s390.md (trunctdsd2): Likewise.
33704 2017-03-02 Richard Biener <rguenther@suse.de>
33706 * wide-int.h (wide_int_storage::operator=): Implement in terms
33709 2017-03-02 Richard Biener <rguenther@suse.de>
33711 PR tree-optimization/79777
33712 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
33713 the to insert expression to sth existing.
33715 2017-03-01 Martin Sebor <msebor@redhat.com>
33717 PR middle-end/79692
33718 * gimple-ssa-sprintf.c
33719 (directive::known_width_and_precision): New function.
33720 (format_integer): Use it.
33721 (get_mpfr_format_length): Consider the full range of precision
33722 when computing %g output with the # flag. Set the likely byte
33723 count to 3 rather than 1 when precision is indeterminate.
33724 (format_floating): Correct the lower bound of precision.
33726 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33728 * doc/invoke.texi: Document default code model for 64-bit Linux.
33730 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33733 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
33734 udiv rather than div since input pattern is unsigned.
33736 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
33738 * config/i386/i386.c (print_reg): Warn for values of
33739 unsupported size in integer register.
33741 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
33744 * config/rs6000/predicates.md (current_file_function_operand): Do
33745 not allow self calls to be local if the function is replaceable.
33747 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
33750 * config/rs6000/altivec.h (vec_ctz and others): Change the
33751 preprocessor macro that controls conditional compilation from
33752 _ARCH_PWR9 to __POWER9_VECTOR__.
33753 (vec_all_ne): Change parameterization of __altivec_scalar_pred
33754 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
33755 control (instead of _ARCH_PWR9 control) so that template
33756 definition uses power9-specific function.
33757 (vec_any_eq): Likewise.
33758 (vec_all_ne): Change macro definition to use a power9-specific
33759 expansion under #ifdef __POWER9_VECTOR__ control (instead of
33760 _ARCH_PWR9 control).
33761 (vec_any_eq) Likewise.
33762 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
33763 expansion for CMPNEF to remove support for xvcmpnesp instruction.
33764 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
33765 support for xvcmpnedp instruction.
33766 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
33767 macro expansion so that Power9 implementation of vec_all_ne does
33768 not use the AltiVec predicate framework.
33769 (VCMPNEH_P): Likewise.
33770 (VCMPNEW_P): Likewise.
33771 (VCMPNED_P): Likewise.
33772 (VCMPNEFP_P): Likewise.
33773 (VCMPNEDP_P): Likewise.
33774 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
33775 implementation of vec_any_eq to not use AltiVec predicate
33777 (VCMPAEH_P): Likewise.
33778 (VCMPAEW_P): Likewise.
33779 (VCMPAED_P): Likewise.
33780 (VCMPAEFP_P): Likewise.
33781 (VCMPAEDP_P): Likewise.
33782 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
33783 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
33784 not use the AltiVec predicate framework.
33785 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
33786 of vec_any_eq to not use AltiVec predicate framework.
33787 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
33788 support for predefined __POWER9_VECTOR__ macro to indicate that
33789 Power9 instruction selection is enabled.
33790 (altivec_overloaded_builtins): Remove extraneous
33791 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
33792 function argument types RS6000_BTI_bool_V16QI and
33793 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
33794 entry for overloaded function argument types RS6000_BTI_bool_V4SI
33795 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
33796 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
33797 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
33798 Power9 for implementations of vec_cmpne. Change the signature for
33799 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
33800 (representing vec_all_ne) to remove the previously described first
33801 argument of type RS6000_BTI_INTSI, as this was an artifact of
33802 reliance on the AltiVec predicate framework, which is no longer
33803 used in the implementation of these functions. Add
33804 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
33805 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
33806 since, unlike the AltiVec predicate framework implementation, we
33807 do not share function descriptors between vec_alle and vec_anyeq.
33808 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
33809 set of modes that receive special treatment even when
33810 TARGET_P9_VECTOR is true. The special treatment emits code that
33811 does not depend on Power9 instructions.
33812 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
33813 define_expand to not rely on AltiVec predicate framework.
33814 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33816 (vector_ne_v2di_p): Change this define_expand to not rely on
33817 AltiVec predicate framework.
33818 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
33820 (vector_ne_<mode>_p): Change this define_expand to not rely on
33821 AltiVec predicate framework.
33822 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33824 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
33825 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
33826 define_insn pattern.
33827 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
33828 define_insn pattern because the xvcmpne<VSs>. instruction is not
33830 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
33831 instruction is not supported.
33833 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33835 * config/nvptx/nvptx.c: Include intl.h.
33837 2017-03-01 Martin Jambor <mjambor@suse.cz>
33840 * ipa-prop.h (ipa_bits): Removed field known.
33841 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
33842 to pointers. Adjusted their comments to warn about their sharing.
33843 (ipcp_transformation_summary): Change bits to a vector of pointers.
33844 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
33845 (ipa_get_ipa_bits_for_value): Declare.
33846 * tree-vrp.h (value_range): Mark as GTY((for_user)).
33847 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
33848 (ipa_bits_hash_table): Likewise.
33849 (ipa_vr_ggc_hash_traits): Likewise.
33850 (ipa_vr_hash_table): Likewise.
33851 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
33852 being pointers and vr_known being removed.
33853 (ipa_set_jf_unknown): Likewise.
33854 (ipa_get_ipa_bits_for_value): New function.
33855 (ipa_set_jfunc_bits): Likewise.
33856 (ipa_get_value_range): New overloaded functions.
33857 (ipa_set_jfunc_vr): Likewise.
33858 (ipa_compute_jump_functions_for_edge): Use the above functions to
33859 construct bits and vr parts of jump functions.
33860 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
33861 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33863 (ipcp_grow_transformations_if_necessary): Also allocate
33864 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33866 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
33867 them. Fix too long lines.
33868 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
33869 vr_known being removed.
33870 (ipa_read_jump_function): Use new setter functions to construct bits
33871 and vr parts of jump functions or set them to NULL.
33872 (write_ipcp_transformation_info): Adjust for bits being pointers.
33873 (read_ipcp_transformation_info): Likewise.
33874 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
33876 Include gt-ipa-prop.h.
33877 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
33879 (ipcp_store_bits_results): Likewise.
33880 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
33881 Do not write to existing jump functions but use a temporary instead.
33883 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33886 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
33887 attempt to use its first operand as BIT_FIELD_REF base.
33889 2017-03-01 Richard Biener <rguenther@suse.de>
33891 PR middle-end/79721
33892 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
33893 interpolating formula in wrapping arithmetic.
33894 (chrec_apply): Convert chrec_evaluate return value to wanted type.
33896 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33898 PR tree-optimization/79734
33899 * tree-vect-generic.c (expand_vector_condition): Optimize
33900 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
33901 Handle VEC_COND_EXPR where comparison has different inner width from
33902 type's inner width.
33904 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
33906 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
33907 markup, and similar issues. Remove @opindex entries for things
33908 that aren't options. Add missing -mmpy-option entries.
33910 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33912 PR tree-optimization/79737
33913 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
33914 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
33915 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
33916 instead of byte_size. Formatting fix.
33917 (shift_bytes_in_array_right): Formatting fix.
33919 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
33922 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
33923 condition on optimize for the leaf function test.
33925 2017-02-28 Martin Liska <mliska@suse.cz>
33928 * read-rtl-function.c (function_reader::handle_unknown_directive):
33929 Bail out when one uses -flto.
33931 2017-02-28 Martin Liska <mliska@suse.cz>
33933 * common.opt: Replace space with tabular for options of <number>
33935 * config/i386/i386.opt: Show <number> value for
33936 -mlarge-data-threshold.
33937 * opts.c (print_filtered_help): Do not display number in hexadecimal
33940 2017-02-28 Martin Liska <mliska@suse.cz>
33942 * common.opt: Fix --help=option -Q for options which are of
33945 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
33947 * config/i386/i386.c (print_reg): Error out for values
33948 of 8-bit size in invalid integer register.
33950 2017-02-28 Martin Sebor <msebor@redhat.com>
33952 PR tree-optimization/79691
33953 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
33955 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33958 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
33959 gcc_unreachable with output_operand_lossage.
33961 2017-02-28 Richard Biener <rguenther@suse.de>
33963 PR tree-optimization/79740
33964 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
33966 (visit_nary_op): Insert the nary into the hashtable if we
33967 pattern-matched sth.
33968 * tree-ssa-pre.c (eliminate_insert): Robustify.
33970 2017-02-28 Richard Biener <rguenther@suse.de>
33972 PR middle-end/79731
33973 * fold-const.c (decode_field_reference): Reject out-of-bound
33976 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33978 * config/i386/i386.c: Include intl.h.
33979 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
33980 instead of just cond ? "..." : "...".
33981 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
33982 * coverage.c (read_counts_file): Likewise.
33983 * omp-offload.c: Include intl.h.
33984 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
33985 of just cond ? "..." : "...".
33986 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
33987 of just cond ? "..." : "...".
33989 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
33992 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
33994 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
33995 'tune for' CPU name.
33996 * config/arm/arm-cpu-data.h: Regenerated.
33998 2017-02-28 Richard Biener <rguenther@suse.de>
34000 PR tree-optimization/79732
34001 * tree-inline.c (expand_call_inline): Do not shadow var.
34003 2017-02-28 Richard Biener <rguenther@suse.de>
34005 PR tree-optimization/79723
34006 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
34007 address-space properly.
34009 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
34011 * doc/optinfo.texi (Optimization groups): Fix option used for
34013 * doc/invoke.texi (-fopt-info): Document "omp".
34014 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
34015 (OPTGROUP_ALL): Add OPTGROUP_OMP.
34016 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
34017 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
34018 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
34020 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
34022 * dumpfile.c (optgroup_options): Instead of "openmp", associate
34023 OPTGROUP_OMP with "omp".
34025 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
34028 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
34029 for arithmetic shift of unsigned V2DI.
34031 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
34033 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
34034 arc/linux.h headers.
34035 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
34036 (LINK_SPEC): Likewise.
34037 (ARC_TLS_EXTRA_START_SPEC): Likewise.
34038 (EXTRA_SPECS): Likewise.
34039 (STARTFILE_SPEC): Likewise.
34040 (ENDFILE_SPEC): Likewise.
34041 (LIB_SPEC): Likewise.
34042 (TARGET_SDATA_DEFAULT): Likewise.
34043 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
34044 (MULTILIB_DEFAULTS): Likewise.
34045 (DWARF2_UNWIND_INFO): Likewise.
34046 * config/arc/big.h: New file.
34047 * config/arc/elf.h: Likewise.
34048 * config/arc/linux.h: Likewise.
34049 * config/arc/t-uClibc: Remove.
34051 2017-02-27 Bin Cheng <bin.cheng@arm.com>
34053 PR tree-optimization/77536
34054 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
34055 (tree_transform_and_unroll_loop): Use above function to compute the
34056 estimated niter of unrolled loop and use it when scaling profile.
34057 Also use count info rather than frequency if it's non-zero.
34058 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
34059 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
34060 (vect_transform_loop): Call above function.
34062 2017-02-27 Richard Biener <rguenther@suse.de>
34064 PR tree-optimization/45397
34065 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
34066 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
34067 (visit_nary_op): Add pattern matching for CSEing sign-changed
34068 or truncated operations with wider ones.
34070 2017-02-27 Richard Biener <rguenther@suse.de>
34072 PR tree-optimization/79690
34073 * tree-vect-stmts.c (vectorizable_store): Use vector type
34074 built from the DR with address-space.
34076 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
34078 * doc/invoke.texi (Optimize Options): Refine the description
34079 of asan-use-after-return.
34081 2017-02-25 Alan Modra <amodra@gmail.com>
34083 PR rtl-optimization/79584
34084 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
34085 base, not ad->base_term, the reg within base. Remove assertion
34086 that ad->base == ad->base_term. Replace gen_int_mode using
34087 bogus mode with const0_rtx.
34089 2017-02-25 Jakub Jelinek <jakub@redhat.com>
34091 PR middle-end/79396
34092 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
34093 FMA_EXPR like tcc_binary or tcc_unary.
34095 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
34098 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
34100 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
34101 (output_loc_operands): Handle DW_OP_call_ref and
34102 DW_OP_GNU_variable_value.
34103 (struct variable_value_struct): New type.
34104 (struct variable_value_hasher): Likewise.
34105 (variable_value_hash): New variable.
34106 (string_types): Remove.
34107 (copy_loc_descr): New function.
34108 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
34109 (prepend_loc_descr_to_each): New function.
34110 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
34111 instead of add_loc_descr_to_each if the first argument is single
34112 location list and the second has multiple.
34113 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
34114 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
34115 when looking for variable value which doesn't have other location info.
34116 (loc_list_from_tree): Formatting fix.
34117 (gen_array_type_die): Simplify DW_AT_string_length handling.
34118 (adjust_string_types): Remove.
34119 (gen_subprogram_die): Don't call adjust_string_types nor test/set
34120 string_types. Call resolve_variable_values.
34121 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
34122 (resolve_addr_in_expr): Likewise. Add A argument.
34123 (copy_deref_exprloc): Remove deref argument. Adjust for the
34124 original expression being DW_OP_GNU_variable_value with optionally
34125 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
34126 optionally after it.
34127 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
34128 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
34129 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
34130 (variable_value_hasher::hash, variable_value_hasher::equal): New
34132 (resolve_variable_value_in_expr, resolve_variable_value,
34133 resolve_variable_values, note_variable_value_in_expr,
34134 note_variable_value): New functions.
34135 (dwarf2out_early_finish): Call note_variable_value on all toplevel
34138 2017-02-24 Jakub Jelinek <jakub@redhat.com>
34141 * opts.h (handle_generated_option): Add GENERATED_P argument.
34142 * opts-common.c (handle_option): Adjust function comment.
34143 (handle_generated_option): Add GENERATED_P argument, pass it to
34145 (control_warning_option): Pass false to handle_generated_option
34147 * opts.c (maybe_default_option): Pass true to handle_generated_option
34149 * optc-gen.awk: Likewise.
34151 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34153 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
34154 a REG, look at the REG it is a SUBREG of.
34155 (splitter for cmpeqsi_t): Ditto.
34157 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34159 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
34160 the special USEs with the pattern of the insn, not the insn itself.
34162 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
34165 * doc/invoke.texi: Document -mload-store-pairs.
34167 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34168 Sandra Loosemore <sandra@codesourcery.com>
34170 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
34171 argument isn't a CONST_INT.
34172 (nios2_alternate_compare_const): Assert op is a CONST_INT.
34173 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
34174 (nios2_validate_compare): Bypass alternate compare logic if *op2
34175 is not a CONST_INT.
34176 (ldstwm_operation_p): Return false if first_base is not a REG or
34177 if first_offset is not a CONST_INT.
34179 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34181 * config/cris/cris.md: Use correct operand in a define_peephole2.
34183 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34185 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
34187 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34189 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
34190 this_insn if it is an INSN or JUMP_INSN.
34191 (force_offsettable): Look at base, not at addr.
34192 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
34193 on things that aren't necessarily CONST_INTs.
34195 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
34197 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
34198 -mfpmath=sse is the default also for x86-32 targets with SSE2
34199 instruction set when @option{-ffast-math} is enabled
34201 2017-02-24 Jeff Law <law@redhat.com>
34203 PR rtl-optimizatoin/79286
34204 * ira.c (update_equiv_regs): Drop may_trap_p exception to
34207 2017-02-24 Richard Biener <rguenther@suse.de>
34209 PR tree-optimization/79389
34210 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
34213 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
34215 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
34216 function comment to reflect reality.
34217 (loop_exits_before_overflow): Fix typo in function description.
34219 2017-02-24 Richard Biener <rguenther@suse.de>
34221 PR tree-optimization/79389
34222 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
34223 properly that a threading opportunity exists. Detect conditional
34224 copy/constant propagation opportunities.
34226 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
34228 * config/visium/visium.md (type): Add trap.
34229 (b): New mode attribute.
34230 (*btst): Rename into...
34231 (*btst<mode>): ...this and adjust.
34232 (*cbranchsi4_btst_insn): Rename into...
34233 (*cbranch<mode>4_btst_insn): ...this and adjust.
34234 (trap): New define_insn.
34236 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34238 PR tree-optimization/79389
34239 * ifcvt.c (struct noce_if_info): Add rev_cond field.
34240 (noce_reversed_cond_code): New function.
34241 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
34242 reversed_comparison_code. Formatting fix.
34243 (noce_try_store_flag): Test rev_cond != NULL in addition to
34244 reversed_comparison_code.
34245 (noce_try_store_flag_constants): Likewise.
34246 (noce_try_store_flag_mask): Likewise.
34247 (noce_try_addcc): Use rev_cond if non-NULL instead of
34248 reversed_comparison_code.
34249 (noce_try_cmove_arith): Likewise. Formatting fixes.
34250 (noce_try_minmax, noce_try_abs): Clear rev_cond.
34251 (noce_find_if_block): Initialize rev_cond.
34252 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
34253 instead of false as last argument never attempt to reverse it
34256 2017-02-23 Bin Cheng <bin.cheng@arm.com>
34258 PR tree-optimization/79663
34259 * tree-predcom.c (combine_chains): Process refs in reverse order
34260 only for ZERO length chains, and add explaining comment.
34262 2017-02-23 Jeff Law <law@redhat.com>
34264 PR tree-optimization/79578
34265 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
34266 in call to operand_equal_p.
34268 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
34271 * config/i386/cpuid.h: Fix another undefined behavior.
34273 2017-02-23 Richard Biener <rguenther@suse.de>
34275 PR tree-optimization/79683
34276 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
34277 vector types for data-refs.
34279 2017-02-23 Martin Liska <mliska@suse.cz>
34281 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
34283 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34285 PR middle-end/79665
34286 * internal-fn.c (get_range_pos_neg): Moved to ...
34287 * tree.c (get_range_pos_neg): ... here. No longer static.
34288 * tree.h (get_range_pos_neg): New prototype.
34289 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
34290 are known to be in between 0 and signed maximum inclusive, try to
34291 expand both unsigned and signed divmod and use the cheaper one from
34294 2017-02-22 Jeff Law <law@redhat.com>
34296 PR tree-optimization/79578
34297 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
34298 to compare base operands.
34300 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
34303 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
34304 gpc_reg_operand instead of fpr_reg_operand.
34306 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
34308 * config/mips/mips.c (mips_return_in_memory): Force FP
34309 vector types to be returned in memory for o32 ABI.
34311 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34313 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
34314 instead of DW_TAG_member for static data member declarations and don't
34315 set no_linkage_name for static inline data members.
34316 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
34319 2017-02-22 Martin Liska <mliska@suse.cz>
34321 * doc/invoke.texi: Replace inequality signs with square brackets
34324 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
34327 * lra-constraints.c (simplify_operand_subreg): Handle
34328 WORD_REGISTER_OPERATIONS targets.
34330 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34333 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
34334 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
34335 elimination by swapping fld*.
34337 2017-02-22 Richard Biener <rguenther@suse.de>
34339 PR tree-optimization/79673
34340 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
34341 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
34342 irrelevant address-space qualifiers and avoiding a
34343 ADDR_SPACE_CONVERT_EXPR from fold_convert.
34345 2017-02-22 Richard Biener <rguenther@suse.de>
34347 PR tree-optimization/79666
34348 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
34349 to not symbolically negate if that may introduce undefined
34352 2017-02-22 Martin Liska <mliska@suse.cz>
34355 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
34356 * data-streamer-out.c (streamer_write_gcov_count_stream):
34358 * value-prof.c (stream_out_histogram_value): Make assert more
34359 precise based on type of counter.
34361 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
34364 * config/i386/i386.md (standard_x87sse_constant_load splitter):
34365 Use nonimmediate_operand instead of memory_operand for operand 1.
34366 (float-extend standard_x87sse_constant_load splitter): Ditto.
34368 2017-02-21 Jeff Law <law@redhat.com>
34370 PR tree-optimization/79621
34371 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
34372 blocks with edges to themselves.
34374 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34377 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
34378 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
34379 Use gimple_call_builtin_p.
34382 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
34383 on temporarily removed DEBUG_INSNs.
34385 PR tree-optimization/79649
34386 * tree-loop-distribution.c (classify_partition): Give up on
34387 non-generic address space loads/stores.
34389 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
34391 * doc/loop.texi (Loop manipulation): Remove nonexistent
34392 tree_ssa_loop_version from the documentation.
34393 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
34395 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34398 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
34399 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
34400 * config/rs6000/rs6000.c: Include except.h.
34401 (rs6000_expand_split_stack_prologue): Call
34402 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
34404 2017-02-21 Martin Jambor <mjambor@suse.cz>
34407 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
34408 have been analyzed.
34410 2017-02-21 Martin Jambor <mjambor@suse.cz>
34412 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
34413 for backward compatibility only.
34414 * doc/invoke.texi (Option Summary): Remove all references to
34415 -fipa-cp-alignment.
34417 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
34421 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34423 * lra-constraints.c (curr_insn_transform): Handle
34424 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34426 2017-02-21 Martin Liska <mliska@suse.cz>
34428 * config/i386/i386.opt: Replace -masm-dialect with -masm.
34430 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
34432 PR translation/79638
34433 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
34435 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
34438 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
34439 (arm_function_ok_for_sibcall): Return false for an indirect call by
34440 descriptor if all the argument registers are used.
34441 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
34442 alignment of the function.
34444 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34446 PR tree-optimization/61441
34447 * simplify-rtx.c (simplify_const_unary_operation): For
34448 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
34449 the sNaN unmodified.
34451 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34453 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34454 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34455 instead of SYSTEM_HEADER_DIR.
34457 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
34458 Martin Liška <mliska@suse.cz>
34460 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
34461 Fix typos and grammar, use active voice, and clarify.
34463 2017-02-20 Marek Polacek <polacek@redhat.com>
34465 PR middle-end/79537
34466 * gimplify.c (gimplify_expr): Handle unused *&&L;.
34469 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
34471 2017-02-20 Jakub Jelinek <jakub@redhat.com>
34474 * config/i386/i386.c (ix86_expand_builtin): Handle
34475 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
34476 ix86_builtins_isa[fcode].isa as a requirement of those
34477 flags and any other flag in the bitmask.
34478 (ix86_init_mmx_sse_builtins): Use 0 instead of
34479 ~OPTION_MASK_ISA_64BIT as mask.
34480 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
34481 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
34482 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
34483 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
34485 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34488 * lra-constraints.c (split_reg): Check requested split mode
34489 is supported by the register.
34491 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34493 * lra-constraints.c (simplify_operand_subreg): Remove early
34496 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34499 * lra-constraints.c (curr_insn_transform): Tighten condition
34500 for converting SUBREG reloads from OP_OUT to OP_INOUT.
34502 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34505 * lra-constraints.c (curr_insn_transform): Handle
34506 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34508 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
34511 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
34513 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
34515 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
34518 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
34521 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34523 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
34526 2017-02-18 Jakub Jelinek <jakub@redhat.com>
34528 * final.c (last_columnnum, override_columnnum): New variables.
34529 (final_start_function): Set last_columnnum, pass it to begin_prologue
34530 hook and pass 0 to dwarf2out_begin_prologue.
34531 (final_scan_insn): Update override_columnnum. Pass last_columnnum
34532 to source_line debug hook.
34533 (notice_source_line): Compute last_columnnum and for debug_column_info
34534 return true on column changes.
34535 * debug.h (struct gcc_debug_hooks): Add column argument to
34536 source_line and begin_prologue hooks.
34537 (debug_nothing_int_charstar_int_bool): Remove prototype.
34538 (debug_nothing_int_int_charstar,
34539 debug_nothing_int_int_charstar_int_bool): New prototypes.
34540 (dwarf2out_begin_prologue): Add column argument.
34541 * debug.c (do_nothing_debug_hooks): Adjust source_line and
34542 begin_prologue hooks.
34543 (debug_nothing_int_charstar_int_bool): Remove.
34544 (debug_nothing_int_int_charstar,
34545 debug_nothing_int_int_charstar_int_bool): New functions.
34546 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
34547 through to dwarf2out_source_line.
34548 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
34549 (dwarf2out_source_line): Add column argument, emit it if requested.
34550 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
34552 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
34553 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
34554 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
34555 through to dwarf2out_begin_prologue.
34556 (vmsdbgout_source_line): Add column argument, pass it through to
34557 dwarf2out_source_line.
34558 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
34559 dbxout_source_line caller.
34560 (dbxout_source_line): Add column argument.
34562 * common.opt (gno-column-info, gcolumn-info): New options.
34563 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
34564 (check_die): Also test for multiple DW_AT_decl_column attributes.
34565 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
34566 DW_AT_decl_column if requested.
34567 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
34569 (gen_variable_die): Likewise.
34570 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
34571 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
34574 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
34575 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
34576 (ix86_handle_option): Handle OPT_m3dnowa.
34577 * doc/invoke.texi (-m3dnowa): Document.
34578 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
34579 -m3dnowa instead of -m3dnow -march=athlon.
34582 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
34583 instead of gcc_assert for K, r and R code checks. Formatting fixes.
34585 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34588 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
34589 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
34590 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
34591 generator for vsx_xxpermdi_<mode>_be.
34592 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
34593 force big-endian semantics.
34594 (vsx_xxpermdi_<mode>_be): New define_expand with same
34595 implementation as previous version of vsx_xxpermdi_<mode>.
34597 2017-02-17 Jakub Jelinek <jakub@redhat.com>
34599 PR tree-optimization/79327
34600 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
34601 variable, its initialization and use.
34603 2017-02-17 Julia Koval <julia.koval@intel.com>
34605 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
34606 (OPTION_MASK_ISA_PKU_UNSET): New.
34607 (ix86_handle_option): Handle -mrdpid.
34608 * config/i386/cpuid.h (bit_RDPID): New.
34609 * config/i386/driver-i386.c (host_detect_local_cpu):
34610 Detect RDPID feature.
34611 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
34612 * config/i386/i386-c.c (ix86_target_macros_internal):
34614 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
34615 (ix86_valid_target_attribute_inner_p): Add "rdpid".
34616 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
34617 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
34618 * config/i386/i386.md (define_insn "rdpid"): New.
34619 * config/i386/i386.opt Add -mrdpid.
34620 * config/i386/immintrin.h (_rdpid_u32): New.
34622 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
34624 PR rtl-optimization/79541
34625 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
34626 instead of transforming it into USE.
34628 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
34630 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
34631 If HONOR_SNANS (SFmode) force the input to a register.
34632 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
34633 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
34634 an frsp or similar insn.
34636 2017-02-17 Martin Liska <mliska@suse.cz>
34638 PR rtl-optimization/79577
34639 * params.def (selsched-max-sched-times): Increase minimum to 1.
34641 2017-02-17 Martin Liska <mliska@suse.cz>
34643 PR rtl-optimization/79574
34644 * gcse.c (want_to_gcse_p): Prevent integer overflow.
34646 2017-02-17 Martin Liska <mliska@suse.cz>
34648 PR tree-optimization/79529
34649 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
34650 ssa_defined_default_def_p to handle cases which are implicitly
34652 * tree-ssa.c (ssa_defined_default_def_p): New function.
34653 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
34654 which are implicitly defined.
34655 * tree-ssa.h (ssa_defined_default_def_p): Declare.
34657 2017-02-17 Richard Biener <rguenther@suse.de>
34659 PR middle-end/79576
34660 * params.def (max-ssa-name-query-depth): Limit to 10.
34662 2017-02-17 Richard Biener <rguenther@suse.de>
34664 PR tree-optimization/79552
34665 * tree-ssa-structalias.c (visit_loadstore): Properly verify
34668 2017-02-17 Richard Biener <rguenther@suse.de>
34671 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
34673 2017-02-17 Marek Polacek <polacek@redhat.com>
34675 PR middle-end/79536
34676 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
34677 (fold_negate_expr): New wrapper.
34679 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
34681 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
34682 Correct terminology and de-emphasize pre-standard behavior.
34684 2017-02-16 Alan Modra <amodra@gmail.com>
34686 PR rtl-optimization/79286
34687 * ira.c (def_dominates_uses): New function.
34688 (update_equiv_regs): Don't create an equivalence for insns that
34689 may trap where the register def does not dominate the use.
34691 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
34693 PR rtl-optimization/78127
34694 * lra.c (lra): Call lra_eliminate before finish the loop after
34697 2017-02-16 Richard Biener <rguenther@suse.de>
34699 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
34701 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
34702 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
34703 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
34704 (isl_val_int_from_wi): New function.
34705 (extract_affine_gmp): Rename to ...
34706 (extract_affine_wi): ... this, take a widest_int.
34707 (extract_affine_int): Just wrap extract_affine_wi.
34708 (add_param_constraints): Use isl_val_int_from_wi.
34709 (add_loop_constraints): Likewise, and extract_affine_wi.
34711 2017-02-15 Jeff Law <law@redhat.com>
34713 PR middle-end/79521
34714 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
34715 ira_init_register_move_cost_if_necessary.
34717 2017-02-15 Martin Sebor <msebor@redhat.com>
34719 PR middle-end/32003
34720 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
34721 removed in a prior commit.
34723 2017-02-15 Bin Cheng <bin.cheng@arm.com>
34725 PR tree-optimization/79347
34726 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
34727 counters during peeling.
34729 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
34731 * Makefile.in (site.exp): Remove "set ISLVER".
34733 2017-02-15 Jakub Jelinek <jakub@redhat.com>
34736 * real.c (real_from_integer): Call real_convert even for decimal.
34738 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
34741 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
34743 2017-02-14 Andrew Pinski <apinski@cavium.com>
34745 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
34746 cores and change the partno/implementer to be correct.
34747 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
34748 the 'B" as the implementer.
34749 * config/aarch64/aarch64-tune.md: Regenerate.
34751 2017-02-14 Carl Love <cel@us.ibm.com>
34753 * config/rs6000/rs6000.c: Add case statement entry to make the
34754 xvcvuxdsp built-in argument unsigned.
34755 * config/rs6000/vsx.md: Fix the source and return operand types so they
34756 match the instruction definitions from the ISA document. Fix typo
34757 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
34760 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
34763 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
34764 member early_clobber_alts.
34765 * lra-lives.c (reg_early_clobber_p): New.
34766 (process_bb_lives): Use it.
34767 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
34768 (debug_operand_data): Initialize early_clobber_alts.
34769 (setup_operand_alternative): Set up early_clobber_alts.
34770 (collect_non_operand_hard_regs): Ditto. Pass early clobber
34771 alternatives to new_insn_reg.
34772 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
34774 (lra_update_insn_regno_info): Pass the new arg.
34776 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34778 PR middle-end/79505
34779 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
34780 (new_oacc_loop_raw): Don't clear already cleared fields.
34783 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
34784 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
34785 _mm512_prefetch_i64gather_ps): New inline functions and macros.
34787 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
34790 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
34792 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
34795 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
34796 the extra instruction to the right place to store 128-bit constant
34799 2017-02-14 Martin Sebor <msebor@redhat.com>
34801 PR middle-end/79448
34802 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
34803 warning for strings of unknown length.
34805 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
34807 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
34809 2017-02-14 Jeff Law <law@redhat.com>
34812 * ira-costs.c (scan_one_insn): Initialize register move costs
34813 for pseudos seen in USE/CLOBBER insns.
34815 PR tree-optimization/79095
34816 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
34817 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
34818 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
34819 if the operands are known to be not equal, then the resulting range
34821 (intersect_ranges): If the new range is ~[0,0] and the old range is
34822 wide, then prefer ~[0,0].
34823 * tree-vrp.c (overflow_comparison_p_1): New function.
34824 (overflow_comparison_p): New function.
34825 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
34826 if NAME is used in an overflow test.
34827 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
34828 overflow check that can be expressed as an equality test, then adjust
34829 ops to be that equality test.
34831 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34833 * config/s390/s390-builtin-types.def: Remove flags argument.
34834 * config/s390/s390.c (s390_init_builtins): Likewise.
34836 2017-02-14 Martin Liska <mliska@suse.cz>
34838 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
34839 vector. Fix trailing white spaces.
34841 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
34843 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
34846 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34848 PR rtl-optimization/68664
34849 * config/arm/arm.c (arm_sched_can_speculate_insn):
34850 New function. Declare prototype.
34851 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34853 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34855 PR rtl-optimization/68664
34856 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
34858 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34860 2017-02-14 Amit Pawar <amit.pawar@amd.com>
34862 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
34863 max skip bytes for function, loop and jump.
34865 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34867 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
34868 ABS_EXPR for gimple dump.
34870 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34873 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
34875 PR tree-optimization/79408
34876 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
34877 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
34878 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
34879 also if rhs1 is INTEGER_CST.
34881 2017-02-14 Richard Biener <rguenther@suse.de>
34883 PR middle-end/79432
34884 * tree-into-ssa.c (insert_phi_nodes): When the function can
34885 have abnormal edges rewrite SSA names with broken use-def
34886 dominance out of SSA and register them for PHI insertion.
34888 2017-02-13 Martin Sebor <msebor@redhat.com>
34890 PR middle-end/79496
34891 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
34892 clearing info.nowrite flag when snprintf size argument is a range.
34894 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34896 * cprop.c (cprop_jump): Add missing space in string literal.
34897 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
34898 (get_constraint_for_component_ref): Likewise.
34899 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
34900 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
34901 * lra-constraints.c (process_alt_operands): Likewise.
34902 * ipa-inline.c (inline_small_functions): Likewise.
34903 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
34904 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
34905 * trans-mem.c (diagnose_tm_1_op): Likewise.
34906 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
34907 (grid_parallel_clauses_gridifiable): Likewise.
34909 * config/nvptx/mkoffload.c (process): Add space in between
34912 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
34913 "MOD4_SSE_REGS" and "ALL_REGS".
34915 * spellcheck.c (test_data): Add , in between "foo" and "food".
34917 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34920 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
34921 boundary crossing check and subsequent code generation agree.
34923 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34925 * config/aarch64/aarch64.c (has_memory_op): Delete.
34926 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
34929 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34931 PR rtl-optimization/79388
34932 PR rtl-optimization/79450
34933 * combine.c (distribute_notes): When removing TEM_INSN for which
34934 corresponding dest has last value recorded, invalidate that last
34937 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34939 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
34940 of explicit '@'. Add missing assembly comment marker on branch costs
34943 2017-02-13 Nathan Sidwell <nathan@acm.org>
34945 * gengtype-lex.l (<in_struct>): Add '/'.
34947 2017-02-13 Martin Liska <mliska@suse.cz>
34950 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
34952 2017-02-13 Richard Biener <rguenther@suse.de>
34954 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
34956 * configure: Re-generate.
34957 * config.in: Likewise.
34958 * graphite-dependences.c: Simplify as if
34959 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
34960 * graphite-isl-ast-to-gimple.c: Likewise.
34961 * graphite-optimize-isl.c: Likewise.
34962 * graphite-poly.c: Likewise.
34963 * graphite-sese-to-poly.c: Likewise.
34964 * graphite.h: Likewise.
34965 * toplev.c: Include isl/version.h and use isl_version () for
34966 printing the ISL version.
34967 * doc/install.texi: Update ISL requirement.
34969 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
34971 * doc/standards.texi (Standards): Update reference to
34974 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
34976 * doc/extend.texi (Named Address Spaces): sourceware.org now
34978 * doc/install.texi (Binaries): Ditto.
34981 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
34983 * doc/cpp.texi: Replace "stringify"/"stringification" with C
34984 standard terminology "stringize"/"stringizing" throughout.
34985 * doc/cppinternals.texi: Likewise.
34987 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
34989 * doc/extend.texi: Fix some spelling mistakes and typos.
34990 * doc/invoke.texi: Likewise.
34992 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34995 * params.def (inline-min-speedup) Change from 10 to 8.
34997 2017-02-11 Jakub Jelinek <jakub@redhat.com>
34999 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
35002 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
35005 * ipa-inline-analysis.c (get_minimal_bb): New function.
35006 (record_modified): Use it.
35007 (remap_edge_change_prob): Handle also ancestor functions.
35009 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
35011 * doc/contrib.texi (Contributors): Remove broken link into
35012 the Mauve CVS repository.
35014 2017-02-11 Jakub Jelinek <jakub@redhat.com>
35016 PR middle-end/79454
35017 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
35018 result computation whenever lhs doesn't have vector mode, not
35019 just when it has BLKmode.
35021 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
35023 * doc/makefile.texi (profiledbootstrap): Refer to the
35024 installation instructions only in textual form.
35026 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35029 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
35031 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
35033 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
35034 (Specific): Update mingw-w64 reference.
35036 (Specific): Remove broken link to Renesas RX processor.
35038 2017-02-10 Richard Biener <rguenther@suse.de>
35040 * toplev.c (process_options): Do not mention obsolete graphite
35041 options when printing sorry message about missing graphite support.
35042 Mention -floop-nest-optimize.
35044 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
35046 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
35047 (vtst_p16): Likewise.
35048 (vtstq_p8): Likewise.
35049 (vtstq_p16): Likewise.
35051 (vtstq_p64): Likewise.
35052 * config/arm/arm_neon.h (vgetq_lane_p64): New.
35053 (vset_lane_p64): New.
35054 (vsetq_lane_p64): New.
35056 2017-02-10 Jakub Jelinek <jakub@redhat.com>
35058 PR tree-optimization/79411
35059 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
35060 stmt operands are SSA_NAMEs used in abnormal phis.
35061 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
35064 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35067 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
35070 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35072 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
35074 2017-02-09 Jakub Jelinek <jakub@redhat.com>
35076 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
35080 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
35081 not arbitrary TREE_CONSTANT.
35084 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
35085 "omp declare target link" attribute unless is_global_var.
35086 * omp-offload.c (find_link_var_op): Likewise.
35088 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
35089 Chung-Lin Tang <cltang@codesourcery.com>
35091 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
35093 (gimplify_adjust_omp_clauses): Don't delete TILE.
35094 (gimplify_omp_for): Deal with TILE.
35095 * internal-fn.c (expand_GOACC_TILE): New function.
35096 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
35098 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
35099 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
35101 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
35102 avoid DIV for outermost collapse var.
35103 (expand_oacc_for): Insert tile element loop as needed. Adjust.
35104 Remove out of date comments, fix whitespace.
35105 * omp-general.c (omp_extract_for_data): Deal with tiling.
35106 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
35107 adjust OLF_DIM_BASE value.
35108 (struct omp_for_data): Add tiling field.
35109 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
35110 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
35111 for auto loops. Remove default auto determining, moved to
35112 oacc_loop_fixed_partitions.
35113 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
35114 stmts, add e_mask field.
35115 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
35116 (oacc_thread_numbers): Use oacc_dim_call.
35117 (oacc_xform_tile): New.
35118 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
35119 (finish_oacc_loop): Adjust for ifns vector.
35120 (oacc_loop_discover_walk): Append loop abstraction sites to list,
35121 add case for GOACC_TILE fns.
35122 (oacc_loop_xform_loop): Delete.
35123 (oacc_loop_process): Iterate over call list directly, and add
35124 handling for GOACC_TILE fns.
35125 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
35127 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
35128 vector partitioning to outer loops. Assign 2 partitions to loops
35129 when available. Add TILE handling.
35130 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
35131 (execite_oacc_device_lower): Process GOACC_TILE fns,
35132 ignore unknown specs.
35133 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
35134 * tree.c (omp_clause_num_ops): Adjust TILE ops.
35135 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
35137 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
35139 * configure.ac (ACX_BUGURL): Update.
35140 * configure: Regenerate.
35142 2017-02-09 Richard Biener <rguenther@suse.de>
35144 PR tree-optimization/69823
35145 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
35146 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
35148 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35150 * config/arc/arc-c.def: Add __NPS400__ definition.
35151 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
35152 (TARGET_NPS400): Define.
35154 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35156 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
35158 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
35159 pointer, arch_info.
35160 (arc_cpu_types): Fill the arch_info field with a pointer into the
35161 arc_arch_types table.
35162 (arc_selected_cpu): Declare.
35163 * config/arc/arc.c (arc_selected_cpu): Make global.
35164 (arc_selected_arch): Delete.
35165 (arc_base_cpu): Delete.
35166 (arc_override_options): Remove references to deleted variables,
35167 update access to arch information.
35168 (ARC_OPT): Update access to arch information.
35169 (ARC_OPTX): Likewise.
35170 * config/arc/arc.h (arc_base_cpu): Remove declaration.
35171 (TARGET_ARC600): Update access to arch information.
35172 (TARGET_ARC601): Likewise.
35173 (TARGET_ARC700): Likewise.
35174 (TARGET_EM): Likewise.
35175 (TARGET_HS): Likewise.
35176 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
35179 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
35182 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
35183 condition/operands for integer GE/LE/GEU/LEU operations.
35185 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
35187 PR translation/79397
35188 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
35191 2017-02-08 Martin Jambor <mjambor@suse.cz>
35194 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
35195 whether allocation happened.
35196 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
35197 nothing was allocated.
35199 2017-02-08 Jakub Jelinek <jakub@redhat.com>
35201 PR tree-optimization/79408
35202 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
35203 constant, but SSA_NAME with a known integer range, use the minimum
35204 of that range instead of op1 to determine if modulo can be replaced
35205 with its first operand.
35207 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35209 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
35211 2017-02-08 Richard Biener <rguenther@suse.de>
35213 PR tree-optimization/71824
35214 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35215 Check all loops contained in the merged region.
35217 2017-02-07 Andrew Pinski <apinski@cavium.com>
35219 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
35221 2017-02-07 Andrew Pinski <apinski@cavium.com>
35223 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
35224 (thunderxt88): Likewise.
35225 (thunderxt81): Disable LSE and change v8.1 to v8.
35226 (thunderxt83): Likewise.
35228 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35229 Richard Biener <rguenther@suse.de>
35231 PR middle-end/79399
35232 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
35233 type from int to size_t.
35234 * ira-costs.c (struct_costs_size): Change type from int to size_t.
35236 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35238 PR rtl-optimization/79386
35239 * cprop.c (bypass_conditional_jumps): Initialize
35240 bypass_last_basic_block already before splitting bbs after
35241 unconditional traps...
35242 (bypass_conditional_jumps): ... rather than here.
35245 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
35246 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
35247 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
35248 fix -masm=intel patterns.
35250 2017-02-07 Richard Biener <rguenther@suse.de>
35252 PR tree-optimization/79256
35253 PR middle-end/79278
35254 * builtins.c (get_object_alignment_2): Use min_align_of_type
35255 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
35256 and ADJUST_FIELD_ALIGN.
35258 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
35260 * doc/tm.texi: Regenerate.
35261 * stor-layout.c (layout_decl): Adjust.
35262 (update_alignment_for_field): Likewise.
35263 (place_field): Likewise.
35264 (min_align_of_type): Likewise.
35265 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
35266 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
35267 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
35268 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
35269 * config/frv/frv.c (frv_adjust_field_align): Likewise.
35270 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
35271 * config/i386/i386.c (x86_field_alignment): Likewise.
35272 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
35273 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
35274 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
35275 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
35276 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
35277 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
35281 2017-01-30 Richard Biener <rguenther@suse.de>
35283 PR tree-optimization/79256
35284 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35285 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35288 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
35290 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
35291 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
35292 builtins to SImode and emit a zero-extend, if necessary.
35294 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35296 * docs/invoke.texi (RISC-V Options): Alphabetize.
35298 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35300 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
35303 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35305 * config/riscv/riscv.c: New file.
35306 * common/config/riscv/riscv-common.c: Likewise.
35307 * config.gcc: Likewise.
35308 * config/riscv/constraints.md: Likewise.
35309 * config/riscv/elf.h: Likewise.
35310 * config/riscv/generic.md: Likewise.
35311 * config/riscv/linux.h: Likewise.
35312 * config/riscv/multilib-generator: Likewise.
35313 * config/riscv/peephole.md: Likewise.
35314 * config/riscv/pic.md: Likewise.
35315 * config/riscv/predicates.md: Likewise.
35316 * config/riscv/riscv-builtins.c: Likewise.
35317 * config/riscv/riscv-c.c: Likewise.
35318 * config/riscv/riscv-ftypes.def: Likewise.
35319 * config/riscv/riscv-modes.def: Likewise.
35320 * config/riscv/riscv-opts.h: Likewise.
35321 * config/riscv/riscv-protos.h: Likewise.
35322 * config/riscv/riscv.h: Likewise.
35323 * config/riscv/riscv.md: Likewise.
35324 * config/riscv/riscv.opt: Likewise.
35325 * config/riscv/sync.md: Likewise.
35326 * config/riscv/t-elf-multilib: Likewise.
35327 * config/riscv/t-linux: Likewise.
35328 * config/riscv/t-linux-multilib: Likewise.
35329 * config/riscv/t-riscv: Likewise.
35330 * configure.ac: Likewise.
35331 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
35332 Waterman as RISC-V maintainers.
35333 * doc/install.texi: Add RISC-V entries.
35334 * doc/invoke.texi: Add RISC-V options section.
35335 * doc/md.texi: Add RISC-V constraints section.
35336 * configure: Regenerated.
35338 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
35341 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
35342 false values to be constant vectors with all 0 or all 1 bits set.
35343 (vcondu<mode><mode>): Likewise.
35344 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
35346 (fpmask_comparison_operator): Update comment.
35347 (vecint_comparison_operator): New predicate.
35348 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
35349 vector conditionals when the true and false values are constant
35350 vectors with all 0 bits or all 1 bits set.
35352 2017-02-06 Martin Sebor <msebor@redhat.com>
35354 PR tree-optimization/79376
35355 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
35357 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
35359 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
35360 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
35361 to simplify split condition.
35363 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35365 * omp-expand.c (oxpand_omp_atomic_fetch_op,
35366 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
35369 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
35371 PR rtl-optimization/68664
35372 * target.def (can_speculate_insn): New hook.
35373 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
35374 * doc/tm.texi: Regenerate.
35375 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
35376 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
35377 (rs6000_sched_can_speculate_insn): New function.
35379 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35381 PR tree-optimization/79284
35382 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
35383 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
35384 vectorizable_mask_load_store, vectorizable_operation,
35385 vect_is_simple_cond, get_same_sized_vectype): Use it instead
35386 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
35387 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
35388 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
35389 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
35390 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
35391 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
35392 is_gimple_assign (stmt). Replace another such test with
35393 is_gimple_assign (stmt).
35395 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35398 * config/avr/avr.c (rtl-iter.h): Include it.
35399 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
35400 (avr_legitimate_combined_insn): ...and implementation.
35402 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
35404 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
35405 * config/s390/s390.c (s390_const_operand_ok)
35406 (s390_canonicalize_comparison, s390_extract_part)
35407 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
35408 (s390_contiguous_bitmask_p, s390_rtx_costs)
35409 (legitimize_pic_address): Likewise.
35410 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
35411 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
35412 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
35413 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
35414 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
35416 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35418 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
35419 REGNO($0) == REGNO($1).
35421 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35423 * config/s390/linux.h(SIZE_TYPE): Add comment.
35425 2017-02-06 Julian Brown <julian@codesourcery.com>
35426 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35427 Virendra Pathak <virendra.pathak@broadcom.com>
35429 * config/aarch64/aarch64-cores.def: Change the scheduler
35431 * config/aarch64/aarch64.md: Include thunderx2t99.md.
35432 * config/aarch64/thunderx2t99.md: New file.
35434 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35436 * doc/standards.texi (Go Language): Update link to language
35439 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35441 * tree-eh.c (lower_resx): Sanitize profile.
35442 (cleanup_empty_eh_move_lp): Likewise.
35444 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35447 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
35449 * cfgloopmanip.h (loop_version): Update prototype.
35450 * modulo-sched.c (sms_schedule): Update call of loop_version.
35451 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
35452 * tree-parloops.c (gen_parallel_loop): Likewise.
35453 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
35454 * tree-ssa-loop-split.c (split_loop): Likewise.
35455 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
35456 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
35458 2017-02-05 Martin Liska <mliska@suse.cz>
35461 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
35463 (print_operand_address): Initialize a struct to zero.
35465 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35467 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
35468 garbage collector only in textual form.
35470 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35472 * doc/extend.texi (x86 specific memory model extensions for
35473 transactional memory): Simplify a phrase.
35475 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
35478 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
35479 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
35480 (atomic_storedi_1): Likewise.
35482 2017-02-04 Jakub Jelinek <jakub@redhat.com>
35484 PR tree-optimization/79338
35485 * tree-parloops.c (gather_scalar_reductions): Don't call
35486 vect_analyze_loop_form for loop->inner before destroying loop's
35489 2017-02-03 Martin Sebor <msebor@redhat.com>
35491 PR tree-optimization/79327
35492 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
35493 when precision has resulted in leading zeros.
35494 (format_integer): Adjust the likely counter to assume an unknown
35495 argument that may be zero is non-zero.
35497 2017-02-03 Jason Merrill <jason@redhat.com>
35500 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
35501 avoid copying non-taken branch.
35503 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35505 PR tree-optimization/79340
35506 * tree-vect-loop.c (vectorizable_reduction): Release
35507 vec_defs elements after safe_splicing them into other vectors.
35510 PR tree-optimization/79327
35511 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
35512 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
35514 (format_integer): Use wide_int_to_tree instead of build_int_cst
35515 + to_?hwi. If argmin is NULL, just set argmin and argmax to
35516 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
35517 of shortest and longest sequence.
35519 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
35521 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
35522 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
35524 2017-02-03 Walter Lee <walt@tilera.com>
35527 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
35528 after initial stackframe link reg save.
35529 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
35531 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35534 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
35535 wu for stxssp alternative.
35537 2017-02-03 Martin Sebor <msebor@redhat.com>
35539 PR tree-optimization/79352
35540 * gimple-fold.c (get_range_strlen): Add argument.
35541 (get_range_strlen): Change return type to bool.
35542 (get_maxval_strlen): Pass in a dummy argument.
35543 * gimple-fold.h (get_range_strlen): Change return type to bool.
35544 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
35545 * tree.h (array_at_struct_end_p): Add argument.
35546 * tree.c (array_at_struct_end_p): Handle it.
35548 2017-02-03 Martin Liska <mliska@suse.cz>
35551 * multiple_target.c (create_dispatcher_calls): Redirect edge
35552 from a caller of a dispatcher.
35553 (expand_target_clones): Make the clones local.
35554 (ipa_target_clone): Do both target clones and resolvers.
35555 (ipa_dispatcher_calls): Remove the pass.
35556 (pass_dispatcher_calls::gate): Likewise.
35557 (make_pass_dispatcher_calls): Likewise.
35558 * passes.def (pass_target_clone): Put as very first IPA early
35561 2017-02-03 Martin Liska <mliska@suse.cz>
35563 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
35564 in case of a function with ifunc attribute.
35566 2017-02-03 Martin Liska <mliska@suse.cz>
35568 * cgraph.c (cgraph_node::dump): Dump function version info.
35569 * symtab.c (symtab_node::dump_base): Add missing new line.
35571 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
35573 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
35574 (ifcombine_ifandif): Use it.
35576 2017-02-03 Martin Liska <mliska@suse.cz>
35578 * doc/invoke.texi: Document default value for
35579 use-after-scope-direct-emission-threshold.
35581 2017-02-03 Martin Liska <mliska@suse.cz>
35583 PR tree-optimization/79339
35584 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
35585 (format_floating): Likewise.
35587 2017-02-03 Martin Liska <mliska@suse.cz>
35590 * ipa-prop.c (ipa_node_params_t::insert): Remove current
35592 (ipa_node_params_t::remove): Likewise.
35593 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
35594 initialization from removed ipa_node_params_t::insert.
35595 (ipa_node_params::~ipa_node_params): Move from removed
35596 ipa_node_params_t::release.
35597 * symbol-summary.h (symbol_summary::m_released): New member.
35598 Do not release a summary twice. Do not allow to call finalizer
35599 for types of a summary that live in GGC memory.
35601 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35603 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
35606 2017-02-02 Martin Sebor <msebor@redhat.com>
35608 PR middle-end/79275
35609 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
35610 (format_string): Tighten up the range of output for non-constant
35611 strings and correct the expected range for wide non-constant strings.
35613 2017-02-02 Martin Sebor <msebor@redhat.com>
35615 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
35617 PR middle-end/32003
35618 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
35620 (-fdump-tree-@var): Add to index and document how to come up
35621 with pass-specific option and dump file names.
35622 (-fdump-passes): Clarify where to look for output.
35624 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
35626 PR middle-end/77445
35627 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
35628 statistics of the analyzed path; allow threading for speed when
35629 any of BBs along the path are optimized for speed.
35631 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
35633 PR middle-end/78468
35634 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
35635 settings of the virtual registers.
35638 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
35640 * explow.c (get_dynamic_stack_size): Take known alignment of stack
35641 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
35644 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35646 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
35647 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
35649 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35651 * config/s390/s390.md: Add missing comments with the expanded
35653 * config/s390/vector.md: Likewise.
35654 * config/s390/vx-builtins.md: Likewise.
35656 2017-02-02 Jakub Jelinek <jakub@redhat.com>
35659 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
35660 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
35661 conditions on a single line.
35663 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35665 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35666 __S390_VX__ to __VX__.
35668 2017-02-01 Andrew Pinski <apinski@cavium.com>
35670 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
35671 stmt_info to record_stmt_cost.
35672 (vect_get_known_peeling_cost): Pass stmt_info if known to
35674 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
35675 cpu_vector_cost field into
35676 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
35677 field into vec_int_stmt_cost and vec_fp_stmt_cost.
35678 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
35679 splitting of scalar_stmt_cost and vec_stmt_cost.
35680 (thunderx_vector_cost): Likewise.
35681 (cortexa57_vector_cost): LIkewise.
35682 (exynosm1_vector_cost): Likewise.
35683 (xgene1_vector_cost): Likewise.
35684 (thunderx2t99_vector_cost): Improve after the splitting of the two
35686 (aarch64_builtin_vectorization_cost): Update for the splitting of
35687 scalar_stmt_cost and vec_stmt_cost.
35689 2017-02-01 Torvald Riegel <triegel@redhat.com>
35690 Richard Henderson <rth@redhat.com>
35692 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
35693 conditional on existance of a fast atomic load.
35694 * optabs-query.c (can_atomic_load_p): New function.
35695 * optabs-query.h (can_atomic_load_p): Declare it.
35696 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
35697 no fast atomic load is available for the particular size of access.
35698 (expand_atomic_compare_and_swap): Likewise.
35699 (expand_atomic_load): Likewise.
35700 (expand_atomic_store): Likewise.
35701 (expand_atomic_fetch_op): Likewise.
35702 * testsuite/lib/target-supports.exp
35703 (check_effective_target_sync_int_128): Remove x86 because it provides
35704 no fast atomic load.
35705 (check_effective_target_sync_int_128_runtime): Likewise.
35707 2017-02-01 Richard Biener <rguenther@suse.de>
35709 * graphite.c: Include tree-vectorizer.h for find_loop_location.
35710 (graphite_transform_loops): Provide opt-info for optimized nests.
35711 * tree-parloop.c (parallelize_loops): Provide opt-info for
35712 parallelized loops.
35714 2017-02-01 Richard Biener <rguenther@suse.de>
35716 PR middle-end/79315
35717 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
35718 was not set before.
35720 2017-02-01 Richard Biener <rguenther@suse.de>
35722 PR tree-optimization/71824
35723 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35724 Verify the loops are valid in the merged SESE region.
35725 (scop_detection::can_represent_loop_1): Check analyzing the
35726 evolution of the number of iterations in the region succeeds.
35728 2017-01-31 Ian Lance Taylor <iant@golang.org>
35730 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
35731 REG_ARGS_SIZE note to 32-bit push insns and call insn.
35733 2017-01-31 David Malcolm <dmalcolm@redhat.com>
35735 PR preprocessor/79210
35736 * input.c (get_substring_ranges_for_loc): Replace line_width
35737 assertion with error-handling.
35739 2017-01-31 Richard Biener <rguenther@suse.de>
35741 PR tree-optimization/77318
35742 * graphite-sese-to-poly.c (extract_affine): Fix assert.
35743 (create_pw_aff_from_tree): Take loop parameter.
35744 (add_condition_to_pbb): Pass loop of the condition to
35745 create_pw_aff_from_tree.
35747 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35749 * config/s390/s390.c (s390_asan_shadow_offset): New function.
35750 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
35752 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
35756 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
35758 (convert_int_to_float128): Likewise.
35759 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
35760 (convert_int_to_float128): Likewise.
35761 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
35762 (UNSPEC_IEEE128_CONVERT): Likewise.
35763 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
35764 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
35765 Use local variables for IBM extended format.
35766 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
35767 (fix_trunc<mode>si2_fprs): Likewise.
35768 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
35769 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
35770 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
35771 to know that we can now have integers of all sizes in vector
35773 (fix<uns>_<mode>di2_hw): Likewise.
35774 (float<uns>_<mode>si2_hw): Likewise.
35775 (fix_<mode>si2_hw): Likewise.
35776 (fixuns_<mode>si2_hw): Likewise.
35777 (float<uns>_<mode>di2_hw): Likewise.
35778 (float_<mode>di2_hw): Likewise.
35779 (float_<mode>si2_hw): Likewise.
35780 (floatuns_<mode>di2_hw): Likewise.
35781 (floatuns_<mode>si2_hw): Likewise.
35782 (xscvqp<su>wz_<mode>): Delete, no longer used.
35783 (xscvqp<su>dz_<mode>): Likewise.
35784 (xscv<su>dqp_<mode>): Likewise.
35785 (ieee128_mfvsrd_64bit): Likewise.
35786 (ieee128_mfvsrd_32bit): Likewise.
35787 (ieee128_mfvsrwz): Likewise.
35788 (ieee128_mtvsrw): Likewise.
35789 (ieee128_mtvsrd_64bit): Likewise.
35790 (ieee128_mtvsrd_32bit): Likewise.
35792 2017-01-31 Martin Liska <mliska@suse.cz>
35795 * ipa-prop.c (ipa_free_all_node_params): Call release method
35796 instead of ~sumbol_summary to not to trigger double times
35799 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
35801 PR tree-optimization/71691
35802 * bitmap.h (class auto_bitmap): New.
35803 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
35804 is_maybe_undefined instead of ssa_undefined_value_p.
35806 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35808 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35809 __S390_ARCH_LEVEL__ to __ARCH__.
35811 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35813 PR tree-optimization/79267
35814 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
35815 if should_remove_lhs_p is true.
35817 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
35820 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
35821 (add_alignment_attribute): New.
35822 (base_type_die): Add alignment attribute.
35823 (subrange_type_die): Likewise.
35824 (modified_type_die): Likewise.
35825 (gen_array_type_die): Likewise.
35826 (gen_descr_array_type_die: Likewise.
35827 (gen_enumeration_type_die): Likewise.
35828 (gen_subprogram_die): Likewise.
35829 (gen_variable_die): Likewise.
35830 (gen_field_die): Likewise.
35831 (gen_ptr_to_mbr_type_die): Likewise.
35832 (gen_struct_or_union_type_die): Likewise.
35833 (gen_subroutine_type_die): Likewise.
35834 (gen_typedef_die): Likewise.
35835 (base_type_cmp): Compare alignment attribute.
35837 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35840 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
35841 (setb_unsigned) New pattern for setb with CCUNS.
35842 * config/rs6000/rs6000.c (expand_block_compare): Use a different
35843 subfc./subfe sequence to avoid overflow problems. Generate a
35844 shorter sequence with cmpld/setb for power9.
35845 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
35846 for generating subfc. instruction.
35847 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
35848 now uses this instruction.
35850 2017-01-30 Ian Lance Taylor <iant@google.com>
35853 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
35854 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
35856 2017-01-30 Martin Sebor <msebor@redhat.com>
35858 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
35859 Move constant to the right of a relational operator.
35860 (get_mpfr_format_length, format_character, format_string): Ditto.
35861 (should_warn_p, maybe_warn): Same.
35863 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
35865 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
35868 * asan.c (get_translation_unit_decl): Remove function.
35869 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
35871 2017-01-30 Martin Liska <mliska@suse.cz>
35873 PR gcov-profile/79259
35874 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
35875 -fprofile-generate.
35877 2017-01-30 Martin Liska <mliska@suse.cz>
35880 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
35881 Initialize variables with NULL value.
35883 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
35886 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
35888 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
35890 2017-01-30 Richard Biener <rguenther@suse.de>
35892 PR tree-optimization/79276
35893 * tree-vrp.c (process_assert_insertions): Properly adjust common
35894 when removing a duplicate.
35896 2017-01-30 Richard Biener <rguenther@suse.de>
35898 PR tree-optimization/79256
35899 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35900 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35902 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
35904 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
35907 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
35908 ("*r<noxa>sbg_<mode>_sll_bitmask")
35909 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
35910 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
35911 Use contiguous_bitmask_nowrap_operand.
35913 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35916 * config/rs6000/altivec.h (vec_xl): Revise #define.
35917 (vec_xst): Likewise.
35919 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
35921 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
35923 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
35925 PR rtl-optimization/79194
35926 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
35927 traps before call to bypass_conditional_jumps.
35929 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35931 PR tree-optimization/71374
35932 * lra-constraints.c (check_conflict_input_operands): New.
35933 (match_reload): Use it.
35935 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35938 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
35939 account to calculate conflict_set.
35941 2017-01-27 Bin Cheng <bin.cheng@arm.com>
35943 PR rtl-optimization/78559
35944 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
35945 other_insn in combine.
35947 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
35949 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
35950 uint16_type_node for BT_UINT16.
35952 2017-01-27 David Malcolm <dmalcolm@redhat.com>
35954 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
35955 "RTL Tests" to menu.
35956 (GIMPLE Tests): New node.
35957 (RTL Tests): New node.
35959 2017-01-27 Richard Biener <rguenther@suse.de>
35961 PR tree-optimization/79245
35962 * tree-loop-distribution.c (distribute_loop): Apply cost
35963 modeling also to detected patterns.
35965 2017-01-27 Richard Biener <rguenther@suse.de>
35967 PR tree-optimization/71433
35968 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
35969 (compare_assert_loc): New function.
35970 (process_assert_insertions): Sort and optimize assert locations
35971 to remove duplicates and push down identical assertions on
35972 edges to their destination block.
35974 2017-01-27 Richard Biener <rguenther@suse.de>
35976 PR tree-optimization/79244
35977 * tree-vrp.c (remove_range_assertions): Forcefully propagate
35978 out SSA names even if abnormal.
35980 2017-01-27 Jakub Jelinek <jakub@redhat.com>
35982 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
35983 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
35984 instead of MPFR_RNDN.
35986 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
35989 * arm.c (arm_option_override): Don't call build_target_option_node
35990 until after doing all option overrides.
35991 (arm_valid_target_attribute_tree): Likewise.
35993 2017-01-27 Martin Liska <mliska@suse.cz>
35995 * doc/invoke.texi (-fprofile-arcs): Document profiling support
35996 for {cd}tors and C++ {cd}tors.
35998 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
36000 * config/s390/s390.md ("*setmem_long_and")
36001 ("*setmem_long_and_31z"): Use zero_extend instead of and.
36003 2017-01-26 Martin Sebor <msebor@redhat.com>
36005 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
36008 2017-01-26 Martin Sebor <msebor@redhat.com>
36010 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
36011 HAVE_DFmode before using XFmode or DFmode.
36012 (parse_directive): Avoid using the z length modifier to avoid
36013 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
36015 PR middle-end/78703
36016 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
36017 to accept adjustment as an array.
36018 (get_int_range): New function.
36019 (struct directive): Make width and prec arrays.
36020 (directive::set_width, directive::set_precision): Call get_int_range.
36021 (format_integer, format_floating): Handle width and precision ranges.
36022 (format_string, parse_directive): Same.
36024 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36027 * dwarf2out.c (generate_skeleton_bottom_up): For children with
36028 comdat_type_p set, just clone them, but keep the children in the
36032 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
36033 which have direct callers with -fvar-tracking-assignments enabled
36035 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
36036 inside of type units.
36038 2017-01-26 Martin Sebor <msebor@redhat.com>
36040 PR middle-end/78703
36041 * gimple-ssa-sprintf.c (struct result_range): Add likely and
36043 (struct format_result): Replace number_chars, number_chars_min,
36044 and number_chars_max with a single member of struct result_range.
36046 (format_result::operator+=): Adjust.
36047 (struct fmtresult): Remove bounded. Handle likely and unlikely
36049 (fmtresult::adjust_for_width_or_precision): New function.
36050 (fmtresult:type_max_digits): New function.
36051 (bytes_remaining): Handle likely and unlikely counters.
36052 (min_bytes_remaining): Remove.
36053 (format_percent): Simplify.
36054 (format_integer, format_floating): Set likely and unlikely counters.
36055 (get_string_length, format_character, format_string): Same.
36056 (format_plain, should_warn_p): New function.
36057 (maybe_warn): Call should_warn_p. Update diagnostic messages
36058 and handle those for all directives, including plain strings.
36059 (format_directive): Handle likely and unlikely counters.
36060 Remove unnecessary quoting from diagnostics. Add an informational
36062 (add_bytes): Remove.
36063 (pass_sprintf_length::compute_format_length): Simplify.
36064 (try_substitute_return_value): Handle likely and unlikely counters.
36066 2017-01-26 Carl Love <cel@us.ibm.com>
36068 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
36069 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
36071 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
36074 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
36075 endianess for subregs into account.
36076 * lra-constraints.c (lra_constraints): Do risky transformations
36077 always on the first iteration.
36078 * lra-lives.c (check_pseudos_live_through_calls): Add arg
36079 last_call_used_reg_set.
36080 (process_bb_lives): Define and use last_call_used_reg_set.
36081 * lra.c (lra): Always continue after lra_constraints on the first
36084 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
36086 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
36088 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
36090 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36092 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
36093 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
36094 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
36095 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
36096 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
36097 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
36098 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
36099 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
36100 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
36102 2017-01-26 Marek Polacek <polacek@redhat.com>
36105 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
36106 for the third operand.
36108 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36110 PR middle-end/79236
36111 * omp-low.c (struct omp_context): Add simt_stmt field.
36112 (scan_omp_for): Return omp_context *.
36113 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
36114 context to the _simt_ SIMD stmt.
36115 (lower_omp_for): For combined SIMD with sibling _simt_
36116 SIMD, make sure to use the same decls in _looptemp_
36117 clauses as in the sibling.
36119 2017-01-26 David Sherwood <david.sherwood@arm.com>
36121 PR middle-end/79212
36122 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
36125 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36128 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
36129 emit fld b; fld a; if possible.
36131 * brig-builtins.def: Update copyright years.
36132 * config/arm/arm_acle_builtins.def: Update copyright years.
36134 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
36137 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
36138 constraint instead of o for the stxsd instruction.
36140 2017-01-25 Carl Love <cel@us.ibm.com>
36142 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
36143 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
36145 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
36147 * doc/invoke.texi (C++ Dialect Options): Fix typo.
36149 2017-01-25 Richard Biener <rguenther@suse.de>
36151 PR tree-optimization/69264
36152 * target.def (vector_alignment_reachable): Improve documentation.
36153 * doc/tm.texi: Regenerate.
36154 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
36156 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
36157 earlier changes with respect to TYPE_USER_ALIGN.
36158 (vector_alignment_reachable_p): Likewise. Improve dumping.
36160 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36163 * config/arm/arm.md (xordi3): Force constant operand into a register
36166 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36168 * doc/invoke.texi (-fstore-merging): Correct default optimization
36169 levels at which it is enabled.
36170 (-O): Move -fstore-merging from list to...
36173 2017-01-25 Richard Biener <rguenther@suse.de>
36176 * omp-expand.c: Include debug.h.
36177 (expand_omp_taskreg): Make sure to generate early debug before
36178 outlining anything from a function.
36179 (expand_omp_target): Likewise.
36180 (grid_expand_target_grid_body): Likewise.
36182 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
36185 * asan.c (get_translation_unit_decl): New function.
36186 (asan_add_global): Extract modules file name from globals
36187 TRANSLATION_UNIT_DECL name.
36189 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
36192 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
36193 for long calls with APCS frame and VFP.
36195 2017-01-24 David Malcolm <dmalcolm@redhat.com>
36197 * cfg.c (original_copy_tables_initialized_p): New function.
36198 * cfg.h (original_copy_tables_initialized_p): New decl.
36199 * cfgrtl.c (relink_block_chain): Guard the call to
36200 free_original_copy_tables with a call to
36201 original_copy_tables_initialized_p.
36202 * cgraph.h (symtab_node::native_rtl_p): New decl.
36203 * cgraphunit.c (symtab_node::native_rtl_p): New function.
36204 (symtab_node::needed_p): Don't assert for early assembly output
36205 for __RTL functions.
36206 (cgraph_node::finalize_function): Set "force_output" for __RTL
36208 (cgraph_node::analyze): Bail out early for __RTL functions.
36209 (analyze_functions): Update assertion to support __RTL functions.
36210 (cgraph_node::expand): Bail out early for __RTL functions.
36211 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
36213 * function.h (struct function): Update comment for field
36215 * gimple-expr.c: Include "tree-pass.h".
36216 (gimple_has_body_p): Return false for __RTL functions.
36217 * Makefile.in (OBJS): Add run-rtl-passes.o.
36218 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
36220 (gcc::pass_manager::get_clean_slate): New accessor.
36221 * passes.c: Include "insn-addr.h".
36222 (should_skip_pass_p): Add logging. Update logic for running
36223 "expand" to be compatible with both __GIMPLE and __RTL. Guard
36224 property-provider override so it is only done for gimple passes.
36226 (skip_pass): New function.
36227 (execute_one_pass): Call skip_pass when skipping passes.
36228 * read-md.c (md_reader::read_char): Support filtering
36229 the input to a subset of line numbers.
36230 (md_reader::md_reader): Initialize fields
36231 m_first_line and m_last_line.
36232 (md_reader::read_file_fragment): New function.
36233 * read-md.h (md_reader::read_file_fragment): New decl.
36234 (md_reader::m_first_line): New field.
36235 (md_reader::m_last_line): New field.
36236 * read-rtl-function.c (function_reader::create_function): Only
36237 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
36238 curr_properties. Set DECL_INITIAL to a dummy block.
36239 (read_rtl_function_body_from_file_range): New function.
36240 * read-rtl-function.h (read_rtl_function_body_from_file_range):
36242 * run-rtl-passes.c: New file.
36243 * run-rtl-passes.h: New file.
36245 2017-01-24 Jeff Law <law@redhat.com>
36247 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
36250 2017-01-24 Bin Cheng <bin.cheng@arm.com>
36252 PR tree-optimization/79159
36253 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
36254 (record_nonwrapping_iv): Improve boundary using above function if no
36255 value range information.
36257 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
36258 Martin Jambor <mjambor@suse.cz>
36260 * brig-builtins.def: New file.
36261 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
36262 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
36263 (DEF_HSAIL_SAT_BUILTIN): Likewise.
36264 (DEF_HSAIL_INTR_BUILTIN): Likewise.
36265 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
36266 * builtin-types.def (BT_INT8): New.
36267 (BT_INT16): Likewise.
36268 (BT_UINT8): Likewise.
36269 (BT_UINT16): Likewise.
36270 (BT_FN_ULONG): Likewise.
36271 (BT_FN_UINT_INT): Likewise.
36272 (BT_FN_UINT_ULONG): Likewise.
36273 (BT_FN_UINT_LONG): Likewise.
36274 (BT_FN_UINT_PTR): Likewise.
36275 (BT_FN_ULONG_PTR): Likewise.
36276 (BT_FN_INT8_FLOAT): Likewise.
36277 (BT_FN_INT16_FLOAT): Likewise.
36278 (BT_FN_UINT32_FLOAT): Likewise.
36279 (BT_FN_UINT16_FLOAT): Likewise.
36280 (BT_FN_UINT8_FLOAT): Likewise.
36281 (BT_FN_UINT64_FLOAT): Likewise.
36282 (BT_FN_UINT16_UINT32): Likewise.
36283 (BT_FN_UINT32_UINT16): Likewise.
36284 (BT_FN_UINT16_UINT16_UINT16): Likewise.
36285 (BT_FN_INT_PTR_INT): Likewise.
36286 (BT_FN_UINT_PTR_UINT): Likewise.
36287 (BT_FN_LONG_PTR_LONG): Likewise.
36288 (BT_FN_ULONG_PTR_ULONG): Likewise.
36289 (BT_FN_VOID_UINT64_UINT64): Likewise.
36290 (BT_FN_UINT8_UINT8_UINT8): Likewise.
36291 (BT_FN_INT8_INT8_INT8): Likewise.
36292 (BT_FN_INT16_INT16_INT16): Likewise.
36293 (BT_FN_INT_INT_INT): Likewise.
36294 (BT_FN_UINT_FLOAT_UINT): Likewise.
36295 (BT_FN_FLOAT_UINT_UINT): Likewise.
36296 (BT_FN_ULONG_UINT_UINT): Likewise.
36297 (BT_FN_ULONG_UINT_PTR): Likewise.
36298 (BT_FN_ULONG_ULONG_ULONG): Likewise.
36299 (BT_FN_UINT_UINT_UINT): Likewise.
36300 (BT_FN_VOID_UINT_PTR): Likewise.
36301 (BT_FN_UINT_UINT_PTR: Likewise.
36302 (BT_FN_UINT32_UINT64_PTR): Likewise.
36303 (BT_FN_INT_INT_UINT_UINT): Likewise.
36304 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
36305 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
36306 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
36307 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
36308 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
36309 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
36310 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
36311 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
36312 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
36313 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
36314 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
36315 * doc/frontends.texi: List BRIG FE.
36316 * doc/install.texi (Testing): Add BRIG tesring requirements.
36317 * doc/invoke.texi (Overall Options): Mention BRIG.
36318 * doc/standards.texi (Standards): Doucment BRIG HSA version.
36320 2017-01-24 Richard Biener <rguenther@suse.de>
36322 PR translation/79208
36323 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
36325 2017-01-24 Martin Jambor <mjambor@suse.cz>
36328 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
36329 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
36330 and known_contexts.
36332 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
36334 PR middle-end/79123
36335 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
36336 casts from signed to unsigned really don't have a range.
36338 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
36340 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
36341 GMP_RNDx for compatiblity.
36343 2017-01-24 Martin Liska <mliska@suse.cz>
36346 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
36347 that would prevent us to call alloca with -1 as argument.
36349 2017-01-24 Jakub Jelinek <jakub@redhat.com>
36351 * dwarf2out.c (output_compilation_unit_header, output_file_names):
36352 Avoid -Wformat-security warning.
36354 2017-01-23 Andrew Pinski <apinski@cavium.com>
36356 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
36359 2017-01-23 Martin Sebor <msebor@redhat.com>
36361 PR middle-end/78703
36362 * gimple-ssa-sprintf.c (warn_level): New global.
36363 (format_integer): Use it here and throughout the rest of the file.
36364 Use the same switch to compute sign as base.
36365 (maybe_warn): New function.
36366 (format_directive): Factor out warnings into maybe_warn.
36367 Add debugging output. Use warn_level.
36368 (add_bytes): Use warn_level.
36369 (pass_sprintf_length::compute_format_length): Add debugging output.
36370 (try_substitute_return_value): Same.
36371 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
36373 PR middle-end/78703
36374 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
36375 (struct fmtresult, format_integer, format_floating): Adjust.
36376 (fmtresult::fmtresult): Set max correctly in two argument ctor.
36377 (get_string_length, format_string,format_directive): Same.
36378 (pass_sprintf_length::compute_format_length): Same.
36379 (try_substitute_return_value): Simplify slightly.
36381 PR middle-end/78703
36382 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
36383 (fmtresult::operator+=): Outlined.
36384 (struct fmtresult): Add ctors.
36385 (struct conversion_spec): Rename...
36386 (struct directive): ...to this. Add and remove data members.
36387 (directive::set_width, directive::set_precision): New functions.
36388 (format_percent): Use fmtresult ctor.
36389 (get_width_and_precision): Remove.
36390 (format_integer): Make naming changes. Avoid computing width and
36392 (format_floating): Same. Adjust indentation.
36393 (format_character, format_none): New functions.
36394 (format_string): Moved character handling to format_character.
36395 (format_directive): Remove arguments, change return type.
36396 (parse_directive): New function.
36397 (pass_sprintf_length::compute_format_length): Move directive
36398 parsing to parse_directive.
36400 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36402 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
36403 (assign_assembler_name_if_needed): ... this.
36404 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
36405 (assign_assembler_name_if_needed): ... this.
36406 (free_lang_data_in_cgraph): Adjust callers.
36407 * cgraphunit.c (cgraph_node::analyze): Likewise.
36408 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
36411 2017-01-23 Richard Biener <rguenther@suse.de>
36413 PR tree-optimization/79088
36414 PR tree-optimization/79188
36415 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
36416 resetting loop bounds after last path deletion. Reset loop
36417 bounds of the target loop, make code match the comments.
36418 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
36419 Make sure loops need no fixups.
36421 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
36423 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
36424 exponent support with double type for first argument.
36425 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
36426 type returned by __builtin_vec_extract_sig,
36427 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
36428 functions from "vector int" to "vector unsigned int" or from
36429 "vector long long int" to "vector unsigned long long int".
36430 Changed type returned by __builtin_vec_extract_exp,
36431 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
36432 functions from "vector int" to "vector unsigned int" or from
36433 "vector long long int" to "vector unsigned long long int".
36434 Changed return type of __builtin_vec_test_data_class,
36435 __builtin_vec_test_data_class_sp, and
36436 __builtin_vec_test_data_class_dp from "vector int" to
36437 "vector bool int" or from "vector long long int" to "vector bool
36438 long long int" and changed second argument type from "unsigned
36439 int" to "int". Added new overloaded function forms "vector float
36440 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
36441 "vector float __builtin_vec_insert_exp_sp (vector float, vector
36442 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
36443 double, vector unsigned long long int)" and "vector double
36444 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
36445 long int)". Changed return type of
36446 __builtin_scalar_test_data_class and
36447 __builtin_scalar_test_data_class_sp and
36448 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
36449 int" and changed second argument from "unsigned int" to "int".
36450 Changed type returned by __builtin_scalar_test_neg,
36451 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
36452 from "int" to "bool int". Added new overloaded function form
36453 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
36454 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
36455 exponent double-precision with floating point first argument.
36456 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
36457 documentation of scalar_test_data_class, scalar_test_neg,
36458 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
36459 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
36460 vec_test_data_class built-in functions to reflect refinements in
36461 their type signatures.
36463 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
36465 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
36467 (aarch64_elf_asm_destructor): Likewise.
36469 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
36471 PR rtl-optimization/78634
36472 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
36473 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
36474 * ifcvt.c (noce_try_cmove): Add missing cost check.
36476 PR rtl-optimization/71724
36477 * combine.c (if_then_else_cond): Look for situations where it is
36478 beneficial to undo the work of one of the recursive calls.
36480 2017-01-23 Bin Cheng <bin.cheng@arm.com>
36482 PR tree-optimization/70754
36483 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
36484 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
36485 combined stmt before it if not NULL.
36486 (combine_chains): Process refs reversely and compute dominance point
36489 2017-01-23 Martin Liska <mliska@suse.cz>
36491 PR tree-optimization/79196
36492 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
36493 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
36495 (strlen_optimize_stmt): Call the renamed function.
36497 2017-01-23 Michael Matz <matz@suse.de>
36499 PR tree-optimization/78384
36500 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
36502 2017-01-23 Richard Biener <rguenther@suse.de>
36504 PR tree-optimization/79186
36505 * tree-vrp.c (register_new_assert_for): Make sure we've seen
36506 both incoming edges before moving an assert.
36508 2017-01-23 Martin Jambor <mjambor@suse.cz>
36510 * ipa-prop.c (load_from_param_1): Removed.
36511 (load_from_unmodified_param): Bits from load_from_param_1 put back
36513 (load_from_param): Removed.
36514 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
36515 with stmt. Reverted back to use of load_from_unmodified_param.
36517 2017-01-23 Martin Jambor <mjambor@suse.cz>
36520 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
36521 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
36522 field a pointer to garbage collected vector, mark lattices and
36523 ipcp_orig_node with GTY((skip)).
36524 (ipa_get_param_count): Adjust to descriptors being a pointer.
36525 (ipa_get_param): Likewise.
36526 (ipa_get_type): Likewise.
36527 (ipa_get_param_move_cost): Likewise.
36528 (ipa_set_param_used): Likewise.
36529 (ipa_get_controlled_uses): Likewise.
36530 (ipa_set_controlled_uses): Likewise.
36531 (ipa_is_param_used): Likewise.
36532 (ipa_node_params_t): Move into garbage collector. New methods insert
36534 (ipa_node_params_sum): Annotate wth GTY(()).
36535 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
36537 (ipa_load_from_parm_agg): Adjust declaration.
36538 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
36539 * ipa-profile.c (ipa_profile): Likewise.
36540 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
36541 (ipa_populate_param_decls): Make descriptors parameter garbage
36543 (ipa_dump_param): Adjust to descriptors being a pointer.
36544 (ipa_alloc_node_params): Likewise.
36545 (ipa_initialize_node_params): Likewise.
36546 (load_from_param_1): Make descriptors parameter garbage collected.
36547 (load_from_unmodified_param): Likewise.
36548 (load_from_param): Likewise.
36549 (ipa_load_from_parm_agg): Likewise.
36550 (ipa_node_params::~ipa_node_params): Removed.
36551 (ipa_free_all_node_params): Remove call to delete operator.
36552 (ipa_node_params_t::insert): New.
36553 (ipa_node_params_t::remove): Likewise.
36554 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
36555 copy known_csts and known_contexts vectors.
36556 (ipa_read_node_info): Adjust to descriptors being a pointer.
36557 (ipcp_modif_dom_walker): Make m_descriptors field garbage
36559 (ipcp_transform_function): Make descriptors variable garbage
36562 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
36564 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
36565 * config/i386/avx512dqintrin.h: Ditto.
36566 * config/i386/avx512fintrin.h: Ditto.
36567 * config/i386/i386.c: Handle new builtins.
36568 * config/i386/i386-builtin.def: Add new builtins.
36569 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
36570 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
36572 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36573 Martin Liska <mliska@suse.cz>
36575 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
36576 * asan.c (asan_expand_poison_ifn): Support stores and use
36577 appropriate ASAN report function.
36578 * internal-fn.c (expand_ASAN_POISON_USE): New function.
36579 * internal-fn.def (ASAN_POISON_USE): Declare.
36580 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
36581 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
36582 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
36583 ASAN_POISON calls w/o LHS.
36584 * tree-ssa.c (execute_update_addresses_taken): Create clobber
36585 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
36586 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
36587 * gimplify.c (asan_poison_variables): Add attribute
36588 use_after_scope_memory to variables that really needs to live
36590 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
36591 having the attribute.
36593 2017-01-23 Martin Liska <mliska@suse.cz>
36595 * asan.c (create_asan_shadow_var): New function.
36596 (asan_expand_poison_ifn): Likewise.
36597 * asan.h (asan_expand_poison_ifn): New declaration.
36598 * internal-fn.c (expand_ASAN_POISON): Likewise.
36599 * internal-fn.def (ASAN_POISON): New builtin.
36600 * sanopt.c (pass_sanopt::execute): Expand
36601 asan_expand_poison_ifn.
36602 * tree-inline.c (copy_decl_for_dup_finish): Make function
36604 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
36605 * tree-ssa.c (is_asan_mark_p): New function.
36606 (execute_update_addresses_taken): Rewrite local variables
36607 (identified just by use-after-scope as addressable) into SSA.
36609 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
36611 * doc/install.texi (Specific): opensource.apple.com uses https
36612 now. Remove trailing slash.
36614 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
36616 * README.Portability: Remove note on an Irix compatibility issue.
36618 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
36620 * gcov.c (INCLUDE_ALGORITHM): Define.
36621 (INCLUDE_VECTOR): Define.
36622 No longer include <vector> and <algorithm> directly.
36624 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
36626 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
36628 * doc/invoke.texi (Code Gen Options): Ditto.
36630 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
36633 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
36635 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
36637 rtl-optimization/79125
36638 * cprop.c (local_cprop_pass): Handle cases where we make an
36639 unconditional trap.
36641 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
36645 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
36646 read from, for big endian.
36648 2017-01-20 Jiong Wang <jiong.wang@arm.com>
36650 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
36651 register pauth builtins for LP64 only.
36653 2017-01-20 Marek Polacek <polacek@redhat.com>
36656 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
36659 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
36661 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
36663 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
36664 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
36665 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
36667 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36670 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
36671 in big-endian mode when they are not a single duplicated value.
36673 2017-01-20 Richard Biener <rguenther@suse.de>
36675 * BASE-VER: Bump to 7.0.1.
36677 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
36679 * omp-low.c (omplow_simd_context): New struct. Use it...
36680 (lower_rec_simd_input_clauses): ...here and...
36681 (lower_rec_input_clauses): ...here to hold common data. Adjust all
36682 references to idx, lane, max_vf, is_simt.
36684 2017-01-20 Graham Markall <graham.markall@embecosm.com>
36686 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
36689 2017-01-20 Martin Jambor <mjambor@suse.cz>
36691 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
36692 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
36694 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
36695 (GTFILES): Rename hsa.c to hsa-common.c.
36696 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
36697 * hsa-dump.c: Likewise.
36698 * hsa-gen.c: Likewise.
36699 * hsa-regalloc.c: Likewise.
36700 * ipa-hsa.c: Likewise.
36701 * omp-expand.c: Likewise.
36702 * omp-low.c: Likewise.
36703 * toplev.c: Likewise.
36705 2017-01-20 Marek Polacek <polacek@redhat.com>
36708 * doc/invoke.texi: Document -Wduplicated-branches.
36709 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
36710 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
36711 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
36712 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
36713 return 0 only when not OEP_LEXICOGRAPHIC.
36714 (fold_build_cleanup_point_expr): Use the expression
36715 location when building CLEANUP_POINT_EXPR.
36716 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
36717 * tree.c (add_expr): Handle error_mark_node.
36719 2017-01-20 Martin Liska <mliska@suse.cz>
36722 * tree-profile.c (init_ic_make_global_vars): Do not call
36724 (gimple_init_gcov_profiler): Likewise.
36726 2017-01-20 Martin Liska <mliska@suse.cz>
36729 * cgraph.h (maybe_create_reference): Remove argument and
36731 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
36733 * ipa-cp.c (create_specialized_node): Likewise.
36734 * symtab.c (symtab_node::maybe_create_reference): Handle
36735 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
36737 2017-01-20 Martin Liska <mliska@suse.cz>
36739 * read-rtl-function.c (function_reader::create_function): Use
36740 build_decl instread of build_decl_stat.
36742 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
36744 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
36745 * config/i386/avx512dqintrin.h: Ditto.
36746 * config/i386/avx512fintrin.h: Ditto.
36747 * config/i386/i386-builtin-types.def: Add new types.
36748 * config/i386/i386.c: Handle new types.
36749 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
36750 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
36751 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
36752 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
36753 (__builtin_ia32_kshiftridi): New.
36754 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
36756 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
36760 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
36761 define to rs6000_init_stack_protect_guard.
36762 (rs6000_init_stack_protect_guard): New function.
36764 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36765 Yunqiang Su <yunqiang.su@imgtec.com>
36767 * config.gcc (supported_defaults): Add madd4.
36768 (with_madd4): Add validation.
36769 (all_defaults): Add madd4.
36770 * config/mips/mips.opt (mmadd4): New option.
36771 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
36773 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
36775 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
36776 (ISA_HAS_FUSED_MADD4): Likewise.
36777 * doc/invoke.texi (-mmadd4): Document the new option.
36778 * doc/install.texi (--with-madd4): Document the new option.
36780 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36782 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
36783 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
36784 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
36785 (aarch64_init_pauth_hint_builtins): New.
36786 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
36787 (aarch64_expand_builtin): Expand new builtins.
36789 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36791 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
36792 * combine-stack-adj.c (no_unhandled_cfa): Handle
36793 REG_CFA_TOGGLE_RA_MANGLE.
36794 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
36795 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
36796 info for return address signing.
36797 (aarch64_expand_epilogue): Likewise.
36799 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36801 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
36802 * config/aarch64/aarch64-protos.h
36803 (aarch64_return_address_signing_enabled): New declaration.
36804 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
36806 (aarch64_expand_prologue): Sign return address before it's pushed onto
36808 (aarch64_expand_epilogue): Authenticate return address fetched from
36810 (aarch64_override_options): Sanity check for ILP32 and ISA level.
36811 (aarch64_attributes): New function attributes for "sign-return-address".
36812 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
36813 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
36814 ("*do_return"): Generate combined instructions according to key index.
36815 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
36816 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
36818 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
36819 * config/aarch64/aarch64.opt (msign-return-address=): New.
36820 * doc/extend.texi (AArch64 Function Attributes): Documents
36821 "sign-return-address=".
36822 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
36824 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36826 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
36827 overall option summary.
36829 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36831 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
36832 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
36833 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
36834 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
36836 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
36838 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
36839 -mpower9-minmax by default for -mcpu=power9.
36840 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
36841 128-bit floating point.
36843 2017-01-20 Alan Modra <amodra@gmail.com>
36845 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
36846 optimizing for size.
36848 2017-01-20 Alan Modra <amodra@gmail.com>
36851 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
36852 for strcmp and strncmp from corresponding builtin decl.
36854 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36856 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
36857 instead of i386/rtems-64.h.
36858 * config/i386/rtems-64.h: Remove.
36860 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36864 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
36866 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
36868 2017-01-19 Tamar Christina <tamar.christina@arm.com>
36870 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
36871 Change int to HOST_WIDE_INT.
36872 * config/aarch64/aarch64-protos.h
36873 (aarch64_simd_gen_const_vector_dup): Likewise.
36874 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
36876 2017-01-19 David Malcolm <dmalcolm@redhat.com>
36878 * langhooks-def.h (lhd_type_for_size): New decl.
36879 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
36880 * langhooks.c (lhd_type_for_size): New function, taken from
36883 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
36885 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
36886 define_bypass for CR latency.
36887 (power9-cracked-alu): Update bypass latency and remove power9-branch.
36888 (power9-alu2): Add define_bypass for CR latency.
36890 (power9-mul): Update insn latency.
36891 (power9-mul-compare): Update insn latency, bypass latency and remove
36894 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36896 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
36898 * config/aarch64/aarch64.md
36899 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
36900 aarch64_nopcrelative_literal_loads.
36901 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
36903 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
36905 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
36906 TARGET_LOONGSON_3A.
36907 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
36909 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
36912 * config.gcc (supported_defaults): Add lxc1-sxc1.
36913 (with_lxc1_sxc1): Add validation.
36914 (all_defaults): Add lxc1-sxc1.
36915 * config/mips/mips.opt (mlxc1-sxc1): New option.
36916 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
36918 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
36919 __mips_no_lxc1_sxc1.
36920 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
36921 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
36922 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
36924 2017-01-19 Richard Biener <rguenther@suse.de>
36926 PR tree-optimization/72488
36927 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
36928 sure to restore SSA info.
36929 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
36931 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
36933 PR rtl-optimization/79121
36934 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
36935 of the inner type when shifting an extended value.
36937 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
36940 * symtab.c (symtab_node::equal_address_to): Fix comparing of
36941 interposable aliases.
36943 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
36946 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
36947 Use the evmergelohi instruction.
36948 (mov_si<mode>_e500_subreg4_2_le): Likewise.
36949 (mov_sitf_e500_subreg8_2_be): Likewise.
36950 (mov_sitf_e500_subreg12_2_le): Likewise.
36951 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
36952 (mov_si<mode>_e500_subreg4_2_be): Likewise.
36953 (mov_sitf_e500_subreg8_2_le): Likewise.
36954 (mov_sitf_e500_subreg12_2_be): Likewise.
36956 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36958 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
36959 attribute from vecsimple to vecperm.
36960 (altivec_vbpermq2): Likewise.
36962 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36965 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
36967 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36968 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
36969 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
36970 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
36971 case where N arg is SIZE_MAX.
36972 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
36973 (cmpstrsi): Add pattern.
36975 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
36977 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
36978 __builtin_vec_revb builtins.
36979 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
36980 built-in functions to support generation of the ISA 3.0 XXBR<x>
36981 vector byte reverse instructions.
36982 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
36983 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
36984 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
36985 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
36986 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
36987 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
36988 (P9V_BUILTIN_VEC_REVB): Likewise.
36989 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
36990 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
36991 (p9_xxbrq_v16qi): Likewise.
36992 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
36993 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
36994 (p9_xxbrh_v8hi): Likewise.
36995 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
36996 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
36997 vec_revb built-in functions.
36999 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
37001 PR rtl-optimization/78952
37002 * config/i386/i386.md (any_extract): New code iterator.
37003 (*insvqi_2): Use any_extract for source operand.
37004 (*insvqi_3): Use any_shiftrt for source operand.
37006 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
37008 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
37010 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
37012 2017-01-18 Matthias Klose <doko@ubuntu.com>
37014 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
37016 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37018 * config/rs6000/altivec.h (vec_bperm): Change #define.
37019 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
37020 (altivec_vbpermq2): New define_insn.
37021 (altivec_vbpermd): Likewise.
37022 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
37023 function interface.
37024 (VBPERMD): Likewise.
37025 (VBPERM): New polymorphic function interface.
37026 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
37027 Add entries for P9V_BUILTIN_VEC_VBPERM.
37028 * doc/extend.texi: Add interfaces for vec_bperm.
37030 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37032 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
37033 first letter of error messages.
37034 (s390_resolve_overloaded_builtin): Likewise.
37035 * config/s390/s390.c (s390_expand_builtin): Likewise.
37036 (s390_invalid_arg_for_unprototyped_fn): Likewise.
37037 (s390_valid_target_attribute_inner_p): Likewise.
37038 * config/s390/s390.md ("tabort"): Likewise.
37040 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
37042 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
37043 (ISA_AVOID_DIV_HILO): New macro.
37044 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
37045 (ISA_HAS_DDIV): Likewise.
37047 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37049 * doc/invoke.texi (fabi-version): Correct number of occurrences.
37051 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37053 * doc/invoke.texi (fabi-version): Spelling fix.
37055 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37058 * doc/invoke.texi (fabi-version): Mention mangling fix for
37061 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37064 * doc/invoke.texi (fabi-version): Document discriminator mangling.
37066 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
37069 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
37070 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
37072 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
37074 (stack_protect_test): Ditto.
37075 * config/rs6000/rs6000.opt (mstack-protector-guard=,
37076 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
37078 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
37079 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
37080 -mstack-protector-guard-offset=.
37081 (RS/6000 and PowerPC Options): Ditto.
37083 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37085 * config/i386/i386.h (MASK_CLASS_P): New define.
37086 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
37087 there are no registers from different register sets also when
37088 mask registers are used. Update function comment.
37089 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
37090 to (*k/*r) and (*k/*km) alternatives.
37092 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
37094 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
37095 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
37096 (EH_RETURN_HANDLER_RTX): New define.
37097 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
37098 Force frame pointer in EH return functions.
37099 (aarch64_expand_epilogue): Add barrier for eh_return.
37100 (aarch64_final_eh_return_addr): Remove.
37101 (aarch64_eh_return_handler_rtx): New function.
37102 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
37104 (aarch64_eh_return_handler_rtx): New prototype.
37106 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37108 * config/rs6000/altivec.h (vec_rlmi): New #define.
37109 (vec_vrlnm): Likewise.
37110 (vec_rlnm): Likewise.
37111 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
37112 (UNSPEC_VRLNM): Likewise.
37113 (VIlong): New mode iterator.
37114 (altivec_vrl<VI_char>mi): New define_insn.
37115 (altivec_vrl<VI_char>nm): Likewise.
37116 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
37118 (VRLDNM): Likewise.
37119 (RLNM): New polymorphic function entry.
37120 (VRLWMI): New monomorphic function entry.
37121 (VRLDMI): Likewise.
37122 (RLMI): New polymorphic function entry.
37123 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
37124 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
37125 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
37128 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37131 * dwarf2out.c (field_byte_offset): Restore the
37132 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
37133 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
37134 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
37137 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37140 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
37142 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37145 * dwarf2out.c (add_data_member_location_attribute): For constant
37146 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
37147 instead of DW_AT_data_member_location, DW_AT_bit_offset and
37148 DW_AT_byte_size attributes.
37150 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37152 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
37153 after forcing to constant memory when the code model is medium.
37155 2017-01-17 Julia Koval <julia.koval@intel.com>
37158 * config/i386/avx512fintrin.h
37159 (_mm512_i32gather_ps): Change __addr type to void const*.
37160 (_mm512_mask_i32gather_ps): Ditto.
37161 (_mm512_i32gather_pd): Ditto.
37162 (_mm512_mask_i32gather_pd): Ditto.
37163 (_mm512_i64gather_ps): Ditto.
37164 (_mm512_mask_i64gather_ps): Ditto.
37165 (_mm512_i64gather_pd): Ditto.
37166 (_mm512_mask_i64gather_pd): Ditto.
37167 (_mm512_i32gather_epi32): Ditto.
37168 (_mm512_mask_i32gather_epi32): Ditto.
37169 (_mm512_i32gather_epi64): Ditto.
37170 (_mm512_mask_i32gather_epi64): Ditto.
37171 (_mm512_i64gather_epi32): Ditto.
37172 (_mm512_mask_i64gather_epi32): Ditto.
37173 (_mm512_i64gather_epi64): Ditto.
37174 (_mm512_mask_i64gather_epi64): Ditto.
37175 (_mm512_i32scatter_ps): Change __addr type to void*.
37176 (_mm512_mask_i32scatter_ps): Ditto.
37177 (_mm512_i32scatter_pd): Ditto.
37178 (_mm512_mask_i32scatter_pd): Ditto.
37179 (_mm512_i64scatter_ps): Ditto.
37180 (_mm512_mask_i64scatter_ps): Ditto.
37181 (_mm512_i64scatter_pd): Ditto.
37182 (_mm512_mask_i64scatter_pd): Ditto.
37183 (_mm512_i32scatter_epi32): Ditto.
37184 (_mm512_mask_i32scatter_epi32): Ditto.
37185 (_mm512_i32scatter_epi64): Ditto.
37186 (_mm512_mask_i32scatter_epi64): Ditto.
37187 (_mm512_i64scatter_epi32): Ditto.
37188 (_mm512_mask_i64scatter_epi32): Ditto.
37189 (_mm512_i64scatter_epi64): Ditto.
37190 (_mm512_mask_i64scatter_epi64): Ditto.
37191 * config/i386/avx512pfintrin.h
37192 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
37193 (_mm512_mask_prefetch_i32gather_ps): Ditto.
37194 (_mm512_mask_prefetch_i64gather_pd): Ditto.
37195 (_mm512_mask_prefetch_i64gather_ps): Ditto.
37196 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
37197 (_mm512_prefetch_i32scatter_ps): Ditto.
37198 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
37199 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
37200 (_mm512_prefetch_i64scatter_pd): Ditto.
37201 (_mm512_prefetch_i64scatter_ps): Ditto.
37202 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
37203 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
37204 * config/i386/avx512vlintrin.h
37205 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
37206 (_mm_mmask_i32gather_ps): Ditto.
37207 (_mm256_mmask_i32gather_pd): Ditto.
37208 (_mm_mmask_i32gather_pd): Ditto.
37209 (_mm256_mmask_i64gather_ps): Ditto.
37210 (_mm_mmask_i64gather_ps): Ditto.
37211 (_mm256_mmask_i64gather_pd): Ditto.
37212 (_mm_mmask_i64gather_pd): Ditto.
37213 (_mm256_mmask_i32gather_epi32): Ditto.
37214 (_mm_mmask_i32gather_epi32): Ditto.
37215 (_mm256_mmask_i32gather_epi64): Ditto.
37216 (_mm_mmask_i32gather_epi64): Ditto.
37217 (_mm256_mmask_i64gather_epi32): Ditto.
37218 (_mm_mmask_i64gather_epi32): Ditto.
37219 (_mm256_mmask_i64gather_epi64): Ditto.
37220 (_mm_mmask_i64gather_epi64): Ditto.
37221 (_mm256_i32scatter_ps): Change __addr type to void*.
37222 (_mm256_mask_i32scatter_ps): Ditto.
37223 (_mm_i32scatter_ps): Ditto.
37224 (_mm_mask_i32scatter_ps): Ditto.
37225 (_mm256_i32scatter_pd): Ditto.
37226 (_mm256_mask_i32scatter_pd): Ditto.
37227 (_mm_i32scatter_pd): Ditto.
37228 (_mm_mask_i32scatter_pd): Ditto.
37229 (_mm256_i64scatter_ps): Ditto.
37230 (_mm256_mask_i64scatter_ps): Ditto.
37231 (_mm_i64scatter_ps): Ditto.
37232 (_mm_mask_i64scatter_ps): Ditto.
37233 (_mm256_i64scatter_pd): Ditto.
37234 (_mm256_mask_i64scatter_pd): Ditto.
37235 (_mm_i64scatter_pd): Ditto.
37236 (_mm_mask_i64scatter_pd): Ditto.
37237 (_mm256_i32scatter_epi32): Ditto.
37238 (_mm256_mask_i32scatter_epi32): Ditto.
37239 (_mm_i32scatter_epi32): Ditto.
37240 (_mm_mask_i32scatter_epi32): Ditto.
37241 (_mm256_i32scatter_epi64): Ditto.
37242 (_mm256_mask_i32scatter_epi64): Ditto.
37243 (_mm_i32scatter_epi64): Ditto.
37244 (_mm_mask_i32scatter_epi64): Ditto.
37245 (_mm256_i64scatter_epi32): Ditto.
37246 (_mm256_mask_i64scatter_epi32): Ditto.
37247 (_mm_i64scatter_epi32): Ditto.
37248 (_mm_mask_i64scatter_epi32): Ditto.
37249 (_mm256_i64scatter_epi64): Ditto.
37250 (_mm256_mask_i64scatter_epi64): Ditto.
37251 (_mm_i64scatter_epi64): Ditto.
37252 (_mm_mask_i64scatter_epi64): Ditto.
37253 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
37254 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
37255 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
37256 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
37257 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
37258 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
37259 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
37260 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
37261 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
37262 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
37263 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
37264 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
37265 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
37266 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
37267 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
37268 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
37269 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
37270 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
37271 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
37272 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
37273 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
37274 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
37275 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
37276 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
37277 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
37278 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
37279 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
37280 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
37281 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
37282 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
37283 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
37284 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
37285 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
37286 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
37287 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
37288 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
37289 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
37290 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
37291 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
37292 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
37293 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
37294 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
37295 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
37296 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
37297 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
37298 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
37299 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
37300 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
37301 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
37302 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
37303 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
37304 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
37305 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
37306 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
37307 definitions accordingly.
37309 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
37310 Kuan-Lin Chen <kuanlinchentw@gmail.com>
37313 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
37316 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
37319 * ira-conflicts.c (ira_build_conflicts): Update total conflict
37320 hard regs for inner regno.
37322 2017-01-17 Martin Liska <mliska@suse.cz>
37325 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
37326 assumption and add comment.
37328 2017-01-17 Nathan Sidwell <nathan@acm.org>
37330 * ipa-visibility.c (localize_node): New function, broken out of ...
37331 (function_and_variable_visibility): ... here. Call it.
37333 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
37335 PR middle-end/77445
37336 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
37337 correctly set frequency of oudgoing edge.
37338 (duplicate_thread_path): Fix profile updating.
37340 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37343 * configure.ac: Add GCC_BASE_VER.
37344 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
37345 version from BASE-VER file.
37346 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
37347 (gcc.o): Depend on $(BASEVER).
37348 * common.opt (dumpfullversion): New option.
37349 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
37350 * doc/invoke.texi: Document -dumpfullversion.
37351 * doc/install.texi: Document --with-gcc-major-version-only.
37352 * configure: Regenerated.
37354 2017-01-17 Richard Biener <rguenther@suse.de>
37356 PR tree-optimization/71433
37357 * tree-vrp.c (register_new_assert_for): Merge same asserts
37358 on all incoming edges.
37359 (process_assert_insertions_for): Handle insertions at the
37362 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
37364 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
37365 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
37367 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
37370 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
37373 2017-01-17 Alan Modra <amodra@gmail.com>
37376 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
37377 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
37378 symbolic stack limit when pic.
37380 2017-01-16 Martin Sebor <msebor@redhat.com>
37382 PR tree-optimization/78608
37383 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
37385 2017-01-16 Jeff Law <law@redhat.com>
37388 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
37389 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
37390 for several include directories that may be relative to sysroot.
37391 * config/i386/x-mingw32 (gplus_includedir): Define.
37392 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
37393 (native_system_includedir): Likewise.
37394 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
37395 override if TARGET_SYSTEM_ROOT is defined.
37396 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
37398 PR tree-optimization/79090
37399 PR tree-optimization/33562
37400 PR tree-optimization/61912
37401 PR tree-optimization/77485
37402 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
37403 and computed trims into the dump file.
37405 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37407 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
37409 2017-01-16 Jakub Jelinek <jakub@redhat.com>
37412 * gimplify.c (gimplify_init_constructor): If want_value and
37413 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
37417 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
37418 sequence. Formatting fixes.
37419 (doloop_optimize): Formatting fixes.
37422 * gcc.c (debug_level_greater_than_spec_func): New function.
37423 (static_spec_functions): Add debug-level-gt spec function.
37424 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
37426 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37427 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37428 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
37429 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
37430 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
37431 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
37433 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
37435 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
37436 QImode fixups to general and mask registers only.
37438 2017-01-16 Carl Love <cel@us.ibm.com>
37440 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
37441 for built-in functions
37442 vector signed char vec_nabs (vector signed char)
37443 vector signed short vec_nabs (vector signed short)
37444 vector signed int vec_nabs (vector signed int)
37445 vector signed long long vec_nabs (vector signed long long)
37446 vector float vec_nabs (vector float)
37447 vector double vec_nabs (vector double)
37448 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
37450 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
37451 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
37452 * doc/extend.texi: Update the documentation file for the new built-in
37455 2017-01-16 Martin Sebor <msebor@redhat.com>
37457 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
37460 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37462 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
37463 UNSPEC_VSX__XXSPLTD to require special splat handling.
37465 2017-01-16 David Malcolm <dmalcolm@redhat.com>
37468 * system.h: Poison strndup.
37470 2017-01-16 Alan Modra <amodra@gmail.com>
37473 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
37476 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
37478 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
37480 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
37482 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
37483 call recog here. Assert that INSN_CODE (insn) is non-negative.
37485 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37488 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
37490 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
37491 in the currently scheduled RTL fragment.
37493 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37495 PR rtl-optimization/78751
37496 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
37499 2017-01-14 Jeff Law <law@redhat.com>
37501 PR tree-optimization/79090
37502 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
37503 variable length stores.
37504 (compute_trims): Delete dead assignment to *trim_tail.
37505 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
37508 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
37510 PR rtl-optimization/78626
37511 PR rtl-optimization/78727
37512 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
37513 of a block, and split such blocks after everything else is finished.
37515 2017-01-14 Alan Modra <amodra@gmail.com>
37518 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
37519 target legitimate_combined_insn.
37520 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
37521 (rs6000_legitimate_combined_insn): New function.
37522 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
37524 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
37525 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
37526 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
37528 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
37530 * doc/frontends.texi (G++ and GCC): Remove references to Java.
37532 2017-01-13 Jeff Law <law@redhat.com>
37534 PR tree-optimization/33562
37535 PR tree-optimization/61912
37536 PR tree-optimization/77485
37537 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
37539 (delete_dead_assignment): Likewise.
37540 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
37541 statement to delete_dead_call and delete_dead_assignment.
37543 2017-01-13 David Malcolm <dmalcolm@redhat.com>
37546 * substring-locations.c (format_warning_va): Strengthen case 1 so
37547 that both endpoints of the substring must be within the format
37548 range for just the substring to be printed.
37550 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
37552 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
37553 * config/i386/i386.c (ix86_target_string): Add missing options
37554 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
37555 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
37556 flags_other and ix86_target_other to flags2_other. Display unknown
37558 (ix86_valid_target_attribute_inner_p): Add missing options and
37559 reorder options by implied ISAs, as in ix86_target_string.
37561 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
37563 * hash-table.h (hash_table::too_empty_p): New function.
37564 (hash_table::expand): Use it.
37565 (hash_table::traverse): Likewise.
37566 (hash_table::empty_slot): Use sizeof (value_type) instead of
37567 sizeof (PTR) to convert bytes to elements. Shrink the table
37568 if the current size is excessive for the current number of
37571 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
37573 * ira-costs.c (record_reg_classes): Break from the inner loop
37574 early once alt_fail is known to be true. Update outer loop
37575 handling accordingly.
37577 2017-01-13 Jeff Law <law@redhat.com>
37579 * tree-ssa-dse.c (decrement_count): New function.
37580 (increment_start_addr, maybe_trim_memstar_call): Likewise.
37581 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
37582 when we know the partially dead statement is a mem* function.
37584 PR tree-optimization/61912
37585 PR tree-optimization/77485
37586 * tree-ssa-dse.c: Include expr.h.
37587 (maybe_trim_constructor_store): New function.
37588 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
37590 PR tree-optimization/33562
37591 PR tree-optimization/61912
37592 PR tree-optimization/77485
37593 * doc/invoke.texi: Document new dse-max-object-size param.
37594 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
37595 * tree-ssa-dse.c: Include params.h.
37596 (dse_store_status): New enum.
37597 (initialize_ao_ref_for_dse): New, partially extracted from
37599 (valid_ao_ref_for_dse, normalize_ref): New.
37600 (setup_live_bytes_from_ref, compute_trims): Likewise.
37601 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
37602 (maybe_trim_partially_dead_store): Likewise.
37603 (maybe_trim_complex_store): Likewise.
37604 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
37605 Track what bytes live from the original store. Return tri-state
37606 for dead, partially dead or live.
37607 (dse_dom_walker): Add constructor, destructor and new private members.
37608 (delete_dead_call, delete_dead_assignment): New extracted from
37610 (dse_optimize_stmt): Make a member of dse_dom_walker.
37611 Use initialize_ao_ref_for_dse.
37613 PR tree-optimization/33562
37614 PR tree-optimization/61912
37615 PR tree-optimization/77485
37616 * sbitmap.h (bitmap_count_bits): Prototype.
37617 (bitmap_clear_range, bitmap_set_range): Likewise.
37618 * sbitmap.c (bitmap_clear_range): New function.
37619 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
37621 2017-01-13 Martin Liska <mliska@suse.cz>
37624 * function.c (set_cfun): Add new argument force.
37625 * function.h (set_cfun): Likewise.
37626 * ipa-inline-transform.c (inline_call): Use the function when
37627 strict alising from is dropped for function we inline to.
37629 2017-01-13 Richard Biener <rguenther@suse.de>
37631 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
37632 for dumping GIMPLE INTEGER_CSTs.
37634 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
37636 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
37637 to 201112L since C++17.
37639 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
37642 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
37643 if -fsanitize=kernel-address is present.
37645 2017-01-13 Richard Biener <rguenther@suse.de>
37647 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
37648 as _Literal ( type ) number in case usual suffixes do not
37649 preserve all information.
37651 2017-01-13 Richard Biener <rguenther@suse.de>
37653 PR tree-optimization/77283
37654 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
37655 and ssa-iterators.h.
37656 (is_feasible_trace): Implement a cost model based on joiner
37659 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
37662 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
37663 char or short to __float128/_Float128 directly.
37665 2017-01-12 Martin Sebor <msebor@redhat.com>
37667 to -Wformat-overflow.
37668 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
37669 (min_bytes_remaining): Same.
37670 (get_string_length): Same.
37671 (format_string): Same.
37672 (format_directive): Same.
37674 (pass_sprintf_length::handle_gimple_call): Same.
37676 2017-01-12 Jakub Jelinek <jakub@redhat.com>
37678 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
37679 info.nowrite calls with no lhs that can't throw. Return bool
37680 whether gsi_remove has been called or not.
37681 (pass_sprintf_length::handle_gimple_call): Return bool whether
37682 try_substitute_return_value called gsi_remove. Formatting fix.
37683 (pass_sprintf_length::execute): Don't use gsi_remove if
37684 handle_gimple_call returned true.
37687 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
37688 be removed due to side-effects, don't remove following barrier nor
37689 turn the successor edge into fallthru edge.
37691 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37694 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
37695 element-reversing loads and stores as not swappable.
37697 2017-01-12 Nathan Sidwell <nathan@acm.org>
37698 Nicolai Stange <nicstange@gmail.com>
37700 * combine.c (try_combine): Don't ignore result of overlap checking
37701 loop. Combine overlap & asm check into single loop.
37703 2017-01-12 Richard Biener <rguenther@suse.de>
37705 * tree-pretty-print.c (dump_generic_node): Provide -gimple
37706 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
37708 2017-01-12 Richard Biener <rguenther@suse.de>
37710 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
37711 and TS_TARGET_OPTION directly derive from TS_BASE.
37712 * tree-core.h (tree_optimization_option): Derive from tree_base.
37713 (tree_target_option): Likewise.
37715 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
37717 * config/i386/i386.c (memory_address_length): Increase len
37718 only when rip_relative_addr_p returns false.
37720 2017-01-11 Julia Koval <julia.koval@intel.com>
37722 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
37723 (OPTION_MASK_ISA_SGX_SET): New.
37724 (ix86_handle_option): Handle OPT_msgx.
37725 * config.gcc: Added sgxintrin.h.
37726 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
37727 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
37728 * config/i386/i386.c (ix86_target_string): Add -msgx.
37730 (ix86_option_override_internal): Handle new options.
37731 (ix86_valid_target_attribute_inner_p): Add sgx.
37732 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
37733 * config/i386/i386.opt: Add msgx.
37734 * config/i386/sgxintrin.h: New file.
37735 * config/i386/x86intrin.h: Add sgxintrin.h.
37737 2017-01-11 Jakub Jelinek <jakub@redhat.com>
37740 * fold-const.c (maybe_nonzero_address): Return 1 for function
37742 (tree_single_nonzero_warnv_p): Don't handle function local objects
37746 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
37749 2017-01-11 David Malcolm <dmalcolm@redhat.com>
37752 * opts.c: Include "spellcheck.h"
37753 (struct string_fragment): New struct.
37754 (struct edit_distance_traits<const string_fragment &>): New
37756 (get_closest_sanitizer_option): New function.
37757 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
37759 2017-01-11 Jakub Jelinek <jakub@redhat.com>
37761 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
37763 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
37764 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
37765 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
37766 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
37767 for initial die_offset if dwarf_split_debug_info.
37768 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
37769 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
37771 (output_skeleton_debug_sections): Formatting fix. Use
37772 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
37773 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
37775 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
37777 * config/arm/cortex-a53.md: Add bypasses for
37778 cortex_a53_r2f_cvt.
37779 (cortex_a53_r2f): Only use for transfers.
37780 (cortex_a53_f2r): Likewise.
37781 (cortex_a53_r2f_cvt): Add reservation for conversions.
37782 (cortex_a53_f2r_cvt): Likewise.
37784 2017-01-11 Tamar Christina <tamar.christina@arm.com>
37786 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
37787 to all inlined functions, change static to extern.
37789 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
37792 * config/arm/arm.c (legitimize_pic_address): Handle reference to
37794 (arm_assemble_integer): Likewise.
37796 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
37798 * config.gcc: Use new awk script to check CPU, FPU and architecture
37799 parameters for --with-... options.
37800 * config/arm/parsecpu.awk: New file
37801 * config/arm/arm-cpus.in: New file.
37802 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
37804 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
37806 * config/arm/t-arm: Update dependency rules.
37807 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
37808 of processing .def files.
37809 * config/arm/genopt.sh: Deleted.
37810 * config/arm/gentune.sh: Deleted.
37811 * config/arm/arm-cores.def: Deleted.
37812 * config/arm/arm-arches.def: Deleted.
37813 * config/arm/arm-fpus.def: Deleted.
37814 * config/arm/arm-tune.md: Regenerated.
37815 * config/arm/arm-tables.opt: Regenerated.
37816 * config/arm/arm-cpu.h: New generated file.
37817 * config/arm/arm-cpu-data.h: New generated file.
37818 * config/arm/arm-cpu-cdata.h: New generated file.
37820 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
37823 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
37825 (input_varpool_node): Unpack dynamically_initialized bit.
37827 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
37829 PR rtl-optimization/79032
37830 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
37831 the alignment of the adjusted memory reference against that of MODE,
37832 instead of the alignment of the original memory reference.
37834 2017-01-11 Martin Jambor <mjambor@suse.cz>
37836 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
37838 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
37839 decorated functions.
37841 2017-01-11 Richard Biener <rguenther@suse.de>
37843 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
37844 set range/nonnull info for PHI results. Do not set it on
37845 stmts marked for removal.
37847 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
37849 * expr.c (store_field): In the bitfield case, fetch the return value
37850 from the registers before applying a single big-endian adjustment.
37851 Always do a final load for a BLKmode value not larger than a word.
37853 2017-01-10 David Malcolm <dmalcolm@redhat.com>
37856 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
37857 that we correctly handle column numbers greater than
37858 LINE_MAP_MAX_COLUMN_NUMBER.
37860 2017-01-10 Martin Sebor <msebor@redhat.com>
37862 PR middle-end/78245
37863 * gimple-ssa-sprintf.c (get_destination_size): Call
37864 {init,fini}object_sizes.
37865 * tree-object-size.c (addr_object_size): Adjust.
37866 (pass_through_call): Adjust.
37867 (pass_object_sizes::execute): Adjust.
37868 * tree-object-size.h (fini_object_sizes): Declare.
37870 2017-01-10 Martin Sebor <msebor@redhat.com>
37872 PR tree-optimization/78775
37873 * builtins.c (get_size_range): Move...
37874 * calls.c: ...to here.
37875 (alloc_max_size): Accept zero argument.
37876 (operand_signed_p): Remove.
37877 (maybe_warn_alloc_args_overflow): Call get_size_range.
37878 * calls.h (get_size_range): Declare.
37880 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
37882 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
37883 from TI's devices.csv file as of September 2016.
37884 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
37886 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
37888 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
37889 * doc/invoke.texi: Likewise.
37890 * doc/md.texi: Likewise.
37891 * doc/objc.texi: Likewise.
37893 2017-01-10 Joshua Conner <joshconner@google.com>
37895 * config/arm/fuchsia-elf.h: New file.
37896 * config/fuchsia.h: New file.
37897 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
37898 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
37900 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
37902 2016-01-10 Richard Biener <rguenther@suse.de>
37904 PR tree-optimization/79034
37905 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
37906 Propagate out degenerate PHIs in the joiner.
37908 2017-01-10 Martin Liska <mliska@suse.cz>
37910 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
37911 (sort_congruence_classes_by_decl_uid): Likewise.
37912 (sort_congruence_class_groups_by_decl_uid): Likewise.
37913 (sem_item_optimizer::merge_classes): Sort class, groups in these
37914 classes and members in the groups by DECL_UID of declarations.
37915 This would make merge operations stable.
37917 2017-01-10 Martin Liska <mliska@suse.cz>
37919 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
37920 usage of m_classes_vec.
37921 (sem_item_optimizer::~sem_item_optimizer): Likewise.
37922 (sem_item_optimizer::get_group_by_hash): Likewise.
37923 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
37924 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
37925 (sem_item_optimizer::verify_classes): Likewise.
37926 (sem_item_optimizer::process_cong_reduction): Likewise.
37927 (sem_item_optimizer::dump_cong_classes): Likewise.
37928 (sem_item_optimizer::merge_classes): Likewise.
37929 * ipa-icf.h (congruence_class_hash): Rename from
37930 congruence_class_group_hash. Remove declaration of m_classes_vec.
37932 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
37934 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
37935 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
37936 * config.gcc: Add avx512vpopcntdqintrin.h.
37937 * config/i386/avx512vpopcntdqintrin.h: New.
37938 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
37939 * config/i386/i386-builtin-types.def: Add new types.
37940 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
37941 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
37942 __builtin_ia32_vpopcountq_v8di_mask): New.
37943 * config/i386/i386-c.c (ix86_target_macros_internal): Define
37944 __AVX512VPOPCNTDQ__.
37945 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
37946 (PTA_AVX512VPOPCNTDQ): Define.
37947 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
37948 TARGET_AVX512VPOPCNTDQ_P): Define.
37949 * config/i386/i386.opt: Add mavx512vpopcntdq.
37950 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
37951 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
37953 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37955 PR middle-end/77484
37956 * predict.def (PRED_CALL): Set to 67.
37958 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
37960 * expr.c (store_field): In the bitfield case, if the value comes from
37961 a function call and is of an aggregate type returned in registers, do
37962 not modify the field mode; extract the value in all cases if the mode
37963 is BLKmode and the size is not larger than a word.
37965 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
37968 * config/i386/cpuid.h: Fix undefined behavior.
37970 2017-01-04 Jeff Law <law@redhat.com>
37972 PR tree-optimization/79007
37973 PR tree-optimization/67955
37974 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
37975 conservative for pt.null when flag_non_call_exceptions is on.
37977 2017-01-09 Jakub Jelinek <jakub@redhat.com>
37979 PR translation/79019
37980 PR translation/79020
37981 * params.def (PARAM_INLINE_MIN_SPEEDUP,
37982 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
37983 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
37985 * config/avr/avr.opt (maccumulate-args): Likewise.
37986 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
37987 * common.opt (freport-bug): Likewise.
37988 * cif-code.def (CIF_FINAL_ERROR): Likewise.
37989 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
37990 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
37991 translatable string.
37992 * config/i386/i386.c (function_value_32): Likewise.
37993 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
37994 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
37996 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
37997 * common/config/msp430/msp430-common.c (msp430_handle_option):
37999 * symtab.c (symtab_node::verify_base): Likewise.
38000 * opts.c (set_debug_level): Likewise.
38001 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
38002 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
38003 missing whitespace to translatable strings.
38004 * config/avr/avr.md (bswapsi2): Fix typo in comment.
38005 * config/sh/superh.h: Likewise.
38006 * config/i386/xopintrin.h: Likewise.
38007 * config/i386/znver1.md: Likewise.
38008 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
38009 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
38010 * double-int.h (struct double_int): Likewise.
38011 * double-int.c (div_and_round_double): Likewise.
38012 * wide-int.cc: Likewise.
38013 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
38014 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
38015 * cfgcleanup.c (crossjumps_occured): Renamed to ...
38016 (crossjumps_occurred): ... this.
38017 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
38020 PR tree-optimization/78899
38021 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
38022 returning bool return struct loop *, NULL for failure and the new
38024 (versionable_outer_loop_p): Don't version outer loop if it has
38025 dont_vectorized bit set.
38026 (tree_if_conversion): When versioning outer loop, ensure
38027 tree_if_conversion is performed also on the inner loop of the
38028 non-vectorizable outer loop copy.
38029 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
38030 LOOP_VECTORIZED in inner loop of the scalar outer loop and
38031 prevent vectorization of it.
38032 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
38033 the outer loop vectorization of the non-scalar version is attempted
38034 before vectorization of the inner loop in scalar version. If
38035 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
38036 vectorization of its inner loop.
38037 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
38038 has 2 inner loops, rename also on edges from bb whose single pred
38039 is outer_loop->header. Fix typo in function comment.
38041 2017-01-09 Martin Sebor <msebor@redhat.com>
38044 * asan.c (asan_emit_stack_protection): Increase local buffer size
38045 to avoid snprintf truncation warning.
38047 2017-01-09 Andrew Pinski <apinski@cavium.com>
38049 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
38050 to reference thunderx2t99 for the tuning structure
38051 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
38053 (thunderx2t99_extra_costs): This.
38054 * config/aarch64/aarch64-tune.md: Regenerate.
38055 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
38056 (vulcan_addrcost_table): This.
38057 (vulcan_regmove_cost): Rename to ...
38058 (thunderx2t99_regmove_cost): This.
38059 (vulcan_vector_cost): Rename to ...
38060 (thunderx2t99_vector_cost): this.
38061 (vulcan_branch_cost): Rename to ...
38062 (thunderx2t99_branch_cost): This.
38063 (vulcan_tunings): Rename to ...
38064 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
38065 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
38067 2017-01-09 Martin Jambor <mjambor@suse.cz>
38071 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
38072 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
38073 (propagate_vr_accross_jump_function): Use the above function for all
38074 value range computations for pass-through jump functions and type
38075 converasion from explicit value range values.
38076 (ipcp_propagate_stage): Do not attempt to deduce types of formal
38077 parameters from TYPE_ARG_TYPES.
38078 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
38079 (ipa_write_node_info): Stream type of the actual argument.
38080 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
38082 2017-01-09 Martin Liska <mliska@suse.cz>
38085 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
38086 (lookup_compiler): Do not show error message with have_E.
38088 2017-01-09 Jakub Jelinek <jakub@redhat.com>
38090 PR tree-optimization/78938
38091 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
38092 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
38093 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
38094 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
38097 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38099 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
38102 2017-01-09 Richard Biener <rguenther@suse.de>
38104 PR tree-optimization/78997
38105 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
38106 name condition properly.
38108 2017-01-09 Richard Biener <rguenther@suse.de>
38111 * dwarf2out.c (is_cxx): New overload with context.
38112 (is_naming_typedef_decl): Use it.
38114 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38116 * invoke.texi (Option Summary): Correct spacing in option lists
38117 and add line breaks to fix over-long lines.
38119 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38121 PR middle-end/17660
38123 * extend.texi (Common Variable Attributes): Add xref to GCC
38124 Internals manual to explain mode attribute keywords.
38126 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38129 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
38130 and Preprocessor Options.
38131 (Options for Linking): Document -pthread here....
38132 (RS/6000 and PowerPC Options): ...not here.
38133 (Solaris 2 Options): ...or here.
38134 * doc/cppopts.texi: Document -pthread.
38136 2017-01-08 Martin Sebor <msebor@redhat.com>
38138 PR middle-end/77708
38139 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
38140 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
38141 New member functions.
38142 (format_directive): Used them.
38144 (pass_sprintf_length::handle_gimple_call): Same.
38145 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
38146 to avoid truncation for any argument.
38147 (extract_affine_mul): Same.
38148 * tree.c (get_file_function_name): Same.
38150 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38152 PR middle-end/77484
38153 * predict.def (PRED_INDIR_CALL): Set to 86.
38155 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38157 PR preprocessor/54124
38158 * doc/cppopts.texi: Reformat -d subtable to list the full name
38159 of the options. Add cross-reference to the docs for the general
38160 compiler -d options.
38161 * doc/invoke.texi (Developer Options): Add cross-reference to the
38162 preprocessor-specific -d option documentation.
38164 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38166 PR preprocessor/13498
38167 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
38168 redudant material, and reflect new command-line options.
38169 (System Headers): Likewise.
38171 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38173 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
38174 -isystem, and -idirafter. Copy-edit.
38175 * doc/cppopts.texi: Copy-edit. Remove contradiction about
38176 default for -ftrack-macro-expansion. Delete obsolete and
38177 badly-formatted implementation details about -fdebug-cpp output.
38178 * doc/cppwarnopts.texi: Copy-edit.
38180 2017-01-07 David Malcolm <dmalcolm@redhat.com>
38183 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
38184 that the transition from a max line width >= 1<<10 to narrower
38185 lines works correctly.
38187 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
38189 * doc/options.texi (PerFunction): New.
38190 * opt-functions.awk (switch_flags): Map both Optimization and
38191 PerFunction to CL_OPTIMIZATION.
38192 * opth-gen.awk: Test for PerFunction flag along with
38194 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
38195 it only when the latter is present. Skip those that don't in
38196 the hash function generator.
38197 * common.opt (fvar-tracking): Mark as PerFunction instead of
38199 (fvar-tracking-assignments): Likewise.
38200 (fvar-tracking-assignments-toggle): Likewise.
38201 (fvar-tracking-uninit): Likewise.
38203 2017-01-07 Jakub Jelinek <jakub@redhat.com>
38205 PR translation/79018
38206 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
38209 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
38212 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
38213 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
38214 TARGET_LONG_JUMP_TABLE_OFFSETS.
38215 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
38216 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
38217 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
38218 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
38219 * config/m68k/m68k.md (tablejump expander): Likewise.
38220 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
38221 TARGET_LONG_JUMP_TABLE_OFFSETS.
38222 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
38223 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
38225 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
38226 David Holsgrove <david.holsgrove@xilinx.com>
38228 * common/config/microblaze/microblaze-common.c
38229 (TARGET_EXCEPT_UNWIND_INFO): Remove.
38230 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
38232 * config/microblaze/microblaze.c (microblaze_must_save_register)
38233 (microblaze_expand_epilogue, microblaze_return_addr): Handle
38235 (microblaze_eh_return): New function.
38236 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
38237 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
38238 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
38239 * config/microblaze/microblaze.md (eh_return): New pattern.
38241 2017-01-06 Jakub Jelinek <jakub@redhat.com>
38243 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
38244 GCC_DIAGNOSTIC_STRINGIFY): Define.
38246 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
38248 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38250 * config/arm/arm.md (<mcrr>): New.
38252 * config/arm/arm.c (arm_arch5te): New.
38253 (arm_option_override): Set arm_arch5te.
38254 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
38256 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
38257 (arm_mcrr_qualifiers): ... this. New.
38258 (MRRC_QUALIFIERS): Define to...
38259 (arm_mrrc_qualifiers): ... this. New.
38260 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
38262 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
38263 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
38264 (MRRCI, mrrc, MRRC): New.
38265 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
38266 VUNSPEC_MRRC2): New.
38268 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38270 * config/arm/arm.md (<mcr>): New.
38272 * config/arm/arm.c (arm_coproc_builtin_available): Add
38273 support for mcr, mrc, mcr2 and mrc2.
38274 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
38275 (arm_mcr_qualifiers): ... this. New.
38276 (MRC_QUALIFIERS): Define to ...
38277 (arm_mrc_qualifiers): ... this. New.
38278 (MCR_QUALIFIERS): Define to ...
38279 (arm_mcr_qualifiers): ... this. New.
38280 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
38282 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
38283 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
38284 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
38285 VUNSPEC_MRC2): New.
38287 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38289 * config/arm/arm.md (*ldc): New.
38293 * config/arm/arm.c (arm_coproc_builtin_available): Add
38294 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
38295 (arm_coproc_ldc_stc_legitimate_address): New.
38296 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
38297 'qualifier_const_pointer'.
38298 (LDC_QUALIFIERS): Define to...
38299 (arm_ldc_qualifiers): ... this. New.
38300 (STC_QUALIFIERS): Define to...
38301 (arm_stc_qualifiers): ... this. New.
38302 * config/arm/arm-protos.h
38303 (arm_coproc_ldc_stc_legitimate_address): New.
38304 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
38305 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
38306 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
38307 stc2, stcl, stc2l): New.
38308 * config/arm/constraints.md (Uz): New.
38309 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
38310 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
38311 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
38312 VUNSPEC_STC2L): New.
38314 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38316 * config/arm/arm.md (<cdp>): New.
38317 * config/arm/arm.c (neon_const_bounds): Rename this ...
38318 (arm_const_bounds): ... this.
38319 (arm_coproc_builtin_available): New.
38320 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
38321 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
38322 (CDP_QUALIFIERS): Define to...
38323 (arm_cdp_qualifiers): ... this. New.
38325 (arm_expand_builtin_args): Add case for 6 arguments.
38326 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
38327 (arm_const_bounds): ... this.
38328 (arm_coproc_builtin_available): New.
38329 * config/arm/arm_acle.h (__arm_cdp): New.
38331 * config/arm/arm_acle_builtins.def (cdp): New.
38333 * config/arm/iterators.md (CDPI,CDP,cdp): New.
38334 * config/arm/neon.md: Rename all 'neon_const_bounds' to
38335 'arm_const_bounds'.
38336 * config/arm/types.md (coproc): New.
38337 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
38338 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
38339 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
38340 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
38342 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38344 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
38345 (UBINOP_QUALIFIERS): New.
38347 (acle_builtin_data): New. Change comment.
38348 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
38349 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
38350 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
38351 arm_acle_builtins.def.
38352 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
38353 (arm_init_acle_builtins): New.
38354 (CRC32_BUILTIN): Remove.
38355 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
38356 crc32cb, crc32ch and crc32cw.
38357 (arm_init_crc32_builtins): Remove.
38358 (arm_init_builtins): Use arm_init_acle_builtins rather
38359 than arm_init_crc32_builtins.
38360 (arm_expand_acle_builtin): New.
38361 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
38362 * config/arm/arm_acle_builtins.def: New.
38364 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38366 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
38367 (arm_builtin_datum): ... this.
38368 (arm_init_neon_builtin): Rename to ...
38369 (arm_init_builtin): ... this. Add a new parameters PREFIX
38370 and USE_SIG_IN_NAME.
38371 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
38372 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
38373 'arm_builtin_datum'.
38374 (arm_init_vfp_builtins): Likewise.
38375 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
38376 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
38377 (arm_expand_neon_args): Rename to ...
38378 (arm_expand_builtin_args): ... this. Rename builtin_arg
38379 enum values and differentiate between ARG_BUILTIN_MEMORY
38380 and ARG_BUILTIN_NEON_MEMORY.
38381 (arm_expand_neon_builtin_1): Rename to ...
38382 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
38383 values, arm_expand_builtin_args and add bool parameter NEON.
38384 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
38385 (arm_expand_vfp_builtin): Likewise.
38386 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
38388 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38390 PR middle-end/77484
38391 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
38392 * predict.c (tree_estimate_probability_bb): Reverse direction of
38393 polymorphic call predictor.
38395 2017-01-06 David Malcolm <dmalcolm@redhat.com>
38397 * passes.c (execute_one_pass): Split out pass-skipping logic into...
38398 (determine_pass_name_match): ...this new function and...
38399 (should_skip_pass_p): ...this new function.
38401 2017-01-06 Nathan Sidwell <nathan@acm.org>
38403 * ipa-visibility.c (function_and_variable_visibility): Reformat
38404 comments and long lines. Remove extrneous if.
38405 * symtab.c (symtab_node::make_decl_local): Fix code format.
38406 (symtab_node::set_section_for_node): Fix comment typo.
38408 2017-01-06 Martin Liska <mliska@suse.cz>
38411 * lra-constraints.c: Rename invariant to lra_invariant.
38412 * predict.c (set_even_probabilities): Initialize e to NULL.
38414 2017-01-05 Martin Sebor <msebor@redhat.com>
38416 PR tree-optimization/78910
38417 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
38418 (format_integer): Correct off-by-one error in the handling
38419 of precision with negative numbers in signed conversions..
38421 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
38423 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
38425 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38427 PR tree-optimization/71016
38428 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
38429 factor_out_conditional_conversion. Formatting fix.
38430 (factor_out_conditional_conversion): Add cond_stmt argument.
38431 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
38432 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
38435 2017-01-05 David Malcolm <dmalcolm@redhat.com>
38437 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
38438 read-rtl-function.o, and selftest-rtl.o.
38439 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
38440 (selftest::aarch64_test_loading_full_dump): New function.
38441 (selftest::aarch64_run_selftests): New function.
38442 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
38443 selftest::aarch64_run_selftests.
38444 * config/i386/i386.c
38445 (selftest::ix86_test_loading_dump_fragment_1): New function.
38446 (selftest::ix86_test_loading_call_insn): New function.
38447 (selftest::ix86_test_loading_full_dump): New function.
38448 (selftest::ix86_test_loading_unspec): New function.
38449 (selftest::ix86_run_selftests): Call the new functions.
38450 * emit-rtl.c (maybe_set_max_label_num): New function.
38451 * emit-rtl.h (maybe_set_max_label_num): New decl.
38452 * function.c (instantiate_decls): Guard call to
38453 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
38454 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
38456 * gensupport.c (gen_reader::gen_reader): Pass "false"
38457 for new "compact" param of rtx_reader.
38458 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
38459 rather than an empty string for NULL strings.
38460 * read-md.c: Potentially include config.h rather than bconfig.h.
38461 Wrap include of errors.h with #ifdef GENERATOR_FILE.
38462 (have_error): New global, copied from errors.c.
38463 (md_reader::read_name): Rename to...
38464 (md_reader::read_name_1): ...this, adding "out_loc" param,
38465 and converting "missing name or number" to returning false, rather
38467 (md_reader::read_name): Reimplement in terms of read_name_1.
38468 (md_reader::read_name_or_nil): New function.
38469 (md_reader::read_string): Handle "(nil)" by returning NULL.
38470 (md_reader::md_reader): Add new param "compact".
38471 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
38472 (md_reader::read_file): New method.
38473 * read-md.h (md_reader::md_reader): Add new param "compact".
38474 (md_reader::read_file): New method.
38475 (md_reader::is_compact): New accessor.
38476 (md_reader::read_name): Convert return type from void to file_location.
38477 (md_reader::read_name_or_nil): New decl.
38478 (md_reader::read_name_1): New decl.
38479 (md_reader::m_compact): New field.
38480 (noop_reader::noop_reader): Pass "false" for new "compact" param
38482 (rtx_reader::rtx_reader): Add new "compact" param.
38483 (rtx_reader::read_rtx_operand): Make virtual and convert return
38484 type from void to rtx.
38485 (rtx_reader::read_until): New decl.
38486 (rtx_reader::handle_any_trailing_information): New virtual function.
38487 (rtx_reader::postprocess): New virtual function.
38488 (rtx_reader::finalize_string): New virtual function.
38489 (rtx_reader::m_in_call_function_usage): New field.
38490 (rtx_reader::m_reuse_rtx_by_id): New field.
38491 * read-rtl-function.c: New file.
38492 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
38493 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
38494 (selftest::verify_three_block_rtl_cfg): New decl.
38495 * read-rtl-function.h: New file.
38496 * read-rtl.c: Potentially include config.h rather than bconfig.h.
38497 For host, include function.h, memmodel.h, and emit-rtl.h.
38498 (one_time_initialization): New function.
38499 (struct compact_insn_name): New struct.
38500 (compact_insn_names): New array.
38501 (find_code): Handle insn codes in compact dumps.
38502 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
38503 (bind_subst_iter_and_attr): Likewise.
38504 (add_condition_to_string): Likewise.
38505 (add_condition_to_rtx): Likewise.
38506 (apply_attribute_uses): Likewise.
38507 (add_current_iterators): Likewise.
38508 (apply_iterators): Likewise.
38509 (initialize_iterators): Guard usage of apply_subst_iterator with
38510 #ifdef GENERATOR_FILE.
38511 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
38512 (md_reader::read_mapping): Likewise.
38513 (add_define_attr_for_define_subst): Likewise.
38514 (add_define_subst_attr): Likewise.
38515 (read_subst_mapping): Likewise.
38516 (check_code_iterator): Likewise.
38517 (rtx_reader::read_rtx): Likewise. Move one-time initialization
38519 (one_time_initialization): New function.
38520 (rtx_reader::read_until): New method.
38521 (read_flags): New function.
38522 (parse_reg_note_name): New function.
38523 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
38524 Handle reuse_rtx ids.
38525 Wrap iterator lookup within #ifdef GENERATOR_FILE.
38526 Add parsing support for RTL dumps, mirroring the special-cases in
38527 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
38528 values, and calling handle_any_trailing_information.
38529 (rtx_reader::read_rtx_operand): Convert return type from void
38530 to rtx, returning return_rtx. Handle case 'e'. Call
38531 finalize_string on XSTR and XTMPL fields.
38532 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
38533 "(nil)" values were omitted. Call the postprocess vfunc on the
38535 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
38536 class ctor. Initialize m_in_call_function_usage. Call
38537 one_time_initialization.
38538 * rtl-tests.c (selftest::test_uncond_jump): Call
38539 set_new_first_and_last_insn.
38540 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
38541 * selftest-rtl.c: New file.
38542 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
38543 (selftest::get_insn_by_uid): New decl.
38544 * selftest-run-tests.c (selftest::run_tests): Call
38545 read_rtl_function_c_tests.
38546 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
38547 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
38550 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
38552 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
38553 operands in a special way. Assert that pos+len <= mode precision.
38555 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38557 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
38558 3 argument Alias with unlimited for the negative form.
38559 (fno-vect-cost-model): Removed.
38561 2017-01-05 Martin Liska <mliska@suse.cz>
38563 * hsa-gen.c (gen_hsa_divmod): New function.
38564 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
38566 2017-01-05 Martin Liska <mliska@suse.cz>
38569 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
38572 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38574 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
38575 small constant length operands.
38577 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38579 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
38580 between loop iterations.
38582 2017-01-05 Martin Liska <mliska@suse.cz>
38585 * gimplify.c (gimplify_decl_expr): Compare to
38586 asan_poisoned_variables instread of checking flags.
38587 (gimplify_target_expr): Likewise.
38588 (gimplify_expr): Likewise.
38589 (gimplify_function_tree): Conditionally initialize
38590 asan_poisoned_variables.
38592 2017-01-04 Jeff Law <law@redhat.com>
38594 PR tree-optimizatin/78812
38595 * rtl.h (contains_mem_rtx_p): Prototype.
38596 * ifcvt.c (containts_mem_rtx_p): Move from here to...
38597 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
38598 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
38599 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
38600 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
38602 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38604 * input.c (assert_char_at_range): Default-initialize actual_range.
38606 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38608 * df-scan.c (df_ref_create_structure): Make regno unsigned,
38609 to match the caller.
38611 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38613 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
38614 insns after final jump in test to emit dummy move.
38616 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38618 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
38619 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
38621 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38623 * multiple_target.c (create_dispatcher_calls): Init e_next.
38624 * tree-ssa-loop-split.c (split_loop): Init border.
38625 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
38628 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
38633 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
38634 (altivec_register_operand): Do not return true if the operand
38635 contains a SUBREG mixing SImode and SFmode.
38636 (vsx_register_operand): Likewise.
38637 (vsx_reg_sfsubreg_ok): New predicate.
38638 (vfloat_operand): Do not return true if the operand contains a
38639 SUBREG mixing SImode and SFmode.
38640 (vint_operand): Likewise.
38641 (vlogical_operand): Likewise.
38642 (gpc_reg_operand): Likewise.
38643 (int_reg_operand): Likewise.
38644 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
38645 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
38646 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
38648 (rs6000_emit_move_si_sf_subreg): New helper function.
38649 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
38650 fixup SUBREGs involving SImode and SFmode.
38651 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
38652 numbers for the new peephole2 optimization.
38653 (peephole2 for SFmode unions): New peephole2 to optimize cases in
38654 the GLIBC math library that do AND/IOR/XOR operations on single
38655 precision floating point.
38656 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
38657 target macros to say whether we need to avoid SUBREGs mixing
38659 (TARGET_ALLOW_SF_SUBREG): Likewise.
38660 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
38661 (UNSPEC_SI_FROM_SF): Likewise.
38662 (iorxor): Change spacing.
38663 (and_ior_xor): New iterator for AND, IOR, and XOR.
38664 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
38665 (movdi_from_sf_zero_ext): Likewise.
38666 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
38667 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
38668 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
38669 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
38670 (fms<mode>4): Likewise.
38671 (fnma<mode>4): Likewise.
38672 (fnms<mode>4): Likewise.
38673 (nfma<mode>4): Likewise.
38674 (nfms<mode>4): Likewise.
38676 2017-01-04 Marek Polacek <polacek@redhat.com>
38679 * doc/invoke.texi: Document -Wpointer-compare.
38681 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38683 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
38686 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
38687 descriptions for -gdwarf-5 and emit them as uleb128 instead of
38690 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
38693 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
38694 documentation of the powerpc_popcntb_ok attribute.
38695 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
38696 code to issue warning messages if a requested CPU configuration is
38697 not supported by the binary (assembler and loader) toolchain.
38698 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
38699 made to define a built-in function that has been disabled.
38700 (paired_init_builtins): Add assertion to prevent ICE if attempt is
38701 made to define a built-in function that has been disabled.
38702 (altivec_init_builtins): Add comment explaining why definition
38703 of the DST built-in functions is not preceded by an assertion
38704 check. Add assertions to prevent ICE if attempts are made to
38705 define an altivec predicate or an abs* built-in function that has
38707 (htm_init_builtins): Add comment explaining why definition of the
38708 htm built-in functions is not preceded by an assertion check.
38710 2017-01-04 Jeff Law <law@redhat.com>
38712 PR tree-optimizatin/67955
38713 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
38714 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
38715 the points-to solution does not include pt_null. Use DECL_PT_UID
38718 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
38720 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
38721 Use gen_int_mode instead of gen_lopwart for const_int operands.
38723 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38725 PR tree-optimization/71563
38726 * match.pd: Simplify X << Y into X if Y is known to be 0 or
38727 out of range value - has low bits known to be zero.
38729 2017-01-04 Alan Modra <amodra@gmail.com>
38731 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
38732 * configure: Regenerate.
38733 * config.in: Regenerate.
38735 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38738 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
38739 a substring of the message, but strcmp with the whole message. Ifdef
38740 ENABLE_NLS, translate the message first using dgettext.
38742 2017-01-03 Jeff Law <law@redhat.com>
38744 PR tree-optimizatin/78856
38745 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
38746 (mark_threaded_blocks): Remove code to truncate thread paths that
38747 cross multiple loop headers. Instead invalidate the cached loop
38748 iteration information and handle case of a thread path walking
38749 into an irreducible region.
38751 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
38754 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
38755 assertions. Add support for doing the signbit if the IEEE 128-bit
38756 floating point value is in a GPR.
38757 * config/rs6000/rs6000.md (Fsignbit): Delete.
38758 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
38759 Update the length attribute if the value is in a GPR.
38760 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
38761 the sign or zero extension instruction, since the value is always 0/1.
38762 (signbit<mode>2_dm2): Delete using <Fsignbit>.
38765 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
38766 extracting SImode to a GPR register so that we can generate a
38767 store, limit the vector to be in a traditional Altivec register
38768 for the vextuwrx instruction.
38770 2017-01-03 Ian Lance Taylor <iant@google.com>
38772 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
38774 2017-01-03 Martin Sebor <msebor@redhat.com>
38776 PR tree-optimization/78696
38777 * gimple-ssa-sprintf.c (format_floating): Correct handling of
38778 precision. Use MPFR for %f for greater fidelity. Correct handling
38780 (pass_sprintf_length::compute_format_length): Set width and precision
38781 specified by asrerisk to void_node for vararg functions.
38782 (try_substitute_return_value): Adjust dump output.
38784 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
38786 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
38788 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
38790 * doc/invoke.texi (SPARC options): Document -mlra as the default.
38791 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
38792 -mlra/-mno-lra was passed to the compiler.
38794 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
38796 PR rtl-optimization/65618
38797 * emit-rtl.c (try_split): Move initialization of "before" and
38798 "after" to just before the call to emit_insn_after_setloc.
38800 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
38802 * doc/md.texi (Standard Names): Remove reference to Java frontend.
38804 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
38806 * dwarf2out.c (gen_enumeration_type_die): When
38807 -gno-strict-dwarf, add a DW_AT_encoding attribute.
38809 2017-01-03 Jakub Jelinek <jakub@redhat.com>
38811 PR tree-optimization/78965
38812 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
38813 Change first argument from const call_info & to call_info &. For %n
38814 set info.nowrite to false.
38816 PR middle-end/78901
38817 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
38818 possibly throwing calls.
38820 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
38821 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
38822 and fns handling, rather than in a separate case SSA_NAME.
38824 2017-01-02 Jeff Law <law@redhat.com>
38826 * config/darwin-driver.c (darwin_driver_init): Const-correctness
38827 fixes for first_period and second_period variables.
38829 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
38832 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
38833 (*insvqi_1): New insn pattern.
38834 (*insvqi_1_mem_rex64): Ditto.
38835 (*insvqi_2): Ditto.
38836 (*insvqi_3): Rename from *insvqi.
38838 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
38840 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
38842 * doc/cfg.texi (Edges): Remove reference to Java.
38843 (Maintaining the CFG): Ditto.
38845 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38847 PR middle-end/77674
38848 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
38849 transparent aliases.
38851 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38853 PR middle-end/77484
38854 * predict.def (PRED_CALL): Update hitrate.
38855 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
38856 * predict.c (tree_estimate_probability_bb): Split CALL predictor
38857 into direct/indirect/polymorphic variants.
38859 2017-01-01 Jakub Jelinek <jakub@redhat.com>
38861 Update copyright years.
38863 * gcc.c (process_command): Update copyright notice dates.
38864 * gcov-dump.c (print_version): Ditto.
38865 * gcov.c (print_version): Ditto.
38866 * gcov-tool.c (print_version): Ditto.
38867 * gengtype.c (create_file): Ditto.
38868 * doc/cpp.texi: Bump @copying's copyright year.
38869 * doc/cppinternals.texi: Ditto.
38870 * doc/gcc.texi: Ditto.
38871 * doc/gccint.texi: Ditto.
38872 * doc/gcov.texi: Ditto.
38873 * doc/install.texi: Ditto.
38874 * doc/invoke.texi: Ditto.
38876 Copyright (C) 2017 Free Software Foundation, Inc.
38878 Copying and distribution of this file, with or without modification,
38879 are permitted in any medium without royalty provided the copyright
38880 notice and this notice are preserved.