1 2017-12-12 Jeff Law <law@redhat.com>
3 PR tree-optimization/83298
4 PR tree-optimization/83362
5 PR tree-optimization/83383
6 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
7 push_value_range a public interface. Add new argument to
8 record_ranges_from_stmt.
9 * gimple-ssa-evrp-analyze.c
10 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
11 Update comments. Handle recording temporary equivalences.
12 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
13 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
14 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
15 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
16 gimple-ssa-evrp-analyze.h.
17 (record_temporary_equivalences_from_phis): Add new argument. When
18 the PHI arg is an SSA_NAME, set the result's range to the range
20 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
22 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
23 Pass it down to children as needed.
24 (thread_outgoing_edges): Likewise.
25 (thread_across_edge): Likewise. Push/pop range state as needed.
26 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
28 2017-12-12 Julia Koval <julia.koval@intel.com>
30 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
31 (PTA_CANNONLAKE): Remove PTA_CLWB.
33 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
37 (vcondv2div2df): Likewise.
38 (vconduv2dfv2di): Likewise.
40 2017-12-12 Jakub Jelinek <jakub@redhat.com>
42 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
43 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
44 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
45 ATTR_MATHFN_FPROUNDING_ERRNO.
47 2017-12-12 Richard Biener <rguenther@suse.de>
49 PR tree-optimization/83385
50 * graphite-scop-detection.c (get_order, order): Remove.
51 (bb_to_rpo): New global.
53 (build_scops): Sort pbbs in RPO order.
55 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
57 * combine.c (simplify_set): Do not transform subregs to zero_extends
58 if the destination mode is a vector mode.
60 2017-12-12 Jakub Jelinek <jakub@redhat.com>
62 PR tree-optimization/83359
63 * tree-cfg.h (fold_loop_internal_call): Declare.
64 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
65 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
66 (find_loop_dist_alias): New function.
67 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
68 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
71 PR tree-optimization/80631
72 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
73 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
74 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
75 hardcoding zero as the value if COND_EXPR is never true. For
76 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
77 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
78 hardcoding MAX_EXPR as the reduction operation.
79 (is_nonwrapping_integer_induction): Allow negative step.
80 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
81 vect_create_epilog_for_reduction, if no value is suitable, don't
82 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
84 2017-12-12 Richard Biener <rguenther@suse.de>
86 PR tree-optimization/81889
87 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
88 range info from the non-wrapping IV instead of just the range
91 2017-12-12 Julia Koval <julia.koval@intel.com>
93 * config.gcc: Add vaesintrin.h.
94 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
96 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
97 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
99 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
100 * config/i386/immintrin.h: Include vaesintrin.h.
101 * config/i386/sse.md (vaesdec_<mode>): New pattern.
102 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
103 _mm_aesdec_epi128): New intrinsics.
105 2017-12-12 Julia Koval <julia.koval@intel.com>
107 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
108 OPTION_MASK_ISA_VAES_UNSET): New.
109 (ix86_handle_option): Handle -mvaes.
110 * config/i386/cpuid.h: Define bit_VAES.
111 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
112 * config/i386/i386-c.c (__VAES__): New.
113 * config/i386/i386.c (ix86_target_string): Add -mvaes.
114 (ix86_valid_target_attribute_inner_p): Ditto.
115 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
116 * config/i386/i386.opt: Add -mvaes.
117 * doc/invoke.texi: Ditto.
119 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
121 * debug.h (gcc_debug_hooks): Add inline_entry.
122 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
123 * debug.c (do_nothing_debug_hooks): Likewise.
124 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
125 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
126 (dwarf2_lineno_debug_hooks): Likewise.
128 * common.opt (gstatement-frontiers): New, setting
129 debug_nonbind_markers_p.
130 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
131 * toplev.c (process_options): Autodetect value for debug statement
133 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
134 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
136 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
137 markers. Integrate source bind into debug stmt expand loop.
138 (pass_expand::execute): Check debug marker limit. Avoid deep
139 TER and expand debug locations for debug bind insns only.
140 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
142 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
143 * final.c (reemit_insn_block_notes): Take current block from
144 nonbind markers. Declare note where it's first set.
145 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
146 begin stmt markers if enabled.
147 (notice_source_line): Handle nonbind markers. Fail if their
148 location is unknown or that of builtins.
149 (rest_of_handle_final): Convert begin stmt markers to notes if
150 var-tracking didn't run.
151 (rest_of_clean_state): Skip begin stmt markers.
152 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
154 * function.c (allocate_struct_function): Set begin_stmt_markers.
155 * function.h (struct function): Add debug_marker_count counter
156 and debug_nonbind_markers flag.
157 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
158 * gimple-low.c (lower_function_body): Adjust
159 debug_nonbind_markers.
160 (lower_stmt): Drop or skip gimple debug stmts.
161 (lower_try_catch): Skip debug stmts.
162 * gimple.c (gimple_build_debug_begin_stmt): New.
163 (gimple_copy): Increment debug_marker_count if copying one.
164 * gimple.h (gimple_build_debug_begin_stmt): Declare.
165 * gimplify.c (rexpr_location): New.
166 (rexpr_has_location): New.
167 (warn_switch_unreachable_r): Handle gimple debug stmts.
168 (shortcut_cond_r): Call expr_location.
170 (find_goto_label): New.
171 (shortcut_cond_expr): Call expr_has_location, expr_location, and
173 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
175 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
176 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
177 (LANG_HOOKS_INITIALIZER): ... this.
178 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
179 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
181 (update_ebb_live_info): Skip debug insn markers.
182 * lra.c (debug_insn_static_data): Rename to...
183 (debug_bind_static_data): ... this.
184 (debug_marker_static_data): New.
185 (lra_set_insn_recog_data): Select one of the above depending
187 (lra_update_isn_regno_info): Don't assume debug insns have
189 (push_insns): Skip debug insns.
190 * lto-streamer-in.c (input_function): Drop debug stmts
191 depending on active options. Adjust debug_nonbind_markers.
192 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
193 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
194 begin stmt marker notes.
195 (print_insn): Likewise.
196 * recog.c (extract_insn): Recognize rtl for debug markers.
197 * rtl.def (DEBUG_MARKER): New.
198 * tree-inline.c: Include params.h.
199 (remap_gimple_stmt): Handle nonbind markers.
200 (maybe_move_debug_stmts_to_successors): Likewise.
201 (copy_debug_stmt): Likewise.
202 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
203 markers regardless of no side effects.
204 (tsi_link_before): Don't update container's side effects when adding
206 (tsi_link_after): Likewise.
207 (expr_first): Skip begin stmt markers.
208 (expr_last): Likewise.
209 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
210 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
211 Disregard nonbind markers.
212 * tree.c (make_node_stat): Don't set side effects for begin stmt
214 (build1_stat): Likewise.
215 * tree.def (DEBUG_BEGIN_STMT): New.
216 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
217 * var-tracking.c (delete_debug_insns): Renamed to...
218 (delete_vta_debug_insns): ... this.
219 (reemit_marker_as_note): New.
220 (vt_initialize): Reemit markers.
221 (delete_vta_debug_insns): Likewise.
222 (vt_debug_insns_local): Reemit or delete markers.
223 (variable_tracking_main_1): Likewise.
224 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
225 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
226 (gimple_debug_nonbind_marker_p): New.
227 (gimple_build_debug_bind): Adjust.
228 (gimple_build_debug_begin_stmt): New.
229 * doc/invoke.texi (max-debug-marker-count): New param.
230 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
231 (debug_parameter_ref, debug_marker): New.
232 (NOTE_INSN_BEGIN_STMT): New.
233 (DEBUG_INSN): Describe begin stmt markers.
235 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
236 without debug insns, we wouldn't, but clean up debug insns
237 after a control flow insn nevertheless.
239 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
240 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
241 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
242 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
243 (rtl_tidy_fallthru_edge): Likewise.
244 (rtl_verify_fallthru): Likewise.
245 (rtl_verify_bb_layout): Likewise.
246 (skip_insns_after_block): Likewise.
247 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
248 * dwarf2out.c: Include print-rtl.h.
249 (dwarf2out_next_real_insn): New.
250 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
251 Dump debug binds in asm comments.
252 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
253 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
254 callers to use gsi_start_nondebug_bb instead.
255 (gsi_after_labels): Skip gimple debug stmts.
256 (gsi_start_nondebug): New.
257 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
258 (proper_loop_form_for_interchange): Adjust.
259 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
260 * gimple.h (gimple_seq_last_nondebug_stmt): New.
261 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
262 (collect_fallthrough_labels): Likewise.
263 (should_warn_for_implicit_fallthrough): Likewise.
264 (warn_implicit_fallthrough_r): Likewise.
265 (expand_FALLTHROUGH_r): Likewise.
266 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
267 (graphite_copy_stmts_from_block): Skip nonbind markers.
268 * haifa-sched.c (sched_extend_bb): Skip debug insns.
269 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
270 * jump.c (clean_barriers): Skip debug insns.
271 * omp-expand.c (expand_parallel_call): Skip debug insns.
272 (expand_task_call): Likewise.
273 (remove_exit_barrier): Likewise.
274 (expand_omp_taskreg): Likewise.
275 (expand_omp_for_init_counts): Likewise.
276 (expand_omp_for_generic): Likewise.
277 (expand_omp_for_static_nochunk): Likewise.
278 (expand_omp_for_static_chunk): Likewise.
279 (expand_omp_simd): Likewise.
280 (expand_omp_taskloop_for_outer): Likewise.
281 (expand_omp_taskloop_for_inner): Likewise.
282 (expand_oacc_for): Likewise.
283 (expand_omp_sections): Likewise.
284 (expand_omp_single): Likewise.
285 (expand_omp_synch): Likewise.
286 (expand_omp_atomic_load): Likewise.
287 (expand_omp_atomic_store): Likewise.
288 (expand_omp_atomic_fetch_op): Likewise.
289 (expand_omp_atomic_pipeline): Likewise.
290 (expand_omp_atomic_mutex): Likewise.
291 (expand_omp_target): Likewise.
292 (grid_expand_omp_for_loop): Likewise.
293 (grid_expand_target_grid_body): Likewise.
294 (build_omp_regions_1): Likewise.
295 * omp-low.c (check_combined_parallel): Skip debug stmts.
296 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
297 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
298 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
300 * tree-cfg.c (make_blobs_1): Skip debug stmts.
301 (make_edges): Likewise.
302 (cleanup_dead_labels): Likewise.
303 (gimple_can_merge_blocks_p): Likewise.
304 (stmt_starts_bb_p): Likewise.
305 (gimple_block_label): Likewise.
306 (gimple_redirect_edge_and_branch): Likewise.
307 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
309 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
311 * tree-pretty-print (print_declaration): Omit initializer in slim
313 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
315 (eliminate_unnecessary_stmts): Stabilize block removal order.
316 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
317 * var-tracking.c (get_first_insn): New.
318 (vt_emit_notes): Call it.
319 (vt_initialize): Walk any insns before the first BB.
320 (delete_debug_insns): Likewise.
322 * gimple.h (enum gimple_debug_subcode): Add
323 GIMPLE_DEBUG_BEGIN_STMT.
324 (gimple_debug_begin_stmt_p): New.
325 (gimple_debug_nonbind_marker_p): New.
326 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
327 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
328 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
329 * insn-notes.def (BEGIN_STMT): New.
330 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
331 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
332 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
333 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
334 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
335 (INSN_DEBUG_MARKER_KIND): New.
336 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
337 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
338 (INSN_VAR_LOCATION_PTR): New.
339 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
341 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
342 deps only if debug bind insns are enabled.
343 (pass_expand::execute): Avoid deep TER and expand
344 debug locations for debug bind insns only.
345 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
346 debug stmts special handling down to debug bind stmts.
347 * combine.c (try_combine): Narrow debug insns special handling
348 down to debug bind insns.
349 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
350 Narrow debug insns preexisting special handling down to debug
352 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
353 handling down to debug bind insns.
354 * function.c (instantiate_virtual_regs): Skip debug markers,
355 adjust handling of debug binds.
356 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
357 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
358 * haifa-sched.c (schedule_insn): Narrow special handling of debug
359 insns to debug bind insns.
360 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
361 special handling of debug stmts to debug bind stmts.
362 * ipa-split.c (split_function): Likewise.
363 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
364 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
366 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
367 * regrename.c (build_def_use): Likewise.
368 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
369 (pass_cprop_hardreg): Narrow special casing of debug insns to
371 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
372 * reload1.c (reload): Likewise.
373 * sese.c (sese_insert_phis_for_liveouts): Narrow special
374 casing of debug stmts to debug bind stmts.
375 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
376 * ssa-iterators.h (num_imm_uses): Likewise.
377 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
378 debug stmts to debug bind stmts.
379 * tree-inline.c (tree_function_versioning): Narrow special casing
380 of debug stmts to debug bind stmts.
381 * tree-loop-distribution.c (generate_loops_for_partition):
382 Narrow special casing of debug stmts to debug bind stmts.
383 * tree-sra.c (analyze_access_subtree): Narrow special casing
384 of debug stmts to debug bind stmts.
385 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
386 stmts to debug bind stmts.
387 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
388 casing of debug stmts to debug bind stmts.
389 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
390 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
391 casing of debug stmts to debug bind stmts.
392 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
394 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
395 debug stmts to debug bind stmts.
396 (gimple_replace_ssa_lhs): Likewise.
397 (insert_debug_temp_for_var_def): Likewise.
398 (insert_debug_temps_for_defs): Likewise.
399 (reset_debug_uses): Likewise.
400 * tree-ssanames.c (release_ssa_name_fn): Likewise.
401 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
402 (adjust_debug_stmts): Likewise.
403 (adjust_phi_and_debug_stmts): Likewise.
404 (vect_do_peeling): Likewise.
405 * tree-vect-loop.c (vect_transform_loop): Likewise.
406 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
407 * var-tracking.c (adjust_mems): Narrow special casing of debug
408 insns to debug bind insns.
409 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
410 (compute_bb_dataflow, vt_find_locations): Likewise.
411 (vt_expand_loc, emit_notes_for_changes): Likewise.
412 (vt_init_cfa_base): Likewise.
413 (vt_emit_notes): Likewise.
414 (vt_initialize): Likewise.
415 (vt_finalize): Likewise.
417 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
418 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
419 (next_nonnote_nondebug_insn_bb): New.
420 (prev_nonnote_nondebug_insn_bb): New.
421 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
422 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
423 (prev_nonnote_nondebug_insn_bb): Declare.
424 (next_nonnote_nondebug_insn_bb): Declare.
425 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
426 * cfgrtl.c (get_last_bb_insn): Likewise.
427 * lra.c (push_insns): Likewise.
429 2017-12-11 David Malcolm <dmalcolm@redhat.com>
432 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
434 * selftest.c (selftest::test_runner::test_runner): New ctor.
435 (selftest::test_runner::~test_runner): New dtor.
436 * selftest.h (class selftest::test_runner): New class.
438 2017-12-11 Carl Love <cel@us.ibm.com>
440 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
441 vec_extract_fp32_from_shortl]): Add #defines.
442 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
443 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
444 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
445 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
446 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
447 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
448 * doc/extend.texi: Add documentation for the added builtins.
450 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
452 PR rtl-optimization/80693
453 PR rtl-optimization/81019
454 PR rtl-optimization/81020
455 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
456 are not mentioned in i3. Place the REG_UNUSED note on i2,
457 possibly modified to REG_DEAD, if it did not originate in i3.
459 2017-12-11 Jakub Jelinek <jakub@redhat.com>
461 * recog.c (store_data_bypass_p_1): New function.
462 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
463 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
465 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
467 PR rtl-optimization/83361
468 * ifcvt.c (if_convert): Call fixup_partitions.
470 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
472 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
473 early folding of splat_u{8,16,32}.
475 2017-12-11 Jakub Jelinek <jakub@redhat.com>
477 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
478 output_operand_lossage first argument with capital letter.
479 (aarch64_override_options): Don't start error and sorry first argument
482 2017-12-11 Andi Kleen <ak@linux.intel.com>
484 PR gcov-profile/83355
485 * auto-profile.c (string_table::get_index_by_decl): Don't
486 recurse when abstract origin points to itself.
488 2017-12-11 Bin Cheng <bin.cheng@arm.com>
490 PR tree-optimization/83320
491 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
492 (prune_datarefs_not_in_loop): Ditto.
494 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
496 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
498 2017-12-10 Jakub Jelinek <jakub@redhat.com>
500 PR tree-optimization/83337
501 * gimple-loop-interchange.cc (compute_access_stride): Handle
502 bitfield DRs properly.
504 2017-12-09 Jakub Jelinek <jakub@redhat.com>
506 PR tree-optimization/83338
507 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
508 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
509 vector of unsigned integers to vector of signed integers.
511 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
513 PR rtl-optimization/83317
514 * lra-constraints.c (process_address_1): Add insn code check.
516 2017-12-08 Michael Matz <matz@suse.de>
518 Fix PR tree-optimization/83323
519 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
520 head-controlled loops and loop BBs.
521 * common.opt (funroll-and-jam): Remove, instead ...
522 (floop-unroll-and-jam): ... reuse this option.
523 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
524 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
525 (-floop-unroll-and-jam): ... this option.
527 2017-12-08 Jakub Jelinek <jakub@redhat.com>
529 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
530 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
531 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
533 2017-12-08 Richard Biener <rguenther@suse.de>
536 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
537 handle accesses outside of zero-sized vars.
539 2017-12-08 Martin Jambor <mjambor@suse.cz>
541 PR tree-optimization/83141
542 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
543 test for MEM_REFs implicitely changing types with padding. Remove
545 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
547 2017-12-08 Jakub Jelinek <jakub@redhat.com>
549 * config/arc/arc.c (arc_attribute_table): Add exclusions to
551 * config/avr/avr.c (avr_attribute_table): Likewise.
552 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
553 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
554 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
555 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
556 of struct attribute_spec.
557 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
559 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
562 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
564 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
567 * profile-count.c (profile_count::from_gcov_type): Move from
568 profile-count.h; handle overflow.
569 * profile-count.h (profile_count::from_gcov_type): Move offline.
571 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
573 PR rtl-optimization/83304
574 * combine.c (move_deaths): If we do not know where a register died,
577 2017-12-08 Richard Biener <rguenther@suse.de>
579 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
580 Provide -fopt-info-loop feedback when we interchange in a nest.
582 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
584 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
585 for armv6 ARM CPU IDs.
587 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
589 * common/config/arm/arm-common.c: Include <algorithm>.
590 (INCLUDE_VECTOR): Define.
591 (compare_opt_names): New function.
592 (arm_rewrite_selected_arch): Only strip out extensions that can be
593 expressed through -mfpu. Sort the remaining extensions
596 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
598 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
599 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
600 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
601 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
603 2017-06-08 Tristan Gingold <gindold@adacore.com>
606 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
607 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
609 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
611 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
612 of struct attribute_spec.
614 2017-12-08 Julia Koval <julia.koval@intel.com>
616 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
617 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
618 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
619 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
620 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
621 _mm_maskz_dpwssds_epi32): Ditto.
623 2017-12-08 Richard Biener <rguenther@suse.de>
625 PR tree-optimization/81303
626 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
627 conditions try to create a comparison vector type matching
628 the data vector type.
629 (vectorizable_condition): Adjust.
630 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
631 Leave invariant conditions alone in case we can vectorize those.
633 2017-12-08 Julia Koval <julia.koval@intel.com>
635 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
636 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
637 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
638 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
639 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
641 2017-12-08 Julia Koval <julia.koval@intel.com>
643 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
644 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
645 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
646 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
647 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
648 _mm_maskz_dpbusds_epi32): New intrinsics.
650 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
652 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
654 (ld<bh>io_signed): Likewise.
655 (st<bhw>io): Likewise.
656 * config/nios2/predicates.md (ldstio_memory_operand): Allow
657 SMALL_INT12 constant integer operand.
659 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
661 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
663 (nios2_large_constant_allowed): ...this. Adjust uses.
664 (nios2_plus_symbolic_constant_p): Rename to...
665 (nios2_plus_large_constant_p): ...this. Adjust uses.
666 (nios2_legitimate_address_p): Correct CONST_INT handling.
667 (nios2_symbolic_memory_operand_p): Rename to...
668 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
669 (nios2_large_constant_p): Check for large constant integers too.
670 (nios2_split_large_constant): Handle constant integers.
671 (nios2_split_symbolic_memory_operand): Rename to...
672 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
673 (nios2_legitimize_constant_address): Handle constant integers.
674 (r0rel_constant_p): Handle small constant integers.
675 (nios2_print_operand_address): Handle r0-relative integer addresses.
676 * config/nios2/nios2-protos.h: Adjust for renamed functions.
677 * config/nios2/nios2.md: Adjust for renamed functions.
679 2017-12-07 Andrew Waterman <andrew@sifive.com>
681 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
682 (TARGET_HAVE_SRODATA_SECTION): New define.
683 (riscv_select_section): New function.
685 2017-12-08 Joseph Myers <joseph@codesourcery.com>
686 Alexander Monakov <amonakov@ispras.ru>
687 Jakub Jelinek <jakub@redhat.com>
690 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
692 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
694 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
695 a bool success value. Don't call output_operand_lossage here.
696 (aarch64_print_ldpstp_address): Return a bool success value.
697 (aarch64_print_operand_address): Call output_addr_const if
698 aarch64_print_address_internal fails.
699 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
700 'y'; call output_operand_lossage instead. Call output_operand_lossage
701 if aarch64_print_ldpstp_address fails.
703 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
705 * tree-vector-builder.h
706 (tree_vector_builder::binary_encoded_nelts): Declare.
707 * tree-vector-builder.c
708 (tree_vector_builder::binary_encoded_nelts): New function.
709 * fold-const.c (negate_expr_p): Likewise.
710 (operand_equal_p, fold_checksum_tree): Likewise.
711 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
712 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
713 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
714 (uniform_vector_p): Likewise.
715 * varasm.c (const_hash_1, compare_constant): Likewise.
716 * tree-ssa-ccp.c: Include tree-vector-builder.h.
717 (valid_lattice_transition): Operate directly on the VECTOR_CST
719 * ipa-icf.c: Include tree-vector-builder.h.
720 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
721 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
723 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
725 * tree.c (build_vector): Delete.
726 * tree.h (build_vector): Make static and move into the self-testing
729 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
731 * vector-builder.h (vector_builder::derived): New const overload.
732 (vector_builder::elt): New function.
733 * tree-vector-builder.h (tree_vector_builder::type): New function.
734 (tree_vector_builder::apply_step): Declare.
735 * tree-vector-builder.c (tree_vector_builder::apply_step): New
737 * gimple-fold.h (tree_vector_builder): Declare.
738 (gimple_build_vector): Take a tree_vector_builder instead of a
739 type and vector of elements.
740 * gimple-fold.c (gimple_build_vector): Likewise.
741 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
743 (get_initial_defs_for_reduction): Likewise.
744 (vectorizable_induction): Likewise.
746 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
748 * tree-vector-builder.h
749 (tree_vector_builder::new_binary_operation): Declare.
750 * tree-vector-builder.c
751 (tree_vector_builder::new_binary_operation): New function.
752 * fold-const.c (fold_relational_const): Use it.
753 (const_binop): Likewise. Check that both input vectors have
754 the same number of elements, thus excluding things like WIDEN_SUM.
755 Check whether it is possible to operate directly on the encodings
758 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
760 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
761 new_unary_operation, operating only on the encoded elements.
762 (const_unop): Likewise.
763 (exact_inverse): Likewise.
764 (distributes_over_addition_p): New function.
765 (const_binop): Use tree_vector_builder and new_unary_operation
766 for combinations of VECTOR_CST and INTEGER_CST. Operate only
767 on the encoded elements unless the encoding is strided and the
768 operation does not distribute over addition.
769 (fold_convert_const): Use tree_vector_builder and
770 new_unary_operation. Operate only on the encoded elements
771 for truncating integer conversions, or for non-stepped encodings.
773 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
775 * config/sparc/sparc.c: Include tree-vector-builder.h.
776 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
777 * expmed.c: Include tree-vector-builder.h.
778 (make_tree): Use tree_vector_builder instead of build_vector.
779 * fold-const.c: Include tree-vector-builder.h.
780 (const_binop): Use tree_vector_builder instead of build_vector.
781 (const_unop): Likewise.
782 (native_interpret_vector): Likewise.
783 (fold_vec_perm): Likewise.
784 (fold_ternary_loc): Likewise.
785 * gimple-fold.c: Include tree-vector-builder.h.
786 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
788 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
789 (simplify_vector_constructor): Use tree_vector_builder instead
791 * tree-vect-generic.c: Include tree-vector-builder.h.
792 (add_rshift): Use tree_vector_builder instead of build_vector.
793 (expand_vector_divmod): Likewise.
794 (optimize_vector_constructor): Likewise.
795 * tree-vect-loop.c: Include tree-vector-builder.h.
796 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
797 of build_vector. Explicitly use a stepped encoding for
799 * tree-vect-slp.c: Include tree-vector-builder.h.
800 (vect_get_constant_vectors): Use tree_vector_builder instead
802 (vect_transform_slp_perm_load): Likewise.
803 (vect_schedule_slp_instance): Likewise.
804 * tree-vect-stmts.c: Include tree-vector-builder.h.
805 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
806 (vect_gen_perm_mask_any): Likewise.
807 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
808 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
810 (build_vector_from_val): Likewise. Explicitly use a duplicate
813 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
815 * doc/generic.texi (VECTOR_CST): Describe new representation of
817 * vector-builder.h: New file.
818 * tree-vector-builder.h: Likewise.
819 * tree-vector-builder.c: Likewise.
820 * Makefile.in (OBJS): Add tree-vector-builder.o.
821 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
822 * tree-core.h (tree_base): Add a vector_cst field to the u union.
823 (tree_vector): Change the number of elements to
824 vector_cst_encoded_nelts.
825 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
826 (VECTOR_CST_ELTS): Delete.
827 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
828 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
829 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
830 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
831 (VECTOR_CST_ENCODED_ELT): Likewise.
832 (vector_cst_encoded_nelts): New function.
833 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
834 VECTOR_CST_NELTS_PER_PATTERN as arguments.
835 (vector_cst_int_elt, vector_cst_elt): Declare.
836 * tree.c: Include tree-vector-builder.h.
837 (tree_code_size): Abort if passed VECTOR_CST.
838 (tree_size): Update for new VECTOR_CST layout.
839 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
840 VECTOR_CST_NELTS_PER_PATTERN as arguments.
841 (build_vector): Use tree_vector_builder.
842 (vector_cst_int_elt, vector_cst_elt): New functions.
843 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
844 encoded elements and then create the vector in the canonical form.
845 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
846 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
847 (tree_c_tests): Call test_vector_cst_patterns.
848 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
850 (hash_tree): Likewise.
851 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
852 (streamer_write_tree_header): Likewise.
853 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
854 (streamer_alloc_tree): Likewise. Update call to make_vector.
855 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
857 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
859 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
860 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
861 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
863 2017-12-07 Bin Cheng <bin.cheng@arm.com>
864 Richard Biener <rguenther@suse.de>
866 PR tree-optimization/81303
867 * Makefile.in (gimple-loop-interchange.o): New object file.
868 * common.opt (floop-interchange): Reuse the option from graphite.
869 * doc/invoke.texi (-floop-interchange): Ditto. New document for
870 -floop-interchange and mention it for -O3.
871 * opts.c (default_options_table): Enable -floop-interchange at -O3.
872 * gimple-loop-interchange.cc: New file.
873 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
874 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
875 * passes.def (pass_linterchange): New pass.
876 * timevar.def (TV_LINTERCHANGE): New time var.
877 * tree-pass.h (make_pass_linterchange): New declaration.
878 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
879 interchange. Record IV before/after increment in new parameters.
880 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
881 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
883 (check_reduction_path): ...New function here.
884 * tree-vectorizer.h (check_reduction_path): New declaration.
886 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
889 PR rtl-optimization/80818
890 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
891 always early clobbered.
892 * lra-lives.c (process_bb_lives): Check input hard regs for early
893 clobbered non-operand hard reg.
895 2017-12-07 Jakub Jelinek <jakub@redhat.com>
898 * tree-cfg.c (verify_gimple_assign_binary): Don't require
899 types_compatible_p, just that TYPE_MODE is the same.
901 2017-12-07 Martin Sebor <msebor@redhat.com>
904 * attribs.c (empty_attribute_table): Initialize new member of
905 struct attribute_spec.
906 (decl_attributes): Add argument. Handle mutually exclusive
907 combinations of attributes.
908 (selftests::test_attribute_exclusions): New function.
909 (selftests::attribute_c_tests): Ditto.
910 * attribs.h (decl_attributes): Add default argument.
911 * selftest.h (attribute_c_tests): Declare.
912 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
913 * tree-core.h (attribute_spec::exclusions, exclude): New type and
915 * doc/extend.texi (Common Function Attributes): Update const and pure.
916 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
917 of struct attribute_spec.
918 * config/arc/arc.c (arc_attribute_table): Same.
919 * config/arm/arm.c (arm_attribute_table): Same.
920 * config/avr/avr.c ( avr_attribute_table): Same.
921 * config/bfin/bfin.c (bfin_attribute_table): Same.
922 * config/cr16/cr16.c (cr16_attribute_table): Same.
923 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
924 * config/h8300/h8300.c (h8300_attribute_table): Same.
925 * config/i386/i386.c (ix86_attribute_table): Same.
926 * config/ia64/ia64.c (ia64_attribute_table): Same.
927 * config/m32c/m32c.c (m32c_attribute_table): Same.
928 * config/m32r/m32r.c (m32r_attribute_table): Same.
929 * config/m68k/m68k.c (m68k_attribute_table): Same.
930 * config/mcore/mcore.c (mcore_attribute_table): Same.
931 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
932 * config/mips/mips.c (mips_attribute_table): Same.
933 * config/msp430/msp430.c (msp430_attribute_table): Same.
934 * config/nds32/nds32.c (nds32_attribute_table): Same.
935 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
936 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
937 * config/rl78/rl78.c (rl78__attribute_table): Same.
938 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
939 * onfig/rx/rx.c (rx_attribute_table): Same.
940 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
941 * config/sh/sh.c (sh_attribute_table): Same.
942 * config/sparc/sparc.c (sparc_attribute_table): Same.
943 * config/spu/spu.c (spu_attribute_table): Same.
944 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
945 * config/v850/v850.c (v850_attribute_table): Same.
946 * config/visium/visium.c (visium_attribute_table): Same.
948 2017-12-07 Tamar Christina <tamar.christina@arm.com>
951 * config/arm/arm.c (INCLUDE_STRING): Define.
952 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
953 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
956 2017-12-07 Michael Matz <matz@suse.de>
958 Add unroll and jam pass
960 * gimple-loop-jam.c: New file.
961 * Makefile.in (OBJS): Add gimple-loop-jam.o.
962 * common.opt (funroll-and-jam): New option.
963 * opts.c (default_options_table): Add unroll-and-jam at -O3.
964 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
965 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
966 * passes.def: Add pass_loop_jam.
967 * timevar.def (TV_LOOP_JAM): Add.
968 * tree-pass.h (make_pass_loop_jam): Declare.
969 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
970 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
971 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
972 to flow_loop_tree_node_add.
973 (duplicate_subloops, copy_loops_to): Append to sibling list.
974 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
975 * doc/invoke.texi (-funroll-and-jam): Document new option.
976 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
978 2017-12-07 Richard Biener <rguenther@suse.de>
980 PR tree-optimization/83296
981 PR tree-optimization/67769
982 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
983 flow sensitive info in an unrelated BB.
984 (value_replacement): Use reset_flow_sensitive_info.
985 (minmax_replacement): Reset flow sensitive info on the def
986 we move. Do not reset flow sensitive info in the whole BB
988 (abs_replacement): Likewise.
990 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
993 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
994 rs6000_cpu to the given -mcpu=, or to the default processor.
996 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
998 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
999 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
1000 instead of rs6000_cpu_attr.
1001 (rs6000_adjust_cost): Ditto.
1002 (is_microcoded_insn): Ditto.
1003 (rs6000_adjust_priority): Ditto.
1004 (rs6000_issue_rate): Ditto.
1005 (rs6000_use_sched_lookahead): Ditto.
1006 (rs6000_use_sched_lookahead_guard): Ditto.
1007 (rs6000_sched_reorder): Ditto.
1008 (force_new_group): Ditto.
1009 * config/rs6000/rs6000.md (cpu attribute): Ditto.
1010 (group_ending_nop): Ditto.
1012 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1014 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
1015 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
1016 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
1017 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
1019 (rs6000_reassociation_width): Ditto.
1020 (rs6000_emit_epilogue): Ditto.
1021 (rs6000_adjust_cost): Ditto.
1022 (is_microcoded_insn): Ditto.
1023 (is_cracked_insn): Ditto.
1024 (rs6000_adjust_priority): Ditto.
1025 (rs6000_sched_reorder): Ditto.
1026 (rs6000_sched_reorder2): Ditto.
1027 (insn_must_be_first_in_group): Ditto.
1028 (insn_must_be_last_in_group): Ditto.
1029 (rs6000_register_move_cost): Ditto.
1030 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
1033 2017-12-07 Julia Koval <julia.koval@intel.com>
1035 * config.gcc: Add vaesintrin.h.
1036 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
1038 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
1039 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
1041 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1042 * config/i386/immintrin.h: Include vaesintrin.h.
1043 * config/i386/sse.md (vaesdec_<mode>): New pattern.
1044 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
1045 _mm_aesdec_epi128): New intrinsics.
1047 2017-12-06 David Malcolm <dmalcolm@redhat.com>
1049 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
1050 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
1051 c-family/c-spellcheck.cc.
1052 (best_macro_match::best_macro_match): Likewise.
1054 (struct edit_distance_traits<cpp_hashnode *>): Move to
1055 c-family/c-spellcheck.h.
1056 (class best_macro_match): Likewise.
1058 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1060 PR tree-optimization/83293
1061 * gimple-ssa-strength-reduction.c (insert_initializers): Use
1062 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
1063 might insert into empty bb.
1066 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
1067 simplify for plus with :c added, and pointer_plus without that.
1068 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
1069 with undefined overflow and the conversion is not widening,
1070 perform negation in utype and only convert to type afterwards.
1071 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
1072 simplify for plus with :c added, and pointer_plus without that.
1073 If type is integral with undefined overflow and the conversion is
1074 not widening, perform minus in utype and only convert to type
1075 afterwards. Move the last pointer_diff_expr simplify into the
1078 2017-12-06 Martin Sebor <msebor@redhat.com>
1080 PR tree-optimization/82646
1081 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
1082 strncpy, not maxlen.
1084 2017-12-06 Martin Sebor <msebor@redhat.com>
1086 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
1089 PR tree-optimization/83075
1090 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
1091 strncat/strncpy don't change length of source string.
1093 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
1096 2017-11-29 Martin Aberg <maberg@gaisler.com>
1098 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1099 to prevent b2bst errata sequence.
1100 (sqrtdf2_fix): Likewise.
1102 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1104 PR tree-optimization/81945
1105 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
1106 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
1107 to dest_cfun has orig_loop_num set, either remap it to the new
1108 loop number if the loop got moved too, or clear it.
1110 2017-12-05 Steve Ellcey <sellcey@cavium.com>
1112 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
1114 (thunderx2t99_nothing): New insn reservation.
1115 (thunderx2t99_mrs): New insn reservation.
1116 (thunderx2t99_multiple): New insn reservation.
1117 (thunderx2t99_alu_basi): Add bfx to reservation.
1118 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
1120 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1123 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
1125 2017-12-05 Bin Cheng <bin.cheng@arm.com>
1127 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
1128 tree-ssa-pre.c::remove_dead_inserted_code.
1129 * tree-ssa-dce.h: New file.
1130 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
1131 (remove_dead_inserted_code): Move and rename to function
1132 tree-ssa-dce.c::simple_dce_from_worklist.
1133 (pass_pre::execute): Update use.
1135 2017-12-05 Richard Biener <rguenther@suse.de>
1137 PR tree-optimization/83277
1138 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
1139 to code-gen liveout vars.
1141 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
1143 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
1144 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
1145 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
1148 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
1150 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
1152 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
1153 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
1156 2017-12-05 Richard Biener <rguenther@suse.de>
1158 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
1159 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
1160 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
1161 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
1163 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1165 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
1167 2017-12-05 Martin Liska <mliska@suse.cz>
1168 Jakub Jelinek <jakub@redhat.com>
1170 * doc/invoke.texi: Document the options.
1171 * flag-types.h (enum sanitize_code): Add
1172 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1173 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
1174 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1175 * opts.c: Define new sanitizer options.
1176 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
1177 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
1179 2017-12-05 Julia Koval <julia.koval@intel.com>
1181 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
1182 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
1183 (ix86_handle_option): Handle -mavx512vnni.
1184 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
1185 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
1186 * config/i386/i386-c (__AVX512VNNI__): New.
1187 * config/i386/i386.c (ix86_target_string): Handle new option.
1188 (ix86_valid_target_attribute_inner_p): Handle new option.
1189 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
1190 * config/i386/i386.opt (mavx512vnni): New option.
1192 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1195 * athlon.md: Disable for generic.
1196 * haswell.md: Enable for generic.
1197 * i386.c (ix86_sched_init_global): Add core hooks for generic.
1198 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
1200 (ix86_adjust_cost): Move generic to haswell path.
1202 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
1204 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
1205 instead of MEM_P in a couple more places. Fix formatting issues.
1207 2017-12-04 Jim Wilson <jimw@sifive.com>
1209 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
1210 instead of GP_REG_LAST-1.
1211 (riscv_adjust_libcall_cfi_prologue): Likewise.
1212 (riscv_adjust_libcall_cri_epilogue): Likewise.
1213 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
1216 2017-12-04 Luis Machado <luis.machado@linaro.org>
1218 * ipa-pure-const.c (check_decl): Add missing newline.
1219 (state_from_flags): Likewise.
1221 2017-12-04 Jeff Law <law@redhat.com>
1223 PR tree-optimizatin/78496
1224 * gimple-ssa-evrp-analyze.h
1225 (evrp_range_analyzer::get_vr_values): Simplify.
1226 * gimple-ssa-evrp-analyze.c: Corresponding changes.
1227 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
1228 and gimple-ssa-evrp-analyze.h.
1229 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
1230 (simplify_stmt_for_jump_threading): Copy a blob of code from
1231 tree-vrp.c to use ranges to simplify statements.
1232 (dom_opt_dom_walker::before_dom_children): Call
1233 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
1234 (dom_opt_dom_walker::after_dom_children): Similarly for
1235 evrp_range_analyzer::leave.
1236 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
1239 * gimple-ssa-evrp-analyze.c
1240 (evrp_range_analyzer::extract_range_from_stmt): Always use
1241 vr_values::update_value_range so preexisting range info is
1242 medged with new range info, even if the new range is VR_VARYING.
1244 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1246 * combine.c: Adjust comment.
1247 (use_crosses_set_p): Delete.
1248 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
1249 (try_combine): Ditto.
1251 2017-12-04 Richard Biener <rguenther@suse.de>
1253 PR tree-optimization/83255
1254 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
1255 Re-add zero-iteration check.
1257 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1259 PR rtl-optimization/83245
1260 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
1261 hard registers as earlyclobber, also if not in an asm.
1263 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1267 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1270 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1271 rs6000_cpu based on cpu_index, not tune_index.
1273 2017-12-04 Richard Biener <rguenther@suse.de>
1275 PR tree-optimization/83238
1276 * graphite-scop-detection.c (scop_detection::merge_sese): Make
1277 code match comment, rejecting invalid SESE regions.
1279 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
1281 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
1282 require base operand is a REG_POINTER prior to reload on targets
1283 with non-equivalent space registers.
1285 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1287 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
1288 (update_specialized_profile): Fix updating of counts.
1289 (perhaps_add_new_callers): Likewise.
1291 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1294 * x86-tune.def: Remove obsolette FIXMEs.
1295 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
1296 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
1297 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
1299 (X86_TUNE_PAD_RETURNS): Disable for generic.
1301 2017-12-02 Jakub Jelinek <jakub@redhat.com>
1303 PR tree-optimization/83170
1304 PR tree-optimization/83241
1305 * gimple-ssa-store-merging.c
1306 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
1307 gimple_vuse (ins_stmt) in case it has changed.
1308 (imm_store_chain_info::output_merged_store): Likewise.
1310 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
1314 * tree-cfg.c (pass_warn_function_return::execute): Handle
1315 __builtin_ubsan_handle_missing_return like __builtin_unreachable
1316 with BUILTINS_LOCATION.
1320 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
1321 is BLKmode for vector field with vector raw mode, use TYPE_MODE
1322 instead of DECL_MODE.
1324 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
1325 last argument to rtx pointer.
1326 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
1327 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
1328 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
1329 depending on the chosen ISAs.
1330 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
1331 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
1332 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
1334 * config/i386/sse.md (mov<mode>_internal): Likewise.
1335 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
1337 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1339 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
1341 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1343 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
1344 parameters from prototype.
1345 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
1346 parameters. Don't print a comment.
1347 (emit_fusion_gpr_load): Adjust.
1348 (emit_fusion_load_store): Adjust.
1349 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
1350 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
1351 comment on the second line.
1353 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1356 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1357 rs6000_cpu based on cpu_index, not tune_index.
1359 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1361 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
1362 which_alternative instead of which_alternative + 1.
1363 (output_asm_insn): Print an extra tab if the template is short.
1365 2017-12-01 Jim Wilson <jimw@sifive.com>
1367 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
1369 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
1370 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
1372 * doc/tm.texi: Regenerate.
1374 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
1377 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
1378 whether we can allocate pseudos before trying to fix an address.
1379 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
1380 memory address is indexed or indirect.
1381 (floatuns_<mode>si2_hw2): Likewise.
1383 2017-12-01 Jason Merrill <jason@redhat.com>
1385 * Makefile.in (TAGS): Add c-family/*.cc.
1387 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
1389 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
1390 (call_value_insn): Likewise.
1391 (sibcall_insn): Likewise.
1392 (sibcall_value_insn): Likewise.
1393 (movsi_aarch64): Likewise.
1394 (movdi_aarch64): Likewise.
1395 (add_losym_): Likewise.
1396 (ldr_got_small_): Likewise.
1397 (ldr_got_small_sidi): Likewise.
1398 (ldr_got_small_28k_): Likewise.
1399 (ldr_got_small_28k_sidi): Likewise.
1400 * config/aarch64/aarch64.c (aarch64_print_address_internal):
1401 Move output_addr_const to symbolic case. Add error check.
1403 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1405 * config/s390/predicates.md (plus16_Q_operand): New predicate.
1406 * config/s390/s390.md: Disable MVC merging peephole if it would
1407 disable operand forwarding.
1408 (new peephole2): Split MVCs if it would turn them into up to 2
1411 2017-12-01 Richard Biener <rguenther@suse.de>
1413 PR tree-optimization/83232
1414 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
1415 detection of same access. Instead of breaking the group here
1416 do not consider the duplicate. Add comment explaining real fix.
1418 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
1420 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
1422 2017-12-01 Sudakshina Das <sudi.das@arm.com>
1424 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
1426 2017-12-01 Jakub Jelinek <jakub@redhat.com>
1428 * function.h (struct function): Remove cilk_frame_decl,
1429 is_cilk_function and calls_cilk_spawn fields.
1430 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
1432 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
1433 cilk_elemental field.
1434 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
1435 * target.def: Adjust comment.
1436 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
1437 Don't test cilk_elemental.
1439 PR tree-optimization/83233
1440 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
1441 bswap_stat name for the struct.
1444 * tree.h (SWITCH_BREAK_LABEL_P): Define.
1445 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
1446 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
1447 SWITCH_BREAK_LABEL_P set on the label.
1448 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
1449 added for default case if it was missing and not all cases covered.
1450 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
1451 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
1453 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
1454 to avoid -Wimplicit-fallthrough warning.
1455 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
1456 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
1458 PR tree-optimization/83221
1459 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
1461 (init_reassoc): Formatting fix.
1464 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
1465 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
1469 * tree-cfg.c: Include asan.h.
1470 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
1472 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1474 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
1476 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1478 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
1479 vpcmpeqd instruction.
1481 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1483 * config/i386/i386.c (standard_sse_constant_opcode): Fix
1484 registers type for 128bit mode.
1486 2017-11-30 Jakub Jelinek <jakub@redhat.com>
1488 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
1490 * gengtype-state.c (read_state_pair): Likewise.
1491 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
1492 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
1493 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
1494 * ifcvt.c (noce_try_store_flag_constants): Likewise.
1495 * tree-ssa-ccp.c (ccp_finalize): Likewise.
1496 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
1497 * builtins.c (fold_builtin_3): Likewise.
1498 * graphite-scop-detection.c
1499 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
1500 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
1503 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
1506 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
1507 parameter. Return true if ignore_decl_rtl_set_p is true and other
1508 conditions are satisfied.
1509 * asan.h (asan_protect_global): Add new parameter.
1510 * varasm.c (categorize_decl_for_section): Pass true as second parameter
1511 to asan_protect_global calls.
1513 2017-11-30 Jim Wilson <jimw@sifive.com>
1515 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
1516 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
1517 -mstrict-align, add info on default value. Delete redundant lines for
1518 -mabi. Add missing -mexplicit-relocs docs.
1520 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1522 * config/arc/arc.md (trap): New pattern.
1524 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1526 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
1527 instruction to end into a delay slot.
1528 * config/arc/arc.md (cond_delay_insn): Check if the instruction
1529 can be placed into a delay slot against reg_note.
1531 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1533 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
1534 labels number of usages.
1536 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1538 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
1540 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
1542 2017-11-30 Jakub Jelinek <jakub@redhat.com>
1545 * internal-fn.c (expand_mul_overflow): Optimize unsigned
1546 multiplication by power of 2 constant into two shifts + comparison.
1548 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
1551 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
1553 2017-11-30 Richard Biener <rguenther@suse.de>
1555 PR tree-optimization/83202
1556 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
1557 allow_peel argument and guard peeling.
1558 (canonicalize_loop_induction_variables): Likewise.
1559 (canonicalize_induction_variables): Pass false.
1560 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
1561 peeling from cunrolli.
1563 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1565 * combine.c (try_combine): Print a message to dump file whenever
1566 I0, I1, or I2 cannot be combined into I3.
1568 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1570 PR rtl-optimization/83156
1571 PR rtl-optimization/82621
1572 * combine.c (try_combine): Don't split an I2 if one of the dests is
1573 set again before I3. Allow unused dests.
1575 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1577 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
1579 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
1581 PR rtl-optimization/80818
1582 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
1583 recursively. Use insn code for clobber.
1584 (lra_set_insn_recog_data): Pass the new arg to
1585 collect_non_operand_hard_regs.
1586 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
1588 (lra_update_insn_regno_info): Pass insn to
1589 add_regs_to_insn_regno_info.
1591 2017-11-29 Jim Wilson <jimw@sifive.com>
1592 Andrew Waterman <andrew@sifive.com>
1594 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
1595 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
1596 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
1597 use SINGLE_SHIFT_COST.
1598 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
1599 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
1601 2017-11-29 Julia Koval <julia.koval@intel.com>
1603 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
1604 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
1605 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
1606 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
1607 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
1608 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
1609 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
1610 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
1611 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
1612 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
1613 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
1614 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
1615 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
1616 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
1617 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
1618 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
1619 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
1620 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
1621 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
1622 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
1623 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
1624 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
1625 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
1626 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
1627 __builtin_ia32_vpshldv_v2di_mask,
1628 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
1629 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
1630 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
1632 2017-11-29 Julia Koval <julia.koval@intel.com>
1634 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
1635 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
1636 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
1637 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
1638 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
1639 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
1640 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
1641 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
1642 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
1643 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
1644 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
1645 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
1646 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
1647 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
1648 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
1649 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
1650 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
1651 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
1652 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1653 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
1654 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
1656 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1658 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
1659 movsi_pic_gotdata_op instruction as a load for the UT699 errata
1662 2017-11-29 Martin Aberg <maberg@gaisler.com>
1664 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1665 to prevent b2bst errata sequence.
1666 (sqrtdf2_fix): Likewise.
1668 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1670 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
1671 (div_sqrt_insn_p): New function.
1672 (sparc_do_work_around_errata): Insert NOP instructions to
1673 prevent sequences that could trigger the TN-0013 errata for
1674 certain LEON3 processors.
1675 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
1676 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
1677 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
1678 (in_branch_delay): Prevent div and sqrt in delay slot if
1680 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
1682 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1684 * config/sparc/sparc.c (atomic_insn_p): New function.
1685 (sparc_do_work_around_errata): Insert NOP instructions to
1686 prevent sequences that could trigger the TN-0010 errata for
1688 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
1689 instruction referable in atomic_insns_p.
1691 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1693 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
1694 (atomic_compare_and_swap_leon3_1): Likewise.
1697 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1699 * config/sparc/sparc.c (fpop_insn_p): New function.
1700 (sparc_do_work_around_errata): Insert NOP instructions to
1701 prevent sequences that could trigger the TN-0012 errata for
1703 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
1704 * config/sparc/sparc.md (fix_gr712rc): New attribute.
1705 (in_branch_annul_delay): Prevent floating-point instructions
1706 in delay slot of annulled integer branch.
1708 2017-11-29 Richard Biener <rguenther@suse.de>
1710 PR tree-optimization/83202
1711 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
1713 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
1714 nodes vectorized to the same stmts multiple times.
1715 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
1716 (vect_analyze_slp_instance): Adjust.
1717 (scalar_stmts_to_slp_tree_map_t): New typedef.
1718 (vect_schedule_slp_instance): Add a map recording the SLP node
1719 representing the vectorized stmts for a set of scalar stmts.
1720 Avoid code-generating redundancies.
1721 (vect_schedule_slp): Allocate map and pass it down.
1723 2017-11-29 Nathan Sidwell <nathan@acm.org>
1726 * tree.c (build_complex_type): Fix canonicalization. Only fill in
1729 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
1731 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
1732 cases for printing LDP/STP memory addresses.
1733 (aarch64_print_address_internal): Renamed from
1734 aarch64_print_operand_address, added parameter, add Pmode check.
1735 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
1736 (aarch64_print_operand_address): Indirect to
1737 aarch64_print_address_internal.
1738 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
1739 'y' operand output specifier.
1741 2017-11-29 Jakub Jelinek <jakub@redhat.com>
1744 * tree.c (build_simple_mem_ref_loc): Handle
1745 get_addr_base_and_unit_offset returning a MEM_REF.
1748 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
1751 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
1754 2017-11-28 David Malcolm <dmalcolm@redhat.com>
1756 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
1758 (layout::move_to_column): Likewise.
1760 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1763 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
1764 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
1766 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1767 Martin Jambor <mjambor@suse.cz>
1770 * tree.h (expr_type_first_operand_type_p): Declare
1771 * tree.c (expr_type_first_operand_type_p): New function.
1772 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
1773 (ipa_value_from_jfunc): Adjust declaration.
1774 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
1775 Use it as result type for arithmetics, unless it is NULL in which case
1776 be more conservative.
1777 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
1778 ipa_get_jf_pass_through_result.
1779 (propagate_vals_across_pass_through): Likewise.
1780 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
1781 is to propagate_vals_across_pass_through.
1782 (propagate_constants_across_call): Pass PARM_TYPE to
1783 propagate_scalar_across_jump_function.
1784 (find_more_scalar_values_for_callers_subset): Pass parameter type to
1785 ipa_value_from_jfunc.
1786 (cgraph_edge_brings_all_scalars_for_node): Likewise.
1787 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
1788 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
1789 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
1790 target_type, pass it to ipa_value_from_jfunc.
1791 (update_indirect_edges_after_inlining): Pass parameter type to
1792 try_make_edge_direct_simple_call.
1794 2017-11-28 Jeff Law <law@redhat.com>
1796 * gimple-ssa-evrp-analyze.c
1797 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
1798 refine ranges if scev_initialized_p returns true.
1799 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
1801 2017-11-28 Julia Koval <julia.koval@intel.com>
1803 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
1804 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
1805 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
1806 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
1807 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
1808 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
1809 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
1810 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
1811 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
1812 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
1813 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
1814 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
1815 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
1816 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
1817 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
1818 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
1819 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
1820 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
1821 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
1822 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
1823 __builtin_ia32_vpshrd_v2di_mask): New builtins.
1824 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
1826 2017-11-28 Julia Koval <julia.koval@intel.com>
1828 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
1829 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
1830 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
1831 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
1832 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
1833 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
1834 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
1835 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
1836 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
1837 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
1838 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
1839 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
1840 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
1841 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
1842 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
1843 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
1844 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
1845 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
1846 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
1847 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
1848 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
1849 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
1850 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
1851 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
1852 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
1853 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
1854 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
1855 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
1856 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
1857 __builtin_ia32_vpshld_v2di_mask): New builtins.
1858 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1859 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
1861 2017-11-28 Richard Biener <rguenther@suse.de>
1863 PR tree-optimization/80776
1864 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
1866 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
1868 (evrp_range_analyzer::record_ranges_from_incoming_edges):
1869 If the incoming edge is an effective fallthru because the other
1870 edge only reaches a __builtin_unreachable () then record ranges
1871 derived from the controlling condition in SSA info.
1872 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
1873 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
1875 2017-11-28 Olivier Hainque <hainque@adacore.com>
1877 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
1878 on mingw build hosts.
1880 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1883 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
1885 2017-11-28 Richard Biener <rguenther@suse.de>
1888 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
1889 copies generated from memcpy use a character array as reference
1892 2017-11-28 Julia Koval <julia.koval@intel.com>
1893 Sebastian Peryt <sebastian.peryt@intel.com>
1895 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
1896 c-family/c-cilkplus.o, c-family/array-notation-common.o,
1897 cilk-common.o, cilk.h, cilk-common.c): Remove.
1899 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
1900 * builtins.c (is_builtin_name): Remove cilkplus condition.
1901 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
1902 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
1903 cilk-builtins.def, cilkplus.def): Remove.
1904 * cif-code.def (CILK_SPAWN): Remove.
1905 * cilk-builtins.def: Delete.
1906 * cilk-common.c: Ditto.
1908 * cilkplus.def: Ditto.
1909 * config/darwin.h (fcilkplus): Delete.
1910 * cppbuiltin.c: Ditto.
1911 * doc/extend.texi: Remove cilkplus doc.
1912 * doc/generic.texi: Ditto.
1913 * doc/invoke.texi: Ditto.
1914 * doc/passes.texi: Ditto.
1915 * gcc.c (fcilkplus): Remove.
1916 * gengtype.c (cilk.h): Remove.
1917 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
1919 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
1921 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
1923 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
1924 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
1925 cilkplus conditions.
1926 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
1927 inline_read_section): Ditto.
1928 * ipa-inline-analysis.c (cilk.h): Remove.
1929 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
1930 * lto-wrapper.c (merge_and_complain, append_compiler_options,
1931 append_linker_options): Remove condition for fcilkplus.
1932 * lto/lto-lang.c (cilk.h): Remove.
1933 (lto_init): Remove condition for fcilkplus.
1934 * omp-expand.c (expand_cilk_for_call): Delete.
1935 (expand_omp_taskreg, expand_omp_for_static_chunk,
1936 expand_omp_for): Remove cilkplus
1938 (expand_cilk_for): Delete.
1939 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
1940 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
1941 execute_lower_omp, diagnose_sb_0): Ditto.
1942 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
1943 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
1944 * tree-nested.c: Ditto.
1945 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
1946 (dump_generic_node): Ditto.
1947 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
1948 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
1950 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
1952 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1954 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
1955 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
1956 for vector mode and !TARGET_SIMD.
1958 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1960 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
1962 * tree.h (SWITCH_LABELS): Remove.
1963 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
1964 assert SWITCH_BODY is non-NULL.
1965 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
1967 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
1969 PR tree-optimization/80788
1970 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
1971 has TREE_OVERFLOW set, call drop_tree_overflow.
1973 2017-11-28 Richard Biener <rguenther@suse.de>
1975 PR tree-optimization/83158
1976 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
1978 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
1981 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
1982 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
1984 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
1986 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
1987 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
1989 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1992 * varasm.c (compare_constant): Take the mode of the constants into
1993 account when comparing floating point constants.
1995 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
1997 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
1998 from explicit instantiation of debug_helper.
1999 * vec.h (DEFINE_DEBUG_VEC): Ditto.
2001 2017-11-27 Richard Biener <rguenther@suse.de>
2003 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
2006 2017-11-27 Richard Biener <rguenther@suse.de>
2008 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
2009 (type_hash_canon): Also clear int_cst_hash_table entry for
2011 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
2013 2017-11-27 Tamar Christina <tamar.christina@arm.com>
2015 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2016 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2017 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2019 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2021 * hash-map.h (gt_cleare_cache): Avoid UB.
2023 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
2025 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
2026 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
2027 (decide_unroll_runtime_iterations): Likewise.
2028 (decide_unroll_stupid): Likewise.
2030 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2033 * config/i386/i386.md: Add a loop with incssp.
2035 2017-11-27 Martin Jambor <mjambor@suse.cz>
2037 PR tree-optimization/81248
2038 * tree-sra.c (splice_param_accesses): Remove size check.
2039 (decide_one_param_reduction): Fix size check.
2040 * gimple-pretty-print.c (dump_profile): Silence warning.
2041 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
2043 2017-11-27 Jakub Jelinek <jakub@redhat.com>
2046 * dbxout.c (lastlineno): New variable.
2047 (dbx_debug_hooks): Use dbxout_switch_text_section as
2048 switch_text_section debug hook.
2049 (dbxout_function_end): Switch to current_function_section
2050 rather than function_section. If crtl->has_bb_partition,
2051 output just one N_FUN, depending on in_cold_section_p.
2052 (dbxout_source_line): Remember last lineno in lastlineno.
2053 (dbxout_switch_text_section): New function.
2054 (dbxout_function_decl): Adjust dbxout_block caller.
2055 (dbx_block_with_cold_children): New function.
2056 (dbxout_block): Return true if any LBRAC/RBRAC have been
2057 emitted. Use dbx_block_with_cold_children at depth == 0
2058 in second partition. Add PARENT_BLOCKNUM argument, pass
2059 it optionally adjusted to children. Output LBRAC/RBRAC
2060 around recursive call only if the block is in the current
2061 partition, if not and anything was output, emit empty
2063 * final.c (final_scan_insn): Compute cold_function_name
2064 before calling switch_text_section debug hook. Call
2065 that hook even if dwarf2out_do_frame if not emitting
2069 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
2070 TREE_READONLY decls.
2072 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2074 PR rtl-optimization/82488
2075 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
2077 2017-11-26 Julia Koval <julia.koval@intel.com>
2079 * config/i386/i386.c (processor_target_table): Add skylake_cost for
2081 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
2084 2017-11-26 Julia Koval <julia.koval@intel.com>
2086 * config/i386/driver-i386.c (host_detect_local_cpu):
2087 Detect skylake-avx512.
2089 2017-11-26 Julia Koval <julia.koval@intel.com>
2091 * config.gcc: Add -march=cannonlake.
2092 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
2093 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
2094 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
2095 (PTA_CANNONLAKE): New.
2096 (processor_target_table): Add cannonlake.
2097 (ix86_option_override_internal): Ditto.
2098 (fold_builtin_cpu): Ditto.
2099 (get_builtin_code_for_version): Handle cannonlake.
2100 (M_INTEL_COREI7_CANNONLAKE): New.
2101 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
2102 * doc/invoke.texi: Add -march=cannonlake.
2104 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
2106 * plugin.c (add_new_plugin): Use platform-specific library extensions.
2107 (try_init_one_plugin): Alternative implementation for MinGW.
2108 * Makefile.in (plugin_implib): New.
2109 (gengtype-lex.c): Fix broken AIX workaround.
2110 * configure: Regenerate.
2111 * doc/plugins.texi: Document support for MinGW.
2113 2017-11-25 Jakub Jelinek <jakub@redhat.com>
2115 PR rtl-optimization/81553
2116 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
2117 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
2118 is a shift where C1 has different mode than the whole shift, use C1's
2119 mode for MULT rather than the shift's mode.
2122 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
2123 builtins not enabled in the currently selected ISA.
2125 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
2127 PR tree-optimization/71026
2128 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
2129 (insert_reciprocals): Change to insert reciprocals before a division
2130 by a square and to insert the square of a reciprocal.
2131 (execute_cse_reciprocals_1): Change to consider division by a square.
2132 (register_division_in): Add importance parameter.
2134 2017-11-24 Richard Biener <rguenther@suse.de>
2136 PR tree-optimization/82402
2137 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
2138 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
2140 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
2142 * match.pd (0-ptr): New transformation.
2144 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
2147 * ipa-inline.c (inline_small_functions): Set current badnes correctly
2148 when skipping checking.
2150 2017-11-24 Richard Biener <rguenther@suse.de>
2152 PR tree-optimization/83128
2153 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
2154 (vn_reference_lookup_3): Likewise.
2156 2017-11-24 Jakub Jelinek <jakub@redhat.com>
2159 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
2160 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
2163 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
2164 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
2165 check for the latter with gimple_call_builtin_p. Do not handle
2166 BUILT_IN_STPNCPY_CHK which is not a pass through call.
2168 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
2170 * config/arm/arm_neon.h: Fix pragma GCC push_options before
2173 2017-11-23 Julia Koval <julia.koval@intel.com>
2175 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
2176 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
2177 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
2178 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
2179 _mm512_maskz_expandloadu_epi16): New intrinsics.
2180 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
2181 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
2182 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
2183 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
2184 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
2185 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
2186 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
2187 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
2188 _mm256_maskz_expandloadu_epi8): New intrinsics.
2189 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
2190 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
2191 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
2192 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
2193 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
2194 * config/i386/sse.md (VI248_VLBW): New iterator.
2195 (expand<mode>_mask, expand<mode>_maskz): New patterns.
2197 2017-11-23 Julia Koval <julia.koval@intel.com>
2199 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
2200 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
2201 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
2202 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
2203 _mm512_mask_compressstoreu_epi16): New.
2204 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
2205 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
2206 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
2207 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
2208 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
2209 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
2210 _mm256_mask_compressstoreu_epi8): New.
2211 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
2212 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
2213 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
2214 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
2215 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
2216 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
2217 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
2218 __builtin_ia32_compresshi128_mask,
2219 __builtin_ia32_compressstoreuqi512_mask,
2220 __builtin_ia32_compressstoreuhi512_mask,
2221 __builtin_ia32_compressstoreuqi256_mask,
2222 __builtin_ia32_compressstoreuqi128_mask,
2223 __builtin_ia32_compressstoreuhi256_mask,
2224 __builtin_ia32_compressstoreuhi128_mask): New builtins.
2225 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
2227 (ix86_expand_special_args_builtin): Handle new types.
2228 (s4fma_expand): Handle new builtin array.
2229 * config/i386/immintrin.h: Include new headers.
2230 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
2231 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
2233 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2235 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
2236 fixes. Declare temp and g variables at the top in order to avoid
2237 {} in most of the cases.
2239 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
2241 * match.pd (ptr-0): New transformation.
2243 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2245 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
2246 (struct addr_mode_cost_table): New.
2247 (struct tune_params): Add field addr_mode_costs.
2248 * config/arm/arm.c (generic_addr_mode_costs): New.
2249 (arm_slowmul_tune): Initialise addr_mode_costs field.
2250 (arm_fastmul_tune): Likewise.
2251 (arm_strongarm_tune): Likewise.
2252 (arm_xscale_tune): Likewise.
2253 (arm_9e_tune): Likewise.
2254 (arm_marvell_pj4_tune): Likewise.
2255 (arm_v6t2_tune): Likewise.
2256 (arm_cortex_tune): Likewise.
2257 (arm_cortex_a8_tune): Likewise.
2258 (arm_cortex_a7_tune): Likewise.
2259 (arm_cortex_a15_tune): Likewise.
2260 (arm_cortex_a35_tune): Likewise.
2261 (arm_cortex_a53_tune): Likewise.
2262 (arm_cortex_a57_tune): Likewise.
2263 (arm_exynosm1_tune): Likewise.
2264 (arm_xgene1_tune): Likewise.
2265 (arm_cortex_a5_tune): Likewise.
2266 (arm_cortex_a9_tune): Likewise.
2267 (arm_cortex_a12_tune): Likewise.
2268 (arm_cortex_a73_tune): Likewise.
2269 (arm_v7m_tune): Likewise.
2270 (arm_cortex_m7_tune): Likewise.
2271 (arm_v6m_tune): Likewise.
2272 (arm_fa726te_tune): Likewise.
2273 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
2275 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2277 * config/arm/arm.c (arm_mem_costs): New function.
2278 (arm_rtx_costs_internal): Use arm_mem_costs.
2280 2017-11-23 Mark Wielaard <mark@klomp.org>
2282 * dwarf2out.c (init_sections_and_labels): Use generation to create
2283 unique ranges_section_label and ranges_base_label. Return generation.
2284 (output_rnglists): Add generation argument. Use generation to create
2285 unique ranges labels.
2286 (dwarf2out_finish): Get generation from init_sections_and_labels
2287 and pass generation to output_rnglists.
2289 2017-11-23 Mike Stump <mikestump@comcast.net>
2290 Eric Botcazou <ebotcazou@adacore.com>
2292 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
2293 * cfgloop.h (struct loop): Add unroll field.
2294 * function.h (struct function): Add has_unroll bitfield.
2295 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
2296 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
2297 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
2298 (pass_rtl_unroll_loops::gate): Likewise.
2299 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
2300 for which loop->unroll==1.
2301 (decide_unroll_constant_iterations): Use note for consistency and
2302 take loop->unroll into account. Return early if loop->unroll is set.
2303 Fix thinko in existing test.
2304 (decide_unroll_runtime_iterations): Use note for consistency and
2305 take loop->unroll into account.
2306 (decide_unroll_stupid): Likewise.
2307 * lto-streamer-in.c (input_cfg): Read loop->unroll.
2308 * lto-streamer-out.c (output_cfg): Write loop->unroll.
2309 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
2311 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
2312 (print_loop): Print loop->unroll if set.
2313 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
2314 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
2315 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
2317 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
2318 loop->unroll is set and smaller than the trip count. Otherwise bypass
2319 entirely the heuristics if loop->unroll is set. Remove dead note.
2320 Fix off-by-one bug in other note.
2321 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
2322 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
2324 (tree_unroll_loops_completely): Make static.
2325 (pass_complete_unroll::execute): Use correct type for variable.
2326 (pass_complete_unrolli::execute): Fix formatting.
2327 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
2329 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
2331 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
2332 enable when TARGET_PREFER_AVX128 is set.
2334 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2336 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
2338 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
2339 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
2341 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2343 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
2344 frequencies. Fix estimation of aggregate parameters.
2346 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2348 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
2351 2017-11-23 Tom de Vries <tom@codesourcery.com>
2353 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
2355 2017-11-23 Tom de Vries <tom@codesourcery.com>
2357 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
2360 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
2363 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
2364 sibcall_value_pcrel_fdpic): Use local variable instead of
2366 (calli_tbr_rel): Add missing operand 2.
2367 (call_valuei_tbr_rel): Add missing operand 3.
2369 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2372 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
2373 bitpos/bitsize covering the whole destination, use store_expr only if
2374 the complex mode is the same. Otherwise, use expand_normal and if
2375 it returns CONCAT, subreg each part separately instead of trying to
2376 subreg the whole result.
2378 2017-11-23 Richard Biener <rguenther@suse.de>
2380 PR tree-optimization/23094
2381 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
2382 come here from walking over backedges in the first iteration.
2383 (vn_reference_lookup_3): Skip clobbers that store the same value.
2385 2017-11-23 Richard Biener <rguenther@suse.de>
2387 PR tree-optimization/81403
2388 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
2389 a block we need a leader relative to.
2390 (phi_translate_1): For nary processing require a leader from
2391 get_representative_for given we run expression simplification
2392 using match-and-simplify. Remove previous fix.
2394 2017-11-22 Jeff Law <law@redhat.com>
2396 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
2397 Use new method allocate_value_range rather than accessing the
2398 vrp_value_range_pool data member directly.
2399 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
2400 to use extract_range_from_stmt method to avoid need for
2401 extract_range_from_assignment method.
2402 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
2403 method rather than setting values_propgated data member directly.
2404 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
2405 and values propagated data members and extract_range_from_assignment
2406 method. Reorder private data members to conform to standards.
2407 Add new methods set_lattice_propagation_complete and
2408 allocate_value_range.
2410 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
2412 PR rtl-optimization/83030
2413 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
2414 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
2415 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
2416 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
2417 CROSSING_JUMP_P flag is already set before setting it.
2418 * cfgrtl.c (fixup_partition_crossing): Likewise.
2419 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
2422 2017-11-22 Jakub Jelinek <jakub@redhat.com>
2424 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
2425 Handle the case where both arguments are using gen_const_vec_series.
2427 2017-11-22 David Malcolm <dmalcolm@redhat.com>
2430 * pretty-print.c (pp_format): Move quoting implementation to
2431 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
2432 to pass address of "quote" local.
2433 (pp_begin_quote): New function.
2434 (pp_end_quote): New function.
2435 * pretty-print.h (printer_fn): Convert penultimate param from bool
2437 (pp_begin_quote): New decl.
2438 (pp_end_quote): New decl.
2439 * tree-diagnostic.c (default_tree_printer): Convert penultimate
2440 param from bool to bool *.
2441 * tree-diagnostic.h (default_tree_printer): Likewise.
2443 2017-11-22 Jeff Law <law@redhat.com>
2445 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
2446 Initialize vr_values.
2447 (evrp_range_analyzer::try_find_new_range): Call methods attached to
2448 vr_values via vr_values class instance rather than delegators.
2449 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
2450 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
2451 (evrp_range_analyzer::push_value_range): Likewise.
2452 (evrp_range_analyzer::pop_value_range): Likewise.
2453 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
2454 most delegators. Those remaining are exposed as public interfaces.
2455 Make vr_values a pointer and private.
2456 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
2458 (evrp_range_analyzer::get_vr_value): New method.
2459 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
2460 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
2461 (class evrp_dom_walker): Attach evrp_folder class, initialize
2462 it in the ctor. Remove temporary delegators.
2463 (evrp_dom_walker::before_dom_children): Call methods in attached
2464 evrp_range_analyzer class via class instance pointer. Use
2465 free value_range_constant_singleton to remove need for
2466 op_with_constant_singleton_value delegator method. Do not
2467 create a vrp_prop class instance for every call! Narrow
2468 scope of a couple variables.
2469 (evrp_dom_walker::cleanup): Call methods in attached
2470 evrp_range_analyzer class via class instance pointer.
2471 * vr-values.h (class vr_values): Privatize many methods and
2474 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2476 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
2477 * cfgexpand.c (expand_debug_expr): Remove handling for them.
2478 * expr.c (expand_expr_real_2): Likewise.
2479 * fold-const.c (const_unop): Likewise.
2480 * optabs-tree.c (optab_for_tree_code): Likewise.
2481 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2482 * tree-inline.c (estimate_operator_cost): Likewise.
2483 * tree-pretty-print.c (dump_generic_node): Likewise.
2484 (op_code_prio): Likewise.
2485 (op_symbol_code): Likewise.
2486 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
2487 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
2488 * internal-fn.c (direct_internal_fn_optab): New function.
2489 (direct_internal_fn_array, direct_internal_fn_supported_p
2490 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
2491 * fold-const-call.c (fold_const_reduction): New function.
2492 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
2494 * tree-vect-loop.c: Include internal-fn.h.
2495 (reduction_code_for_scalar_code): Rename to...
2496 (reduction_fn_for_scalar_code): ...this and return an internal
2498 (vect_model_reduction_cost): Take an internal_fn rather than
2500 (vect_create_epilog_for_reduction): Likewise. Build calls rather
2502 (vectorizable_reduction): Use internal functions rather than tree
2503 codes for the reduction operation. Update calls to the functions
2505 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
2506 Use calls to internal functions rather than REDUC tree codes.
2507 * config/aarch64/aarch64-simd.md: Update comment accordingly.
2509 2017-11-22 Olivier Hainque <hainque@adacore.com>
2511 * config/vxworks.c (vxworks_override_options): Pick default
2512 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
2513 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
2514 DWARF_GNAT_ENCODINGS_DEFAULT.
2515 * config/vxworksae.h: Likewise.
2517 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
2519 PR tree-optimization/83104
2520 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
2523 2017-11-22 Marek Polacek <polacek@redhat.com>
2524 H.J. Lu <hongjiu.lu@intel.com>
2525 Jason Merrill <jason@redhat.com>
2530 * calls.c (initialize_argument_information): Call
2531 warn_parameter_passing_abi target hook.
2532 (store_one_arg): Use 0 for empty record size. Don't push 0 size
2533 argument onto stack.
2534 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
2535 * common.opt: Update -fabi-version description.
2536 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
2537 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
2539 (ix86_is_empty_record): New function.
2540 (ix86_warn_parameter_passing_abi): New function.
2541 (TARGET_EMPTY_RECORD_P): Redefine.
2542 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
2543 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
2544 * doc/tm.texi: Regenerated.
2545 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
2546 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
2547 * dwarf2out.c (get_ultimate_context): Move to tree.c.
2548 * explow.c (hard_function_value): Call arg_int_size_in_bytes
2549 instead of int_size_in_bytes.
2550 * expr.c (copy_blkmode_to_reg): Likewise.
2551 * function.c (aggregate_value_p): Return 0 for empty types.
2552 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
2554 (locate_and_pad_parm): Call arg size_in_bytes instead
2556 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
2557 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
2558 * target.def (empty_record_p, warn_parameter_passing_abi): New target
2560 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
2561 (std_gimplify_va_arg_expr): Skip empty records. Call
2562 arg_size_in_bytes instead size_in_bytes.
2563 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
2564 * tree-core.h (tree_type_common): Add empty_flag.
2565 (tree_decl_common): Update comments.
2566 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
2568 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
2569 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
2571 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
2572 * tree.c (default_is_empty_type): New function.
2573 (default_is_empty_record): New function.
2574 (arg_int_size_in_bytes): New function.
2575 (arg_size_in_bytes): New function.
2576 (get_ultimate_context): New function.
2577 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
2578 TRANSLATION_UNIT_WARN_EMPTY_P.
2579 (default_is_empty_record, arg_int_size_in_bytes,
2580 arg_size_in_bytes, get_ultimate_context): Declare.
2582 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2584 * config/arm/arm.c (cmse_clear_registers): New function.
2585 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
2586 code by call to cmse_clear_registers.
2587 (cmse_nonsecure_entry_clear_before_return): Likewise.
2589 2017-11-22 Tamar Christina <tamar.christina@arm.com>
2591 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
2592 (vdot_s32, vdotq_s32): New.
2593 (vdot_lane_u32, vdotq_lane_u32): New.
2594 (vdot_lane_s32, vdotq_lane_s32): New.
2597 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2600 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
2601 for unsigned values with fewer HWIs than the precision.
2602 (test_overflow): New function.
2603 (wide_int_cc_tests): Call it.
2605 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2607 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
2608 has the same mode class as Pmode.
2610 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2612 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
2613 padding_bits_to_clear_ptr.
2614 (cmse_nonsecure_entry_clear_before_return): Likewise.
2616 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2618 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
2619 auto_sbitap instead of integer bitfield to control register needing
2622 2017-11-22 Jakub Jelinek <jakub@redhat.com>
2624 PR tree-optimization/83044
2625 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
2626 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
2627 that need the upper bound. Subtract offset from
2628 get_addr_base_and_unit_offset only if positive and subtract it
2629 before division by eltsize rather than after it.
2632 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
2633 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
2634 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
2638 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
2639 expand_binop with *mul_widen_optab, make sure at least one of the
2640 operands doesn't have VOIDmode.
2643 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
2645 PR rtl-optimization/82044
2646 PR tree-optimization/82042
2647 * dse.c (record_store): Check for overflow.
2648 (check_mem_read_rtx): Properly check for overflow if width == -1, call
2649 add_wild_read instead of clear_rhs_from_active_local_stores on
2650 overflow and log it into dump_file.
2652 2017-11-22 Richard Biener <rguenther@suse.de>
2654 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
2655 fake edges to exit when looking for a place to insert.
2656 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
2658 (insert_into_preds_of_block): Commit edge insertion immediately,
2659 assert that doesn't require new BBs.
2660 (fini_pre): Release expressions.
2661 (pass_pre::execute): Shuffle things around a bit, if the fn
2662 is too large do not compute AVAIL either as this is really the
2665 2017-11-22 Richard Biener <rguenther@suse.de>
2667 PR tree-optimization/83089
2668 * tree-if-conv.c (pass_if_conversion::execute): If anything
2669 changed reset SCEV and free the number of iteration estimates.
2671 2017-11-21 Martin Sebor <msebor@redhat.com>
2673 PR tree-optimization/82945
2674 * calls.h (warn_nonstring_bound): Remove unused function.
2676 2017-11-21 Martin Sebor <msebor@redhat.com>
2678 PR tree-optimization/82945
2679 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
2680 * calls.h (maybe_warn_nonstring_arg): Declare new function.
2681 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
2683 (initialize_argument_information): Call maybe_warn_nonstring_arg.
2684 * calls.h (get_attr_nonstring_decl): Declare new function.
2685 * doc/extend.texi (attribute nonstring): Update.
2686 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
2687 get_attr_nonstring_decl and handle it.
2688 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
2689 detection of nul-termination.
2690 (strlen_to_stridx): Change to a pointer.
2691 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
2692 (pass_strlen::execute): Same.
2694 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
2696 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
2697 for the new option -mprefer-vector-width=[none|128|256|512].
2698 * config/i386/i386.c (ix86_target_string): remove old style options
2699 -mprefer-avx256 and make -mprefer-avx128 as alias.
2700 (ix86_option_override_internal): Apply defaults for the
2701 -mprefer-vector-width=[128|256] option.
2702 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
2703 Implement macros to work with -mprefer-vector-width=.
2704 * config/i386/i386.opt: Implemented option
2705 -mprefer-vector-width=[none|128|256|512].
2706 * doc/invoke.texi: Documentation for
2707 -mprefer-vector-width=[none|128|256|512].
2709 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
2711 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
2713 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
2715 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
2717 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
2719 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
2721 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
2722 * expr.c (expand_expr_real_2): Likewise.
2723 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
2724 fold_binary_loc): Likewise.
2725 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
2726 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
2727 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
2728 MINUS_EXPR transformations.
2729 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
2730 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
2731 * tree-inline.c (estimate_operator_cost): Likewise.
2732 * tree-pretty-print.c (dump_generic_node, op_code_prio,
2733 op_symbol_code): Likewise.
2734 * tree-vect-stmts.c (vectorizable_operation): Likewise.
2735 * vr-values.c (extract_range_from_binary_expr): Likewise.
2736 * varasm.c (initializer_constant_valid_p_1): Likewise.
2737 * tree.def: New tree code POINTER_DIFF_EXPR.
2739 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
2741 * config/i386/i386.md (*bswap<mode>2_movbe): Add
2742 integer suffix to movbe mnemonic.
2743 (*bswaphi2_movbe): Ditto.
2744 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
2746 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2749 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
2750 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
2751 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
2753 2017-11-21 Martin Liska <mliska@suse.cz>
2755 * tree-inline.c (expand_call_inline): Remove not needed
2758 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
2759 Jakub Jelinek <jakub@redhat.com>
2762 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
2763 Remove static keyword from f variable.
2765 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2767 PR tree-optimization/83086
2768 * gimple-ssa-store-merging.c
2769 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
2770 rather than n.base_addr.
2772 2017-11-21 Martin Liska <mliska@suse.cz>
2774 PR rtl-optimization/82044
2775 PR tree-optimization/82042
2776 * dse.c (check_mem_read_rtx): Check for overflow.
2778 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2780 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
2783 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2785 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
2786 and its corresponding call arg location note.
2788 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2789 Andrew Burgess <andrew.burgess@embecosm.com>
2791 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
2793 (arc_return_slot_offset): Likewise.
2794 (arc_eh_return_address_location): New declaration.
2795 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
2796 (MUST_SAVE_REGISTER): Add exception handler case.
2797 (MUST_SAVE_RETURN_ADDR): Likewise.
2798 (arc_frame_pointer_required): Likewise.
2799 (arc_frame_pointer_needed): New function.
2800 (arc_compute_frame_size): Changed.
2801 (arc_expand_prologue): Likewise.
2802 (arc_expand_epilogue): Likewise.
2803 (arc_initial_elimination_offset): Likewise.
2804 (arc_return_slot_offset): Delete.
2805 (arc_eh_return_address_location): New function.
2806 (arc_builtin_setjmp_frame_value): Likewise.
2807 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
2808 (EH_RETURN_STACKADJ_RTX): Define.
2809 (EH_RETURN_HANDLER_RTX): Likewise.
2810 * config/arc/arc.md (eh_return): Delete.
2812 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2814 * print-tree.h (debug_vec_tree): Remove prototype.
2815 * gdbinit.in (pvt): Remove macro.
2817 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2819 PR tree-optimization/83047
2820 * gimple-ssa-store-merging.c
2821 (imm_store_chain_info::output_merged_store): If the loads with the
2822 same vuse are in different basic blocks, for load_gsi pick a load
2823 location that is dominated by the other loads.
2826 * config/i386/i386.c (ix86_memmodel_check): Start
2827 -Winvalid-memory-model diagnostics with lowercase letter.
2830 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
2831 set in_cold_section_p to the partition containing loc_list->first.
2832 When seeing loc_list->last_before_switch node, update secname and
2833 perform range_across_switch second partition handling only after that.
2836 * run-rtl-passes.c: Include debug.h.
2837 (run_rtl_passes): Call debug_hooks->assembly_start.
2838 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
2842 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
2843 OPTAB_DIRECT in calls to expand_simple_binop.
2845 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2848 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
2850 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2853 * spellcheck.h (best_match::blithely_get_best_candidate): New
2856 2017-11-20 Jakub Jelinek <jakub@redhat.com>
2858 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
2859 with lower case letter.
2861 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
2863 * config/i386/i386.md (bswaphi2): New expander.
2864 (*bswaphi2_movbe): New insn pattern.
2865 (bswaphi -> rorhi pepehole2): New peephole pattern.
2867 2017-11-20 Jeff Law <law@redhat.com>
2869 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
2870 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
2871 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
2872 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
2873 gimple-ssa-evrp-analyze.h.
2875 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
2876 set BB_VISITED here.
2877 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
2879 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2881 * tree-predcom.c: Add general comment on Store-Store chains.
2882 (split_data_refs_to_components): Postpone clearing eliminate_store_p
2884 (get_chain_last_ref_at): Rename into...
2885 (get_chain_last_write_at): ...this.
2886 (get_chain_last_write_before_load): New function.
2887 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
2888 CT_STORE_STORE when write reference is added.
2889 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
2890 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
2891 (initialize_root_vars_store_elim_1): Ditto.
2892 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
2893 definition is created.
2894 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
2895 chain by replacing it with dominant stored value.
2897 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2899 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
2901 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
2903 * vr-values.c (extract_range_from_binary_expr): Use a full range
2906 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
2908 * config/arm/arm.md (R4_REGNUM): Define constant.
2909 (nonsecure_call_internal): Remove r4 clobber.
2910 (nonsecure_call_value_internal): Likewise.
2911 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
2912 clobber and resequence match_operands.
2913 (nonsecure_call_value_reg_thumb1_v5): Likewise.
2914 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
2915 (nonsecure_call_value_reg_thumb2): Likewise.
2917 2017-11-20 Jakub Jelinek <jakub@redhat.com>
2919 PR tree-optimization/78821
2920 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
2921 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
2922 to the address of the base rather than the base itself.
2923 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
2924 (find_bswap_or_nop_finalize): New function.
2925 (find_bswap_or_nop): Use it.
2926 (bswap_replace): Return a tree rather than bool, change first
2927 argument from gimple * to gimple_stmt_iterator, allow inserting
2928 into an empty sequence, allow ins_stmt to be NULL - then emit
2929 all stmts into gsi. Fix up MEM_REF address gimplification.
2930 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
2931 (struct store_immediate_info): Add N and INS_STMT non-static
2933 (store_immediate_info::store_immediate_info): Initialize them
2934 from newly added ctor args.
2935 (merged_store_group::apply_stores): Formatting fixes. Sort by
2937 (stmts_may_clobber_ref_p): For stores call also
2938 refs_anti_dependent_p.
2939 (gather_bswap_load_refs): New function.
2940 (imm_store_chain_info::try_coalesce_bswap): New method.
2941 (imm_store_chain_info::coalesce_immediate_stores): Use it.
2942 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
2943 (imm_store_chain_info::output_merged_store): Fail if number of
2944 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
2945 and NOP_EXPR rhs_code.
2946 (pass_store_merging::process_store): Compute n and ins_stmt, if
2947 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
2948 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
2950 (pass_store_merging::execute): Calculate dominators.
2952 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
2953 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
2954 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
2955 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
2956 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
2957 class pass_optimize_bswap, bswap_replace,
2958 pass_optimize_bswap::execute): Moved to ...
2959 * gimple-ssa-store-merging.c: ... this file.
2960 Include optabs-tree.h.
2961 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
2962 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
2963 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
2964 anonymous namespace, remove static keywords.
2965 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
2966 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
2968 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
2971 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
2973 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
2975 * vec.h (debug_helper): New function.
2976 (DEFINE_DEBUG_VEC): New macro.
2977 * hash-set.h (debug_helper): New function.
2978 (DEFINE_DEBUG_HASH_SET): New macro.
2979 * cfg.c (debug_slim (edge)): New function.
2980 Call DEFINE_DEBUG_VEC for edges.
2981 Call DEFINE_DEBUG_HASH_SET for edges.
2982 * cfghooks.c (debug_slim (basic_block)): New function.
2983 Call DEFINE_DEBUG_VEC for basic blocks.
2984 Call DEFINE_DEBUG_HASH_SET for basic blocks.
2985 * print-tree.c (debug_slim): New function to handle trees.
2986 Call DEFINE_DEBUG_VEC for trees.
2987 Call DEFINE_DEBUG_HASH_SET for trees.
2988 (debug (vec<tree, va_gc>) &): Remove.
2989 (debug (<vec<tree, va_gc>) *): Remove.
2990 * print-rtl.c (debug_slim): New function to handle const_rtx.
2991 Call DEFINE_DEBUG_VEC for rtx_def.
2992 Call DEFINE_DEBUG_VEC for rtx_insn.
2993 Call DEFINE_DEBUG_HASH_SET for rtx_def.
2994 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
2995 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
2996 (debug (vec<rtx_insn *> *ptr): Remove.
2997 (debug_insn_vector): Remove.
2998 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
3000 2017-11-20 Tom de Vries <tom@codesourcery.com>
3002 PR rtl-optimization/82020
3003 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
3004 IF_THEN_ELSE condition.
3006 2017-11-19 Jeff Law <law@redhat.com>
3008 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
3009 of degenerates resulting from ignoring an edge.
3011 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3014 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
3016 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3019 * profile-count.c (profile_count::to_sreal_scale): Fix return value
3020 for uninitialied counts.
3022 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3025 * tree-inline.c (estimate_num_insns): Set to 1 at least.
3027 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3030 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
3032 2017-11-19 Tom de Vries <tom@codesourcery.com>
3034 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
3037 2017-11-19 Tom de Vries <tom@codesourcery.com>
3039 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
3042 2017-11-19 Tom de Vries <tom@codesourcery.com>
3044 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
3047 2017-11-19 Tom de Vries <tom@codesourcery.com>
3049 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
3051 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
3053 2017-11-19 Tom de Vries <tom@codesourcery.com>
3055 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
3057 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
3058 semicolon after MCORE_EXPORT_NAME call.
3060 2017-11-19 Tom de Vries <tom@codesourcery.com>
3063 * vmsdbgout.c (vmsdbgout_early_finish): New function.
3064 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
3066 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
3068 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
3069 (cgraph_edge::clone): Cleanup updating of profile.
3070 * ipa-cp.c (update_profiling_info): Likewise.
3071 * ipa-inline-transform.c (inline_transform): Likewise.
3072 * ipa-inline.c (inline_small_functions): Add missing space to dump.
3073 * ipa-split.c (execute_split_functions): Do not split when function
3075 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
3076 * profile-count.c (profile_count::dump): Add global0.
3077 (profile_count::to_cgraph_frequency): Do not ICE when entry is
3079 (profile_count::to_sreal_scale): Likewise.
3080 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
3081 (profile_count::combine_with_ipa_count): New function.
3082 * profile-count.h (profile_guessed_global0adjusted): New.
3083 (profile_count::adjusted_zero): New.
3084 (profile_count::global0adjusted): New.
3085 (profile_count::combine_with_ipa_count): New.
3086 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
3087 correct profile of return block of split functions.
3088 (copy_cfg_body): Remove unused profile_count.
3089 (copy_body): Likewise.
3090 (expand_call_inline): Update.
3091 (tree_function_versioning): Update.
3093 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
3095 * hash-set.h (hash_set::empty): New.
3096 * tree-ssa-threadbackward.h: Delete.
3097 * tree-ssa-threadbackward.c (class thread_jumps): New.
3098 Move max_threaded_paths into class.
3099 (fsm_find_thread_path): Remove arguments that are now in class.
3100 (profitable_jump_thread_path): Rename to...
3101 (thread_jumps::profitable_jump_thread_path): ...this.
3102 (convert_and_register_jump_thread_path): Rename to...
3103 (thread_jumps::convert_and_register_current_path): ...this.
3104 (check_subpath_and_update_thread_path): Rename to...
3105 (thread_jumps::check_subpath_and_update_thread_path): ...this.
3106 (register_jump_thread_path_if_profitable): Rename to...
3107 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
3108 (handle_phi): Rename to...
3109 (thread_jumps::handle_phi): ...this.
3110 (handle_assignment): Rename to...
3111 (thread_jumps::handle_assignment): ...this.
3112 (fsm_find_control_statement_thread_paths): Rename to...
3113 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
3114 (find_jump_threads_backwards): Rename to...
3115 (thread_jumps::find_jump_threads_backwards): ...this.
3116 Initialize path local data.
3117 (pass_thread_jumps::execute): Call find_jump_threads_backwards
3118 from within thread_jumps class.
3119 (pass_early_thread_jumps::execute): Same.
3121 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3123 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
3125 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3127 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
3128 to canonical location.
3130 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3133 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
3135 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
3137 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
3139 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
3140 rcrt1.o%s/grcrt1.o%s for -static-pie.
3142 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3144 * i386.c (ix86_multiplication_cost, ix86_division_cost,
3145 ix86_shift_rotate_cost): Break out from ...
3146 (ix86_rtx_costs): ... here.
3147 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
3150 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3152 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
3153 when entry block was promoted unlikely.
3154 (estimate_bb_frequencies): Increase frequency scale.
3155 * profile-count.h (profile_count): Export precision info.
3157 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3159 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
3160 disturbin profile of entry block.
3162 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3164 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
3167 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3169 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
3170 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
3172 2017-11-17 Jeff Law <law@redhat.com>
3174 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
3175 from evrp_dom_walker class. Various methods moved into new class.
3176 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
3177 (evrp_range_analyzer::enter): New method.
3178 (evrp_range_analyzer::leave): New method.
3179 (evrp_dom_walker): Remove delegators no longer needed by this class.
3180 Replace vr_values data member with evrp_range_analyzer
3182 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
3183 method extracted from evrp_dom_walker::before_dom_children.
3184 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
3185 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
3187 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
3188 Add private copy constructor and move assignment operators.
3189 Privatize methods and class data where trivially possible.
3190 (evrp_dom_walker::cleanup): New function, extracted from
3191 execute_early_vrp. Simplify access to class data.
3193 * vr-values.h (get_output_for_vrp): Prototype.
3194 * vr-values.c (get_output_for_vrp): New function extracted from
3195 vrp_visit_assignment_or_call and extract_range_from_stmt.
3196 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
3198 2017-11-17 Luis Machado <luis.machado@linaro.org>
3200 * config/aarch64/aarch64.c
3201 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
3202 (qdf24xx_tunings) <autoprefetcher_model>: Set to
3203 tune_params::AUTOPREFETCHER_WEAK.
3205 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3208 * config/arm/arm.c (arm_valid_target_attribute_rec):
3209 Parse "arch=" and "+<ext>".
3210 (arm_valid_target_attribute_tree): Re-init global options.
3211 (arm_option_override): Make non-static.
3212 (arm_options_perform_arch_sanity_checks): Make errors fatal.
3213 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
3214 (__ARM_FEATURE_CRC32): Support undef.
3215 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
3216 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
3218 2017-11-17 David Malcolm <dmalcolm@redhat.com>
3220 * gdbinit.in (break-on-diagnostic): New command.
3222 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3224 * config/i386/i386.c (ix86_expand_epilogue): Change simple
3225 return to indirect jump for EH return if control-flow
3226 protection is enabled. Change explicit 'false' argument in
3227 pro_epilogue_adjust_stack with a value of
3229 * config/i386/i386.md (simple_return_indirect_internal):
3230 Remove SImode restriction to support 64-bit.
3232 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3234 * combine.c (added_notes_insn): New.
3235 (try_combine): Handle added_notes_insn like added_links_insn.
3236 Rewrite return value code.
3237 (distribute_notes): Set added_notes_insn to the earliest insn we added
3240 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3242 PR rtl-optimization/82621
3243 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
3244 dest of one of those SETs is unused.
3246 2017-11-17 Richard Biener <rguenther@suse.de>
3249 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
3250 * tree-pretty-print.c (dump_generic_node): Handle
3251 annot_expr_parallel_kind.
3252 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
3253 * gimplify.c (gimple_boolify): Likewise.
3255 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3257 * config.gcc (extra_headers): Add cet.h for x86 targets.
3258 * config/i386/cet.h: New file.
3259 * doc/install.texi: Add --enable-cet/--disable-cet.
3261 2017-11-17 Richard Biener <rguenther@suse.de>
3263 PR tree-optimization/83017
3264 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
3265 (gen_parallel_loop): Properly count iterations.
3266 (parallelize_loops): Handle loop->can_be_parallel independent
3267 of flag_loop_parallelize_all. Make static profitability test match
3269 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
3270 * invoke.texi (parloops-min-per-thread): Document.
3272 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
3274 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
3275 upstreaming review comments.
3277 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3279 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
3280 with fast unaligned access.
3281 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
3283 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
3285 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
3286 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
3288 2017-11-17 Richard Biener <rguenther@suse.de>
3290 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
3291 folding of references.
3293 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
3296 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
3297 of replacing call to strncmp with corresponding call to strcmp when
3300 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
3302 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
3303 option prefer-avx256 for skylake-avx512 configuration.
3304 * config/i386/i386.c (ix86_option_override_internal): Ditto.
3305 (get_builtin_code_for_version): Ditto.
3307 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3308 Monk Chiang <sh.chiang04@gmail.com>
3310 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
3311 (FIXED_REGISTERS): Reserve more register numbers.
3312 (CALL_USED_REGISTERS): Likewise.
3313 (REG_ALLOC_ORDER): Likewise.
3314 (REG_CLASS_CONTENTS): Likewise.
3315 (REGISTER_NAMES): Likewise.
3317 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3318 Kito Cheng <kito.cheng@gmail.com>
3320 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
3322 * config/nds32/iterators.md: Add vector mode iterators and attributes.
3324 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
3326 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
3327 parameter list for vec_splats.
3329 2017-11-16 Joseph Myers <joseph@codesourcery.com>
3331 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
3333 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
3335 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
3338 * doc/install.texi: Document incompatibility of Perl >=5.6.26
3339 with the required version of automake 1.11.6.
3341 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
3343 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
3345 (DU_C2_3_power9): Correct reservation combinations.
3346 (FP_DIV_power9, VEC_DIV_power9): New.
3347 (power9-alu): Split out rotate/shift...
3348 (power9-rot): ...to here, correct dispatch resource.
3349 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
3351 (power9-fp): Correct latency.
3352 (power9-sdiv): Add div/sqrt resource.
3353 (power9-ddiv): Correct latency, add div/sqrt resource.
3354 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
3355 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
3357 (power9-qpdiv, power9-qpmul): Adjust resource usage.
3359 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
3361 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
3362 switch statement mapping KF built-ins to TF built-ins if we don't
3363 have the proper ISA 3.0 assembler support.
3365 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3367 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
3368 (lower_emutls_function_body): Do not compute it.
3370 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3372 * ipa-split.c (split_bb_info): Turn time to sreal.
3373 (split_point): Likewise.
3374 (dump_split_point): Likewise.
3375 (fine_split_points): Likewise.
3376 (execute_split_functions): Only zero split_bbs; turn time to sreals.
3378 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3380 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
3383 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3385 * predict.c (combine_predictions_for_bb): Preserve zero predicted
3387 (expensive_function_p): Remove useless assert.
3388 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
3390 2017-11-16 Martin Sebor <msebor@redhat.com>
3392 PR tree-optimization/82588
3393 PR tree-optimization/82583
3394 * tree-vrp.c (check_array_ref): Handle flexible array members,
3395 string literals, and inner indices.
3396 (search_for_addr_array): Add detail to diagnostics.
3398 2017-11-16 Nathan Sidwell <nathan@acm.org>
3402 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
3403 (SET_DECL_ASSEMBLER_NAME): Forward to
3404 overwrite_decl_assembler_name.
3405 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
3406 (overwrite_decl_assembler_name): Declare.
3407 * tree.c (overwrite_decl_assembler_name): New.
3408 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
3409 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
3410 (LANG_HOOKS_INITIALIZER): Add it.
3411 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
3412 * langhooks.c (lhd_set_decl_assembler_name): Use
3413 SET_DECL_ASSEMBLER_NAME.
3414 (lhd_overwrite_decl_assembler_name): Default implementation.
3416 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
3417 Jackson Woodruff <jackson.woodruff@arm.com>
3419 PR tree-optimization/71026
3420 * match.pd: Canonicalize constant multiplies in division.
3422 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
3424 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
3425 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
3426 Remove OPT_fomit_frame_pointer entry.
3427 * common/config/arc/arc-common.c: Likewise.
3428 * common/config/arm/arm-common.c: Likewise.
3429 * common/config/avr/avr-common.c: Likewise.
3430 * common/config/c6x/c6x-common.c: Likewise.
3431 * common/config/cr16/cr16-common.c: Likewise.
3432 * common/config/cris/cris-common.c: Likewise.
3433 * common/config/epiphany/epiphany-common.c: Likewise.
3434 * common/config/fr30/fr30-common.c: Likewise.
3435 * common/config/frv/frv-common.c: Likewise.
3436 * common/config/ia64/ia64-common.c: Likewise.
3437 * common/config/iq2000/iq2000-common.c: Likewise.
3438 * common/config/lm32/lm32-common.c: Likewise.
3439 * common/config/m32r/m32r-common.c: Likewise.
3440 * common/config/mcore/mcore-common.c: Likewise.
3441 * common/config/microblaze/microblaze-common.c: Likewise.
3442 * common/config/mips/mips-common.c: Likewise.
3443 * common/config/mmix/mmix-common.c: Likewise.
3444 * common/config/mn10300/mn10300-common.c: Likewise.
3445 * common/config/nios2/nios2-common.c: Likewise.
3446 * common/config/pa/pa-common.c: Likewise.
3447 * common/config/pdp11/pdp11-common.c: Likewise.
3448 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
3449 * common/config/riscv/riscv-common.c: Likewise.
3450 * common/config/rs6000/rs6000-common.c: Likewise.
3451 * common/config/rx/rx-common.c: Likewise.
3452 * common/config/s390/s390-common.c: Likewise.
3453 * common/config/sh/sh-common.c: Likewise.
3454 * common/config/sparc/sparc-common.c: Likewise.
3455 * common/config/tilegx/tilegx-common.c: Likewise.
3456 * common/config/tilepro/tilepro-common.c: Likewise.
3457 * common/config/v850/v850-common.c: Likewise.
3458 * common/config/visium/visium-common.c: Likewise.
3459 * common/config/xstormy16/xstormy16-common.c: Likewise.
3460 * common/config/xtensa/xtensa-common.c: Likewise.
3461 * invoke.texi (-fomit-frame-pointer): Update documentation.
3463 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3465 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
3467 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3469 * cfg.c (scale_bbs_frequencies_int,
3470 cale_bbs_frequencies_gcov_type): Remove.
3471 * cfg.h (scale_bbs_frequencies_int,
3472 cale_bbs_frequencies_gcov_type): Remove.
3474 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3476 * tree-ssa-loop-manip.c
3477 (scale_dominated_blocks_in_loop): Update to profile counts.
3478 (tree_transform_and_unroll_loop): Likewise.
3480 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3482 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
3483 scale_bbs_frequencies_int.
3485 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3487 * final.c (compute_alignments): Use counts rather than frequencies.
3489 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3491 * cfgloopanal.c: Include sreal.h
3492 (average_num_loop_insns): Use counts and sreal for accounting.
3494 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3496 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
3499 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3500 Kito Cheng <kito.cheng@gmail.com>
3502 * config/nds32/constraints.md: Provide more constraints.
3503 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
3504 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
3505 support constraints usage.
3507 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3509 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
3511 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3512 Kito Cheng <kito.cheng@gmail.com>
3514 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
3515 * config/nds32/nds32.opt: Refine the layout.
3516 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
3517 TARGET_EXT_STRING): Support new options.
3518 * config/nds32/nds32.h: Likewise.
3519 * config/nds32/nds32.md: Likewise.
3520 * config/nds32/nds32-predicates.c: Likewise.
3521 * config/nds32/constraints.md: Likewise.
3522 * common/config/nds32/nds32-common.c: Likewise.
3524 2017-11-16 Julia Koval <julia.koval@intel.com>
3527 * config/i386/gfniintrin.h: Add sse check.
3528 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
3530 2017-11-16 Julia Koval <julia.koval@intel.com>
3532 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
3533 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
3534 (ix86_handle_option): Handle -mavx512vbmi2.
3535 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
3536 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3537 * config/i386/i386-c.c (__AVX512VBMI2__): New.
3538 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
3539 (ix86_valid_target_attribute_inner_p): Ditto.
3540 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
3541 * config/i386/i386.opt (mavx512vbmi2): New option.
3542 * doc/invoke.texi: Add new option.
3544 2017-11-16 Julia Koval <julia.koval@intel.com>
3546 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
3547 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
3548 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
3549 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
3550 _mm512_gf2p8mul_epi8): New intrinsics.
3551 * config/i386/i386-builtin-types.def
3552 (V64QI_FTYPE_V64QI_V64QI): New type.
3553 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
3554 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
3555 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
3556 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
3557 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
3558 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3560 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
3562 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
3563 explicitly as a stream of bytes.
3565 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3567 * config/rs6000/altivec.h (vec_xst_be): New #define.
3568 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
3569 and externalize from *altivec_vperm_<mode>_internal.
3570 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
3572 (XL_BE_V8HI): Likewise.
3573 (XL_BE_V4SI): Likewise.
3574 (XL_BE_V4SI): Likewise.
3575 (XL_BE_V2DI): Likewise.
3576 (XL_BE_V4SF): Likewise.
3577 (XL_BE_V2DF): Likewise.
3578 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
3579 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
3580 all array entries with these keys: VSX_BUILTIN_VEC_XL,
3581 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
3582 VSX_BUILTIN_VEC_XST_BE.
3583 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
3584 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
3586 (altivec_init_builtins): Replace conditional calls to def_builtin
3587 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
3588 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
3589 with unconditional calls. Remove calls to def_builtin for
3590 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
3591 __builtin_vec_xst_be.
3592 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
3593 to define_expand, and add alternate RTL generation for P8.
3594 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
3595 vsx_ld_elemrev_v8hi.
3596 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
3597 add alternate RTL generation for P8.
3598 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
3599 vsx_ld_elemrev_v16qi.
3600 (vsx_st_elemrev_v8hi): Convert define_insn
3601 to define_expand, and add alternate RTL generation for P8.
3602 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
3603 vsx_st_elemrev_v8hi.
3604 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
3605 add alternate RTL generation for P8.
3606 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
3607 vsx_st_elemrev_v16qi.
3609 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
3612 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
3613 TARGET_AVX512ER check.
3614 (ix86_option_override_internal): Set MASK_VZEROUPPER if
3615 neither -mzeroupper nor -mno-zeroupper is used and
3616 TARGET_EMIT_VZEROUPPER is set.
3617 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
3618 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
3620 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
3622 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3623 folding of vector compares.
3624 (fold_build_vec_cmp): New helper function.
3625 (fold_compare_helper): New helper function.
3626 (builtin_function_type): Add compare builtins to the list of functions
3627 having unsigned arguments. Cosmetic updates to comment indentation.
3628 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
3629 the not+eq combination.
3631 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3633 PR tree-optimization/82726
3634 PR tree-optimization/70754
3635 * tree-predcom.c (order_drefs_by_pos): New function.
3636 (combine_chains): Move code setting has_max_use_after to...
3637 (try_combine_chains): ...here. New parameter. Sort combined chains
3638 according to position information.
3639 (tree_predictive_commoning_loop): Update call to above function.
3640 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
3642 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3644 PR tree-optimization/82726
3646 2017-01-23 Bin Cheng <bin.cheng@arm.com>
3648 PR tree-optimization/70754
3649 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
3650 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
3651 combined stmt before it if not NULL.
3652 (combine_chains): Process refs reversely and compute dominance point
3656 2017-02-23 Bin Cheng <bin.cheng@arm.com>
3658 PR tree-optimization/79663
3659 * tree-predcom.c (combine_chains): Process refs in reverse order
3660 only for ZERO length chains, and add explaining comment.
3662 2017-11-15 Tamar Christina <tamar.christina@arm.com>
3664 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
3665 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
3666 * doc/invoke.texi (ARM Options): Add armv8.3-a.
3668 2017-11-15 Tamar Christina <tamar.christina@arm.com>
3670 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
3672 2017-11-15 Martin Liska <mliska@suse.cz>
3674 * tree-cfg.c (pass_warn_function_return::execute):
3675 Compare warn_return_type for greater than zero.
3677 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
3681 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
3682 to return true on Xeon and not on Xeon Phi.
3683 (ix86_check_avx256_register): Changed to ...
3684 (ix86_check_avx_upper_register): ... this. Add extra check for
3685 VALID_AVX512F_REG_OR_XI_MODE.
3686 (ix86_avx_u128_mode_needed): Changed
3687 ix86_check_avx256_register to ix86_check_avx_upper_register.
3688 (ix86_check_avx256_stores): Changed to ...
3689 (ix86_check_avx_upper_stores): ... this. Changed
3690 ix86_check_avx256_register to ix86_check_avx_upper_register.
3691 (ix86_avx_u128_mode_after): Changed
3692 avx_reg256_found to avx_upper_reg_found. Changed
3693 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
3694 (ix86_avx_u128_mode_entry): Changed
3695 ix86_check_avx256_register to ix86_check_avx_upper_register.
3696 (ix86_avx_u128_mode_exit): Ditto.
3697 * config/i386/i386.h: (host_detect_local_cpu): New define.
3699 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3701 * config/arm/xgene1.md (xgene1): Split into automatons
3702 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
3703 (xgene1_f_load): Adjust reservations and/or types.
3704 (xgene1_f_store): Likewise.
3705 (xgene1_load_pair): Likewise.
3706 (xgene1_store_pair): Likewise.
3707 (xgene1_fp_load1): Likewise.
3708 (xgene1_load1): Likewise.
3709 (xgene1_store1): Likewise.
3710 (xgene1_move): Likewise.
3711 (xgene1_alu): Likewise.
3712 (xgene1_simd): Likewise.
3713 (xgene1_bfm): Likewise.
3714 (xgene1_neon_load1): Likewise.
3715 (xgene1_neon_store1): Likewise.
3716 (xgene1_neon_logic): Likewise.
3717 (xgene1_neon_st1): Likewise.
3718 (xgene1_neon_ld1r): Likewise.
3719 (xgene1_alu_cond): Added.
3720 (xgene1_shift_reg): Likwise.
3721 (xgene1_bfx): Likewise.
3722 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
3724 2017-11-15 Jakub Jelinek <jakub@redhat.com>
3727 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
3729 (can_widen_mult_without_libcall): New function.
3730 (expand_mul_overflow): If only checking unsigned mul overflow,
3731 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
3732 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
3733 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
3734 (expand_DIVMOD): Formatting fix.
3735 * expmed.h (expand_mult): Add NO_LIBCALL argument.
3736 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
3737 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
3739 PR tree-optimization/82977
3740 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
3741 constructed temporary to strlen_to_stridx.put.
3743 2017-11-15 Martin Liska <mliska@suse.cz>
3745 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
3746 * configure: Regenerate.
3748 2017-11-15 Martin Liska <mliska@suse.cz>
3751 * config/sh/sh-mem.cc: Use proper probability for
3754 2017-11-14 Jeff Law <law@redhat.com>
3756 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
3757 the red zone for stack_clash_protection_final_dynamic_probe targets
3758 when the total dynamic stack size is zero bytes.
3760 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
3761 blocks is post order.
3763 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
3765 * dumpfile.h (TDF_COMPARE_DEBUG): New.
3766 * final.c (rest_of_clean_state): Set it for the
3767 -fcompare-debug dump.
3768 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
3769 class when TDF_COMPARE_DEBUG is set.
3771 * dwarf2out.c (gen_producer_string): Discard
3774 2017-11-15 Joseph Myers <joseph@codesourcery.com>
3777 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
3778 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
3779 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
3780 (__TGMATH_REAL_2_3): Remove macros.
3781 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
3782 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
3784 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
3786 (remquo): Define using __TGMATH_REAL_3.
3788 2017-11-14 Jeff Law <law@redhat.com>
3790 * vr-values.c: New file with contents extracted from tree-vrp.c.
3791 * Makefile.in (OBJS): Add vr-values.o
3792 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
3793 (set_value_range, set_and_canonicalize_value_range): Likewise.
3794 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
3795 (value_range_constant_singleton, symbolic_range_p): Likewise.
3796 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
3797 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
3798 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
3799 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
3800 (find_case_label_range, find_case_label_index): Likewise.
3801 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
3802 (range_int_cst_singleton_p, value_inside_range): Likewise.
3803 (get_single_symbol): Likewise.
3804 (switch_update): Move structure definition here.
3805 (to_remove_edges, to_update_switch_stmts): Provide externs.
3806 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
3807 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
3808 (vrp_val_is_min, set_value_range): Likewise.
3809 (set_and_canonicalize_value_range, copy_value_range): Likewise.
3810 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
3811 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
3812 (range_is_nonnull, range_int_cst_p): Likewwise.
3813 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
3814 (get_single_symbol, operand_less_p): Likewise
3815 (compare_values_warnv, compare_values): Likewise.
3816 (value_inside_range, value_range_constant_singleton): Likewise.
3817 (zero_nonzero_bitgs_from_vr): Likewise.
3818 (extract_range_from_binary_expr_1): Likewise.
3819 (overflow_comparison_p): Likewise.
3820 (to_remove_edges, to_update_switch_stmts): Likewise.
3821 (find_case_label-index, find_case_label_range): Likewise.
3822 (switch_update, set_value_range_to_nonnegative): Remove.
3823 (set_value_range_to_truthvalue): Likewise.
3824 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
3825 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
3826 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
3827 (find_case_label_ranges, test_for_singularity): Likewise.
3828 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
3829 (x_vr_values): Move to its remaining use site.
3831 2017-11-10 Jeff Law <law@redhat.com>
3833 * vr-values.h (VR_INITIALIZER): Move #define here.
3834 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
3835 * Makefile.in (OBJS): Add tree-evrp.o
3836 * tree-vrp.h (assert_info): Move structure definition here.
3837 (set_value_range_to_varying): Prototype.
3838 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3839 (infer_value_range, register_edge_assert_for): Likewise.
3840 (stmt_interesting_for_vrp): Likewise.
3841 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
3842 (set_value_range_to_varying): No longer static.
3843 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3844 (infer_value_range, register_edge_assert_for): Likewise.
3846 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
3848 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
3849 generate the XXBRD instruction.
3851 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
3852 (rs6000_builtin_type_compatible): Treat _Float128 and long double
3853 as being compatible if -mabi=ieeelongdouble.
3854 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
3855 to setup float128 built-ins with hardware support.
3856 (BU_FLOAT128_HW_2): Likewise.
3857 (BU_FLOAT128_HW_3): Likewise.
3858 (BU_FLOAT128_HW_VSX_1): Likewise.
3859 (BU_FLOAT128_HW_VSX_2): Likewise.
3860 (scalar_extract_expq): Change float128 built-in functions to
3861 accommodate having both KFmode and TFmode functions. Use the
3862 KFmode variant as the default.
3863 (scalar_extract_sigq): Likewise.
3864 (scalar_test_neg_qp): Likewise.
3865 (scalar_insert_exp_q): Likewise.
3866 (scalar_insert_exp_qp): Likewise.
3867 (scalar_test_data_class_qp): Likewise.
3868 (sqrtf128_round_to_odd): Delete processing the round to odd
3869 built-in functions as special built-in functions, and define them
3870 as float128 built-ins. Use the KFmode variant as the default.
3871 (truncf128_round_to_odd): Likewise.
3872 (addf128_round_to_odd): Likewise.
3873 (subf128_round_to_odd): Likewise.
3874 (mulf128_round_to_odd): Likewise.
3875 (divf128_round_to_odd): Likewise.
3876 (fmaf128_round_to_odd): Likewise.
3877 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
3878 support for KFmode and TFmode xststdcqp calls.
3879 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
3880 point, switch the built-in handlers for the get/set float128
3881 exponent, get float128 mantissa, float128 test built-ins, and the
3882 float128 round to odd built-in functions. Eliminate creating the
3883 float128 round to odd built-in functions as special built-ins.
3884 (rs6000_init_builtins): Eliminate special creation of the float128
3885 round to odd built-in functions.
3886 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
3887 function insns to support both TFmode and KFmode variants.
3888 (xsxsigqp_<mode>): Likewise.
3889 (xsiexpqpf_<mode>): Likewise.
3890 (xsiexpqp_<mode>): Likewise.
3891 (xststdcqp_<mode>): Likewise.
3892 (xststdcnegqp_<mode>): Likewise.
3893 (xststdcqp_<mode>): Likewise.
3895 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3897 * tree-ssa-threadupdate.c (compute_path_counts): Remove
3898 unused path_in_freq_ptr parameter.
3899 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
3901 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3903 * ipa-inline.c (edge_badness): Dump sreal frequency.
3904 (compute_inlined_call_time): Match natural implementaiton ...
3905 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
3906 forgotten division by CGRAPH_FREQ_BASE.
3908 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3910 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
3911 Solaris 11. Update comment.
3912 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
3914 * config/sol2.h (STARTFILE_SPEC): Likewise.
3915 * configure: Regenerate.
3917 2017-11-14 Carl Love <cel@us.ibm.com>
3919 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
3920 le_ and be_ prefixes to swap* variables. Remove
3921 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
3924 2017-11-14 Jason Merrill <jason@redhat.com>
3926 Support GTY((cache)) on hash_map.
3927 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
3928 (ggc_cache_remove): Override it instead of ggc_mx.
3929 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
3930 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
3931 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
3932 (simple_cache_map_traits): Override maybe_mx.
3933 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
3934 (hash_map): Friend gt_cleare_cache.
3935 (gt_cleare_cache): New.
3936 * tree.h (tree_cache_traits): New hash_map traits class.
3937 (tree_cache_map): New typedef.
3939 2017-11-14 Richard Biener <rguenther@suse.de>
3941 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
3942 paramter and handling.
3943 (cleanup_control_flow_bb): Likewise.
3944 (cleanup_control_flow_pre): New helper performing a DFS walk
3945 to call cleanup_control_flow_bb in PRE order.
3946 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
3947 via cleanup_control_flow_pre.
3949 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
3951 * config/aarch64/aarch64-simd.md
3952 (aarch64_simd_bsl<mode>_internal): Remove DImode.
3953 (*aarch64_simd_bsl<mode>_alt): Likewise.
3954 (aarch64_simd_bsldi_internal): New.
3955 (aarch64_simd_bsldi_alt): Likewise.
3957 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3959 * tracer.c (better_p): Do not compare frequencies.
3960 * reg-stack.c (better_edge): Likewise.
3961 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
3964 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3966 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
3967 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
3968 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
3969 * lto-streamer-in.c (input_function): Use update_max_bb_count.
3970 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
3971 * predict.c (maybe_hot_frequency_p): Inline to ...
3972 (maybe_hot_count_p): ... here; rewrite to counts.
3973 (counts_to_freqs): Rename to ...
3974 (update_max_bb_count): ... this one.
3975 (expensive_function_p): Use counts.
3976 (estimate_bb_frequencies): Update.
3977 (rebuild_frequencies): Update.
3978 * predict.h (counts_to_freqs): Rename to ...
3979 (update_max_bb_count): ... this one.
3980 * profile.c (compute_branch_probabilities): Add debug info
3981 * tree-inline.c (expand_call_inline): Update debug info.
3982 (optimize_inline_calls): Use update_max_bb_count..
3983 (tree_function_versioning): Use update_max_bb_count..
3984 * value-prof.c (gimple_value_profile_transformations):
3985 Do not use update_max_bb_count.
3987 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3989 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
3990 always use frequencies.
3992 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3994 * bb-reorder.c: Remove frequencies from comments.
3995 (better_edge_p): Use profile counts.
3996 (find_traces): Dump profile counts.
3997 (rotate_loop): Use profile counts.
3998 (find_traces_1_round): Likewise.
3999 (connect_better_edge_p): Use counts instead of probabilities for
4001 (copy_bb_p): Drop early check for non-0 frequency.
4002 (sanitize_hot_paths): Update comments.
4004 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4006 * ipa-split.c (struct split_point): Add count.
4007 (consider_split): Do not compute incoming frequency; compute incoming
4008 count and store it to split_point.
4009 (split_function): Set count of the call to split part correctly.
4011 2017-11-13 Carl Love <cel@us.ibm.com>
4013 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
4015 2017-11-13 Tom Tromey <tom@tromey.com>
4017 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
4019 2017-11-13 Carl Love <cel@us.ibm.com>
4021 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
4022 Add support for builtins:
4023 unsigned int vec_first_{,miss}_match_{,or_eos}index,
4024 vector {un,}signed {char,int,short},
4025 vector {un,}signed {char,int,short}) arguments.
4026 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
4027 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
4028 Add BU_P9V_AV_2 expansions for the builtins.
4029 * config/rs6000/altivec.h (vec_first_match_index,
4030 vec_first_mismatch_index, vec_first_match_or_eos_index,
4031 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
4032 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
4033 new extern declaration.
4034 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
4035 * config/rs6000/vsx.md (first_match_index_<mode>,
4036 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
4037 first_mismatch_or_eos_index_<mode>): Add define expand.
4038 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
4039 * doc/extend.texi: Update the built-in documenation file for the new
4042 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
4044 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
4045 into the min/max operations for _Float<N> and _Float<N>X types.
4047 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4050 * dwarf2out.c (do_eh_frame): New static variable.
4051 (dwarf2out_begin_prologue): Set it.
4052 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
4054 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4056 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
4058 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
4060 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4063 * debug.h (dwarf2out_do_eh_frame): Declare.
4064 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
4065 (dwarf2out_do_frame): Use it.
4066 (dwarf2out_do_cfi_asm): Likewise.
4067 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
4068 (dwarf2out_assembly_start): Likewise.
4069 (dwarf2out_begin_prologue): Fix comment.
4070 * toplev.c (compile_file): Always call dwarf2out_frame_finish
4071 if the target needs either debug or unwind DWARF2 info.
4072 * lto-opts.c (lto_write_options): Do not save -fexceptions,
4073 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
4074 -ftrapping-math, -ftrapv and -fwrapv.
4076 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4078 * cgraph.c (cgraph_edge::sreal_frequency): New function.
4079 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
4080 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
4081 (estimate_edge_size_and_time): Likewise.
4082 (ipa_merge_fn_summary_after_inlining): Likewise.
4083 * ipa-inline.c (cgraph_freq_base_rec): Remove.
4084 (compute_uninlined_call_time): Use sreal_frequency.
4085 (compute_inlined_call_time): Likewise.
4086 (ipa_inline): Do not initialize cgraph_freq_base_rec.
4087 * profile-count.c: Include sreal.h.
4088 (profile_count::to_sreal_scale): New.
4089 * profile-count.h: Forward declare sreal.
4090 (profile_count::to_sreal_scale): Declare.
4092 2017-11-13 Nathan Sidwell <nathan@acm.org>
4094 * diagnostic.c (maybe_line_and_column): New.
4095 (diagnostic_get_location_text): Use it.
4096 (diagnostic_report_current_module): Likewise.
4097 (test_diagnostic_get_location_text): Add tests.
4099 2017-11-13 Luis Machado <luis.machado@linaro.org>
4101 * doc/md.texi (Specifying processor pipeline description): Fix
4102 incorrect latency for the div instruction example.
4104 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4106 PR tree-optimization/78821
4107 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
4108 that bit_not_p is the same.
4109 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
4110 (split_group): Count precisely bit_not_p bits in each statement.
4111 (invert_op): New function.
4112 (imm_store_chain_info::output_merged_store): Use invert_op to
4113 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
4114 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
4116 2017-11-13 Martin Liska <mliska@suse.cz>
4118 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
4119 (struct source_info): Likewise.
4120 (add_branch_counts): Likewise.
4121 (add_line_counts): Likewise.
4122 (function_summary): Likewise.
4123 (output_intermediate_line): Likewise.
4124 (generate_results): Likewise.
4126 2017-11-13 Martin Liska <mliska@suse.cz>
4128 * gcov.c (struct block_info): Remove typedef for block_t.
4129 (struct line_info): Likewise.
4130 (line_info::has_block): Likewise.
4131 (EXIT_BLOCK): Likewise.
4132 (unblock): Likewise.
4133 (circuit): Likewise.
4134 (get_cycles_count): Likewise.
4135 (process_file): Likewise.
4136 (read_graph_file): Likewise.
4137 (solve_flow_graph): Likewise.
4138 (find_exception_blocks): Likewise.
4139 (add_line_counts): Likewise.
4140 (accumulate_line_info): Likewise.
4141 (output_line_details): Likewise.
4143 2017-11-13 Martin Liska <mliska@suse.cz>
4145 * gcov.c (struct arc_info): Remove typedef for arc_t.
4146 (struct line_info): Likewise.
4147 (add_branch_counts): Likewise.
4148 (output_branch_count): Likewise.
4149 (function_info::~function_info): Likewise.
4150 (circuit): Likewise.
4151 (output_intermediate_line): Likewise.
4152 (read_graph_file): Likewise.
4153 (solve_flow_graph): Likewise.
4154 (find_exception_blocks): Likewise.
4155 (add_line_counts): Likewise.
4156 (accumulate_line_info): Likewise.
4157 (output_line_details): Likewise.
4158 (output_function_details): Likewise.
4160 2017-11-13 Martin Liska <mliska@suse.cz>
4162 * gcov.c (struct function_info): Remove typedef for function_t.
4163 (struct source_info): Likewise.
4164 (source_info::get_functions_at_location): Likewise.
4165 (solve_flow_graph): Likewise.
4166 (find_exception_blocks): Likewise.
4167 (add_line_counts): Likewise.
4168 (output_intermediate_file): Likewise.
4169 (process_file): Likewise.
4170 (generate_results): Likewise.
4171 (release_structures): Likewise.
4172 (read_graph_file): Likewise.
4173 (read_count_file): Likewise.
4174 (accumulate_line_counts): Likewise.
4175 (output_lines): Likewise.
4177 2017-11-13 Martin Liska <mliska@suse.cz>
4179 * gcov.c (function_info::function_info): Remove num_counts
4180 and add vector<gcov_type>.
4181 (function_info::~function_info): Use the vector.
4182 (process_file): Likewise.
4183 (read_graph_file): Likewise.
4184 (read_count_file): Likewise.
4185 (solve_flow_graph): Likewise.
4187 2017-11-13 Martin Liska <mliska@suse.cz>
4189 * gcov.c (function_info::is_artificial): New function.
4190 (process_file): Erase all artificial early.
4191 (generate_results): Skip as all artificial are already
4194 2017-11-13 Martin Liska <mliska@suse.cz>
4196 * gcov.c (read_graph_file): Store to global vector of functions.
4197 (read_count_file): Iterate the vector.
4198 (process_file): Likewise.
4199 (generate_results): Likewise.
4200 (release_structures): Likewise.
4202 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4204 PR tree-optimization/82954
4205 * gimple-ssa-store-merging.c
4206 (imm_store_chain_info::coalesce_immediate_stores): If
4207 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
4209 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
4211 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
4212 Upddate call to ENDIAN_LANE_N.
4213 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
4214 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
4215 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
4216 and use aarch64_endian_lane_rtx.
4217 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
4219 2017-11-12 Tom de Vries <tom@codesourcery.com>
4221 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
4223 2017-11-12 Tom de Vries <tom@codesourcery.com>
4225 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
4227 2017-11-12 Tom de Vries <tom@codesourcery.com>
4229 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
4231 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4232 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4233 * defaults.h (ASM_OUTPUT_LABELREF): Same.
4235 2017-11-11 Martin Sebor <msebor@redhat.com>
4238 * doc/extend.texi (attribute nonstring): Remove spurious argument.
4241 * prefic.c (translate_name): Replace strncpy with memcpy to
4242 avoid -Wstringop-truncation.
4244 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4246 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
4248 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4250 * predict.c (maybe_hot_frequency_p): Do not use cfun.
4252 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4254 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
4257 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
4259 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
4260 enable generating XXBRH if the value is in a vector register.
4261 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
4262 value is in a vector register.
4263 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
4264 register to register bswap64's instead of doing the GPR sequence
4265 used on previous machines.
4266 (bswapdi2_xxbrd): New insn.
4267 (bswapdi2_reg): Disallow on ISA 3.0.
4268 (register to register bswap64 splitter): Do not split the insn on
4269 ISA 3.0 systems that use XXBRD.
4271 2017-11-10 Martin Sebor <msebor@redhat.com>
4274 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
4276 (find_subframework_file): Same.
4278 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4280 * auto-profile.c (afdo_indirect_call): Drop frequency.
4281 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
4282 (cgraph_node::create_edge): Drop frequency argument.
4283 (cgraph_node::create_indirect_edge): Drop frequency argument.
4284 (cgraph_edge::make_speculative): Drop frequency arguments.
4285 (cgraph_edge::resolve_speculation): Do not update frequencies
4286 (cgraph_edge::dump_edge_flags): Do not dump frequency.
4287 (cgraph_node::dump): Check consistency in IPA mode.
4288 (cgraph_edge::maybe_hot_p): Use IPA counter.
4289 (cgraph_edge::verify_count_and_frequency): Rename to ...
4290 (cgraph_edge::verify_count): ... this one; drop frequency checking.
4291 (cgraph_node::verify_node): Update.
4292 * cgraph.h (struct cgraph_edge): Drop frequency.
4293 (cgraph_edge::frequency): New function.
4294 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
4296 (cgraph_edge::rebuild_edges): Likewise.
4297 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
4298 (duplicate_thunk_for_node): Do not pass frequency.
4299 (cgraph_node::create_clone): Scale only counts.
4300 (cgraph_node::create_virtual_clone): Do not pass frequency.
4301 (cgraph_node::create_edge_including_clones): Do not pass frequency.
4302 (cgraph_node::create_version_clone): Do not pass frequency.
4303 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
4304 (cgraph_node::expand_thunk): Do not pass frequency.
4305 (cgraph_node::create_wrapper): Do not pass frequency.
4306 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
4308 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
4309 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
4310 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
4311 (gather_caller_stats): Use frequency function.
4312 (ipcp_cloning_candidate_p): Use frequency function.
4313 (ipcp_propagate_stage): Use frequency function.
4314 (get_info_about_necessary_edges): Use frequency function.
4315 (update_profiling_info): Update only IPA profile.
4316 (update_specialized_profile): Use frequency functoin.
4317 (perhaps_add_new_callers): Update only IPA profile.
4318 * ipa-devirt.c (ipa_devirt): Use IPA profile.
4319 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
4320 (dump_ipa_call_summary): Use frequency function.
4321 (estimate_edge_size_and_time): Use frequency function.
4322 (ipa_merge_fn_summary_after_inlining): Use frequency function.
4323 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
4324 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
4325 (update_noncloned_counts): ... ths one; scale counts only.
4326 (clone_inlined_nodes): Do not scale frequency.
4327 (inline_call): Do not pass frequency.
4328 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
4329 (compute_inlined_call_time): Use IPA profile.
4330 (want_inline_small_function_p): Use IPA profile.
4331 (want_inline_self_recursive_call_p): Use IPA profile.
4332 (edge_badness): Use IPA profile.
4333 (lookup_recursive_calls): Use IPA profile.
4334 (recursive_inlining): Do not pass frequency.
4335 (resolve_noninline_speculation): Do not update frequency.
4336 (inline_small_functions): Collect max of IPA profile.
4337 (dump_overall_stats): Dump IPA porfile.
4338 (dump_inline_stats): Dump IPA porfile.
4339 (ipa_inline): Collect IPA stats.
4340 * ipa-inline.h (clone_inlined_nodes): Update prototype.
4341 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
4342 (ipa_propagate_frequency): Use frequency function.
4343 (ipa_profile): Cleanup.
4344 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
4345 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
4346 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
4347 (input_node): Do not stream frequency.
4348 (input_edge): Do not stream frequency.
4349 (merge_profile_summaries): Scale only IPA profiles.
4350 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
4351 * predict.c (drop_profile): Do not recompute frequency.
4352 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
4353 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
4354 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
4355 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
4356 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
4357 * tree-inline.c (copy_bb): Do not scale frequency.
4358 (expand_call_inline): Do not scale frequency.
4359 (tree_function_versioning): Do not scale frequency.
4360 * ubsan.c (ubsan_create_edge): Do not pass frequency.
4362 2017-11-10 Julia Koval <julia.koval@intel.com>
4364 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
4365 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
4366 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
4367 (_mm256_maskz_gf2p8affine_epi64_epi8)
4368 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
4369 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
4370 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
4371 (__builtin_ia32_vgf2p8affineqb_v32qi)
4372 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
4373 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
4375 2017-11-10 Tamar Christina <tamar.christina@arm.com>
4379 (arm_option_override): Refactor.
4380 (arm_option_reconfigure_globals): New.
4381 (arm_options_perform_arch_sanity_checks): New.
4382 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
4384 (arm_options_perform_arch_sanity_checks): Likewise
4386 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
4388 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
4389 (power9-qpmul): New.
4390 * rs6000/rs6000.md ("type" attr): Add qmul.
4391 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
4392 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
4393 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
4395 2017-11-10 Martin Sebor <msebor@redhat.com>
4398 * builtins.c (compute_objsize): Handle arrays that
4399 compute_builtin_object_size likes to fail for. Make extern.
4400 * builtins.h (compute_objsize): Declare.
4401 (check_strncpy_sizes): New function.
4402 (expand_builtin_strncpy): Call check_strncpy_sizes.
4403 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
4404 -Wstringop-truncation.
4405 (gimple_fold_builtin_strncat): Same.
4406 * gimple.c (gimple_build_call_from_tree): Set call location.
4407 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
4408 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
4409 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
4410 (handle_builtin_strlen): Use strlen_to_stridx.
4411 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
4413 Use strlen_to_stridx.
4414 (pass_strlen::execute): Release strlen_to_stridx.
4415 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
4416 (-Wstringop-truncation): Document new option.
4418 2017-11-10 Martin Liska <mliska@suse.cz>
4420 PR gcov-profile/82702
4421 * gcov.c (main): Handle intermediate files in a different
4423 (get_gcov_intermediate_filename): New function.
4424 (output_gcov_file): Remove support of intermediate files.
4425 (generate_results): Allocate intermediate file.
4426 (release_structures): Clean-up properly fn_end.
4427 (output_intermediate_file): Start iterating with line 1.
4429 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4431 PR tree-optimization/82929
4432 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
4433 ops_swapped_p non-static data member.
4434 (store_immediate_info::store_immediate_info): Clear it.
4435 (imm_store_chain_info::coalesce_immediate_stores): If swapping
4436 ops set ops_swapped_p.
4437 (count_multiple_uses): Handle ops_swapped_p.
4439 2017-11-10 Martin Liska <mliska@suse.cz>
4441 * coverage.c (coverage_init): Stream information about
4442 support of has_unexecuted_blocks.
4443 * doc/gcov.texi: Document that.
4444 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
4445 * gcov.c (read_graph_file): Likewise.
4446 (output_line_beginning): Fix a small issue with
4449 2017-11-10 Bin Cheng <bin.cheng@arm.com>
4451 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
4452 reference of trivial component.
4454 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4457 * gimple-ssa-store-merging.c
4458 (pass_store_merging::terminate_all_aliasing_chains): For
4459 gimple_store_p stmts also call refs_output_dependent_p.
4461 PR rtl-optimization/82913
4462 * compare-elim.c (try_merge_compare): Punt if def_insn is not
4465 2017-11-09 Jeff Law <law@redhat.com>
4467 * vr-values.h: New file with vr_values class.
4468 * tree-vrp.c: Include vr-values.h
4469 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
4470 data objects into the vr_values class.
4471 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
4472 (get_value_range): Make it a member function within vr_values class.
4473 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
4474 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
4475 (op_with_constant_singleton_value_range): Likewise.
4476 (extract_range_for_var_from_comparison_expr): Likewise.
4477 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
4478 (extract_range_from_binary_expr): Likewise.
4479 (extract_range_from_unary_expr): Likewise.
4480 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
4481 (check_for_binary_op_overflow, extract_range_basic): Likewise.
4482 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
4483 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
4484 (compare_name_with_value, compare_names): Likewise.
4485 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
4486 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
4487 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
4488 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
4489 (extract_range_from_phi_node): Likewise.
4490 (simplify_truth_ops_using_ranges): Likewise.
4491 (simplify_div_or_mod_using_ranges): Likewise.
4492 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
4493 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
4494 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
4495 (simplify_float_conversion_using_ranges): Likewise.
4496 (simplify_internal_call_using_ranges): Likewise.
4497 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
4498 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
4499 poitner via x_vr_values for calls into gimple folder.
4500 (vrp_initialize_lattice): Make this the vr_values ctor.
4501 (vrp_free_lattice): Make this the vr_values dtor.
4502 (set_vr_value): New function.
4503 (class vrp_prop): Add vr_values data member. Add various member
4504 functions as well as member functions that delegate to vr_values.
4505 (check_array_ref): Make a member function within vrp_prop class.
4506 (search_for_addr_array, vrp_initialize): Likewise.
4507 (vrp_finalize): Likewise. Revamp to avoid direct access to
4508 vr_value, values_propagated, etc.
4509 (check_array_bounds): Extract vrp_prop class instance pointer from
4510 walk info structure. Use it to call member functions.
4511 (check_all_array_refs): Make a member function within vrp_prop class.
4512 Smuggle class instance pointer via walk info structure.
4513 (x_vr_values): New local static.
4514 (vrp_valueize): Use x_vr_values to get class instance.
4515 (vr_valueize_1): Likewise.
4516 (class vrp_folder): Add vr_values data member. Add various member
4517 functions as well as member functions that delegate to vr_values.
4518 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
4519 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
4520 class instance from vr_values. Use it to call member functions.
4521 (vrp_dom_walker): Add vr_values data member.
4522 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
4523 instance via x_vr_values.
4524 (identify_jump_threads): Accept vr_values as argument. Store
4525 it into the walker structure.
4526 (evrp_dom_walker): Add vr_values class data member. Add various
4528 (evrp_dom_walker::try_find_new_range): Use vr_values data
4529 member to access the memory allocator.
4530 (evrp_dom_walker::before_dom_children): Store vr_values class
4531 instance into the vrp_folder class.
4532 (evrp_dom_walker::push_value_range): Rework to avoid direct
4533 access to num_vr_values and vr_value.
4534 (evrp_dom_walker::pop_value_range): Likewise.
4535 (execute_early_vrp): Remove call to vrp_initialize_lattice.
4536 Use vr_values to get to dump_all_value_ranges member function.
4537 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
4538 and simplify_cond_using_ranges_2 via vrp_prop class instance.
4539 Pass vr_values class instance down to identify_jump_threads.
4540 Remove call to vrp_free_lattice.
4541 (debug_all_value_ranges): Remove.
4543 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
4544 (vrp_folder): Likewise.
4546 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
4547 Get it from the existing bitmap instead.
4548 (vrp_intersect_ranges_1): Likewise.
4550 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4552 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
4554 (store_immediate_info::store_immediate_info): Add bitnotp argument,
4555 set bit_not_p to it.
4556 (imm_store_chain_info::coalesce_immediate_stores): Break group
4557 if bit_not_p is different.
4558 (count_multiple_uses, split_group,
4559 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
4560 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
4561 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
4562 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
4563 to store_immediate_info ctor.
4565 2017-11-09 Jim Wilson <jimw@sifive.com>
4567 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
4568 (scan_prog_file): Likewise.
4570 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4572 * bb-reorder.c (max_entry_frequency): Remove.
4573 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
4574 connect_traces, push_to_next_round_p): Remove prototypes.
4575 (find_traces_1_round): Use counts only.
4576 (push_to_next_round_p): Likewise.
4577 (find_traces): Likewise.
4578 (rotate_loop): Likewise.
4579 (find_traces_1_round): Likewise.
4580 (connect_traces): Likewise.
4581 (edge_order): Likewise.
4583 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
4585 * config/arm/arm.c (output_return_instruction): Add comments to
4586 indicate requirement for cmse_nonsecure_entry return to account
4587 for the size of clearing instruction output here.
4588 (thumb_exit): Likewise.
4589 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
4590 return in hardfloat mode.
4592 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
4594 * config/rs6000/rs6000.c (machine_function): Add a bool,
4595 "toc_is_wrapped_separately".
4596 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
4597 if it wasn't explicitly set or unset, we are optimizing for speed, and
4598 doing separate shrink-wrapping.
4599 (rs6000_get_separate_components): Enable the TOC component if
4600 saving the TOC register in the prologue.
4601 (rs6000_components_for_bb): Handle the TOC component.
4602 (rs6000_emit_prologue_components): Store the TOC register where needed.
4603 (rs6000_set_handled_components): Mark TOC as handled, if handled.
4604 (rs6000_emit_prologue): Don't save the TOC if that is already done.
4606 2017-11-09 Martin Jambor <mjambor@suse.cz>
4608 * ipa-param-manipulation.c: New file.
4609 * ipa-param-manipulation.h: Likewise.
4610 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
4611 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
4612 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
4613 (ipa_parm_adjustment): Likewise.
4614 (ipa_parm_adjustment_vec): Likewise.
4615 (ipa_get_vector_of_formal_parms): Moved declaration to
4616 ipa-param-manipulation.h.
4617 (ipa_get_vector_of_formal_parm_types): Likewise.
4618 (ipa_modify_formal_parameters): Likewise.
4619 (ipa_modify_call_arguments): Likewise.
4620 (ipa_combine_adjustments): Likewise.
4621 (ipa_dump_param_adjustments): Likewise.
4622 (ipa_modify_expr): Likewise.
4623 (ipa_get_adjustment_candidate): Likewise.
4624 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
4625 ipa-param-manipulation.c.
4626 (ipa_get_vector_of_formal_parm_types): Likewise.
4627 (ipa_modify_formal_parameters): Likewise.
4628 (ipa_modify_call_arguments): Likewise.
4629 (ipa_modify_expr): Likewise.
4630 (get_ssa_base_param): Likewise.
4631 (ipa_get_adjustment_candidate): Likewise.
4632 (index_in_adjustments_multiple_times_p): Likewise.
4633 (ipa_combine_adjustments): Likewise.
4634 (ipa_dump_param_adjustments): Likewise.
4635 * tree-sra.c: Also include ipa-param-manipulation.h
4636 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
4639 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4640 Alan Hayward <alan.hayward@arm.com>
4641 David Sherwood <david.sherwood@arm.com>
4643 * doc/sourcebuild.texi (vect_masked_store): Document.
4645 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4646 Alan Hayward <alan.hayward@arm.com>
4647 David Sherwood <david.sherwood@arm.com>
4649 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
4651 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4652 Alan Hayward <alan.hayward@arm.com>
4653 David Sherwood <david.sherwood@arm.com>
4655 * doc/sourcebuild.texi (vect_variable_length): Document.
4657 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4658 Alan Hayward <alan.hayward@arm.com>
4659 David Sherwood <david.sherwood@arm.com>
4661 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
4663 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4664 Alan Hayward <alan.hayward@arm.com>
4665 David Sherwood <david.sherwood@arm.com>
4667 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
4669 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4670 Alan Hayward <alan.hayward@arm.com>
4671 David Sherwood <david.sherwood@arm.com>
4673 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
4674 previously undocumented selectors.
4675 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
4677 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4679 * doc/rtl.texi (const_vector): Say that elements can be
4680 const_wide_ints too.
4681 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
4682 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
4683 (gen_vec_duplicate): Use it instead of CONSTANT_P.
4684 * optabs.c (expand_vector_broadcast): Likewise.
4686 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4687 Alan Hayward <alan.hayward@arm.com>
4688 David Sherwood <david.sherwood@arm.com>
4690 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
4691 scaled index even if the unscaled address was invalid.
4692 Don't increase the complexity of using a scale in that case.
4694 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4695 Alan Hayward <alan.hayward@arm.com>
4696 David Sherwood <david.sherwood@arm.com>
4698 * doc/rtl.texi: Rewrite the subreg rules so that they partition
4699 the inner register into REGMODE_NATURAL_SIZE bytes rather than
4700 UNITS_PER_WORD bytes.
4701 * emit-rtl.c (validate_subreg): Divide subregs into blocks
4702 based on REGMODE_NATURAL_SIZE of the inner mode.
4703 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
4704 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
4705 * expmed.c (lowpart_bit_field_p): Divide the value up into
4706 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
4707 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
4708 whether something is likely to occupy more than one register.
4710 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4713 * ipa-inline-transform.c (update_noncloned_frequencies): Use
4714 profile_count::adjust_for_ipa_scaling.
4715 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
4716 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
4718 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
4720 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4722 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
4723 (split_group): Add total_orig and total_new arguments, estimate the
4724 number of statements related to the store group without store merging
4725 and with store merging.
4726 (imm_store_chain_info::output_merged_store): Adjust split_group
4727 callers, punt if estimated number of statements with store merging
4728 is not smaller than estimated number of statements without it.
4730 (handled_load): Remove has_single_use checks.
4731 (pass_store_merging::process_store): Likewise.
4733 2017-11-09 Richard Biener <rguenther@suse.de>
4735 PR tree-optimization/82902
4736 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
4738 2017-11-09 Martin Liska <mliska@suse.cz>
4741 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
4744 2017-11-09 Martin Liska <mliska@suse.cz>
4746 PR tree-optimization/82669
4747 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
4749 2017-11-09 Martin Liska <mliska@suse.cz>
4751 PR gcov-profile/48463
4752 * coverage.c (coverage_begin_function): Output also end locus
4753 of a function and information whether the function is
4755 * gcov-dump.c (tag_function): Parse and print the information.
4756 * gcov.c (INCLUDE_MAP): Add include.
4757 (INCLUDE_SET): Likewise.
4758 (struct line_info): Move earlier in the source file because
4759 of vector<line_info> in function_info structure.
4760 (line_info::line_info): Likewise.
4761 (line_info::has_block): Likewise.
4762 (struct source_info): Add new member index.
4763 (source_info::get_functions_at_location): New function.
4764 (function_info::group_line_p): New function.
4765 (output_intermediate_line): New function.
4766 (output_intermediate_file): Use the mentioned function.
4767 (struct function_start): New.
4768 (struct function_start_pair_hash): Likewise.
4769 (process_file): Add code that identifies group functions.
4770 Assign lines either to global or function scope.
4771 (generate_results): Skip artificial functions.
4772 (find_source): Assign index for each source file.
4773 (read_graph_file): Read new flag artificial and end_line.
4774 (add_line_counts): Assign it either to global of function scope.
4775 (accumulate_line_counts): Isolate core of the function to
4776 accumulate_line_info and call it for both function and global
4778 (accumulate_line_info): New function.
4779 (output_line_beginning): Fix GNU coding style.
4780 (print_source_line): New function.
4781 (output_line_details): Likewise.
4782 (output_function_details): Likewise.
4783 (output_lines): Iterate both source (global) scope and function
4785 (struct function_line_start_cmp): New class.
4786 * doc/gcov.texi: Reflect changes in documentation.
4788 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4791 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
4792 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
4793 and similarly for not instead of neg.
4795 2017-11-08 Andi Kleen <ak@linux.intel.com>
4797 * config/i386/i386.opt: Add -mforce-indirect-call.
4798 * config/i386/predicates.md: Check for flag_force_indirect_call.
4799 * doc/invoke.texi: Document -mforce-indirect-call
4801 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
4803 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
4805 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
4806 config/riscv/riscv.c (predict.h): New include.
4807 (riscv_slow_unaligned_access_p): No longer static.
4808 (riscv_block_move_straight): Add require.
4809 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
4811 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4814 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
4815 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
4816 nonimmediate_operand predicate for operand 1 instead of
4819 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4821 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
4823 * config/aarch64/constraints.md (Uml): New constraint.
4824 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
4827 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4829 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
4830 of two vec_duplicates into a vec_concat.
4832 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4834 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4835 Simplify vec_merge of vec_duplicate and vec_concat.
4836 * config/aarch64/constraints.md (Utq): New constraint.
4837 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
4840 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4842 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4843 Simplify vec_merge of vec_duplicate and const_vector.
4844 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
4846 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
4847 mode iterator. Update predicate on operand 1 to
4848 handle non-const_vec constants. Delete constraints.
4849 (*aarch64_combinez_be<mode>): Likewise for operand 2.
4851 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4853 PR tree-optimization/78821
4854 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
4856 (store_operand_info::store_operand_info): Initialize it to false.
4857 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
4858 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
4859 store in the group, and if chain_info is non-NULL, to ignore altogether
4861 (compatible_load_p): Fail if bit_not_p does not match.
4862 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
4863 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
4864 (pass_store_merging::process_store): Adjust
4865 terminate_all_aliasing_chains calls to pass NULL in all current spots,
4866 call terminate_all_aliasing_chains newly when adding a store into
4867 a chain with non-NULL chain_info.
4869 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
4871 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
4873 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
4875 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4877 (aarch64_layout_frame): Initialise emit_frame_chain.
4878 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
4879 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
4881 2017-11-08 Martin Liska <mliska@suse.cz>
4883 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
4884 of gimple_call_internal_p.
4886 2017-11-07 Tom de Vries <tom@codesourcery.com>
4888 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
4890 2017-11-07 Tom de Vries <tom@codesourcery.com>
4892 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
4895 2017-11-08 Martin Liska <mliska@suse.cz>
4898 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
4900 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
4902 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
4904 2017-11-07 Jakub Jelinek <jakub@redhat.com>
4907 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
4908 (*cmp<mode>_ccz_1): New insn with $k alternative.
4911 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
4912 RTX_COMM_COMPARE as commutative as well.
4913 (ix86_binary_operator_ok): Formatting fix.
4914 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
4915 *<code><mode>3<mask_name><round_saeonly_name>,
4916 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
4917 *<s>mul<mode>3_highpart<mask_name>,
4918 *vec_widen_umult_even_v16si<mask_name>,
4919 *vec_widen_umult_even_v8si<mask_name>,
4920 *vec_widen_umult_even_v4si<mask_name>,
4921 *vec_widen_smult_even_v16si<mask_name>,
4922 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
4923 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
4924 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
4925 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
4926 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
4927 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
4928 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
4929 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
4930 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
4931 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
4932 ix86_binary_operator_ok. Formatting fixes.
4933 (*<plusminus_insn><mode>3<mask_name><round_name>,
4934 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
4937 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
4939 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
4940 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
4941 eq and ne if TARGET_ISEL.
4942 (cmp): New code_iterator.
4943 (UNS, UNSU_, UNSIK): New code_attrs.
4944 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
4945 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
4947 ("eq<mode>3"): ... this.
4948 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
4950 ("ne<mode>3"): ... this.
4952 2017-11-07 Julia Koval <julia.koval@intel.com>
4955 * common/config/i386/i386-common.c
4956 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
4957 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
4958 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
4959 * config/i386/i386.opt: Ditto.
4960 * config/i386/i386.c (ix86_target_string): Ditto.
4961 (ix86_option_override_internal): Ditto.
4962 (ix86_init_mpx_builtins): Move MPX to args2.
4963 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
4964 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
4965 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
4966 __builtin_ia32_vgf2p8affineinvqb_v32qi,
4967 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
4968 __builtin_ia32_vgf2p8affineinvqb_v16qi,
4969 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
4971 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
4974 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
4975 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
4976 (zero-extendsidi peephole2): Remove peephole.
4978 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
4981 * stor-layout.c: Include attribs.h.
4982 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
4983 explicit lookup of "aligned" attribute.
4985 2017-11-07 Andrew Waterman <andrew@sifive.com>
4987 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
4988 (riscv_expand_block_move): Likewise.
4989 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
4991 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
4992 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
4993 * config/riscv/riscv.c (riscv_block_move_straight): New function.
4994 (riscv_adjust_block_mem): Likewise.
4995 (riscv_block_move_loop): Likewise.
4996 (riscv_expand_block_move): Likewise.
4997 * config/riscv/riscv.md (movmemsi): New pattern.
4999 2017-11-07 Michael Clark <michaeljclark@mac.com>
5001 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
5002 (MUSL_DYNAMIC_LINKER): Likewise.
5004 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
5006 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
5007 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
5010 2017-11-07 Richard Biener <rguenther@suse.de>
5012 * match.pd: Fix build.
5014 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
5015 Jackson Woodruff <jackson.woodruff@arm.com>
5017 PR tree-optimization/71026
5018 * match.pd: Canonicalize negate in division.
5020 2017-11-07 Sudakshina Das <sudi.das@arm.com>
5023 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
5025 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5027 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
5028 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
5030 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5032 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
5033 non-scalar integral types.
5034 * match.pd (negate_expr_p): Handle MINUS_EXPR.
5035 (-(A-B), -(~A)): New transformations.
5037 2017-11-07 Tom de Vries <tom@codesourcery.com>
5039 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
5040 semicolon after "do {} while (0)".
5041 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5042 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5043 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5044 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5045 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5047 2017-11-07 Tom de Vries <tom@codesourcery.com>
5049 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
5050 after "do {} while (0)".
5051 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5052 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5053 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5054 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5055 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5057 2017-11-07 Tom de Vries <tom@codesourcery.com>
5060 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
5062 (arm_rtx_costs_internal): Add missing semicolon after
5063 HANDLE_NARROW_SHIFT_ARITH call.
5065 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5067 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
5068 disable isel if it was not set explicitly.
5070 2017-11-06 James Bowman <james.bowman@ftdichip.com>
5072 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
5073 (add_type_attribute) likewise.
5075 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
5077 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
5078 of struct ix86_frame.
5079 (ix86_initial_elimination_offset): Likewise.
5080 (ix86_expand_split_stack_prologue): Likewise.
5082 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
5084 * tree-vrp.h (enum value_range_type): Update stale comment.
5086 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5087 Alan Hayward <alan.hayward@arm.com>
5088 David Sherwood <david.sherwood@arm.com>
5090 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
5091 (aarch64_expand_vec_perm_const): Take the number of units too.
5092 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
5093 (aarch64_expand_vec_perm_const): Likewise.
5094 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
5095 (vec_perm<mode>): Update accordingly.
5097 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5098 Alan Hayward <alan.hayward@arm.com>
5099 David Sherwood <david.sherwood@arm.com>
5101 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
5102 Take the number of units too.
5103 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
5104 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
5105 but check for a vector mode before rather than after the call.
5106 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
5107 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
5108 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
5109 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
5110 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
5111 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
5112 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
5113 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
5114 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
5115 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
5116 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
5117 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
5118 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
5119 (aarch64_sqdmull2_n<mode>): Update accordingly.
5121 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5122 Alan Hayward <alan.hayward@arm.com>
5123 David Sherwood <david.sherwood@arm.com>
5125 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
5126 the number of units too.
5127 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
5128 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
5129 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
5130 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
5131 (vec_store_lanesxi<mode>): Update accordingly.
5133 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5134 Alan Hayward <alan.hayward@arm.com>
5135 David Sherwood <david.sherwood@arm.com>
5137 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
5138 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
5139 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
5140 of units rather than the mode.
5141 * config/aarch64/iterators.md (nunits): New mode attribute.
5142 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
5143 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
5144 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
5145 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
5146 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
5147 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
5148 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
5149 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
5150 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
5151 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
5152 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5153 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
5154 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
5155 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
5156 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
5157 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
5158 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
5159 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
5160 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
5161 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
5162 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
5163 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
5164 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
5165 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
5166 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
5167 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
5168 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
5169 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
5170 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
5171 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
5172 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
5173 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
5174 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
5175 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
5176 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
5177 (aarch64_simd_vec_setv2di): Likewise.
5179 2017-11-06 Carl Love <cel@us.ibm.com>
5181 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
5183 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
5184 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
5186 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
5188 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
5189 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
5190 Add power 8 macro expansions.
5191 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
5192 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
5193 power 8 instructions. (VSX_XXBR): Add iterator.
5195 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
5197 * config/arm/arm.md (predicable_short_it): Change default to "no",
5198 improve documentation, remove uses that are identical to the default.
5199 (enabled_for_depr_it): Rename to enabled_for_short_it.
5200 * gcc/config/arm/arm-fixed.md (predicable_short_it):
5201 Remove default uses.
5202 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
5203 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
5204 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
5205 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
5207 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
5210 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
5211 float128 helper macros, which are no longer used after deleting
5212 the old 'q' built-in functions, and moving the round to odd
5213 built-in functions to being special built-in functions.
5214 (BU_FLOAT128_2): Likewise.
5215 (BU_FLOAT128_1_HW): Likewise.
5216 (BU_FLOAT128_2_HW): Likewise.
5217 (BU_FLOAT128_3_HW): Likewise.
5218 (FABSQ): Delete old 'q' built-in functions.
5219 (COPYSIGNQ): Likewise.
5220 (SQRTF128_ODD): Move round to odd built-in functions to be
5221 special built-in functions, so that we can handle
5222 -mabi=ieeelongdouble.
5223 (TRUNCF128_ODD): Likewise.
5224 (ADDF128_ODD): Likewise.
5225 (SUBF128_ODD): Likewise.
5226 (MULF128_ODD): Likewise.
5227 (DIVF128_ODD): Likewise.
5228 (FMAF128_ODD): Likewise.
5229 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
5230 built-in names to 'f128'.
5231 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
5232 old 'q' built-in functions, as the machine independent code for
5233 'f128' built-in functions handles this.
5234 (rs6000_expand_builtin): Add expansion for float128 round to odd
5235 functions, keying off on -mabi=ieeelongdouble of whether to use
5236 the KFmode or TFmode variant.
5237 (rs6000_init_builtins): Initialize the _Float128 round to odd
5239 * doc/extend.texi (PowerPC Built-in Functions): Document the old
5240 _Float128 'q' built-in functions are now mapped into the new
5241 'f128' built-in functions.
5243 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
5245 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
5246 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
5248 2017-11-06 David Malcolm <dmalcolm@redhat.com>
5251 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
5252 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
5253 * toplev.c: Include "ipa-fnsummary.h".
5254 (toplev::finalize): Call ipa_fnsummary_c_finalize.
5256 2017-11-06 Jakub Jelinek <jakub@redhat.com>
5258 PR tree-optimization/82838
5259 * gimple-ssa-store-merging.c
5260 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
5261 on a separate gimple_seq which is then appended to seq.
5263 2017-11-06 Jeff Law <law@redhat.com>
5266 * config/i386/i386.c (PROBE_INTERVAL): Remove.
5267 (get_probe_interval): New functions.
5268 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
5269 (ix86_adjust_stack_and_probe): Likewise.
5270 (output_adjust_stack_and_probe): Likewise.
5271 (ix86_emit_probe_stack_range): Likewise.
5272 (ix86_expand_prologue): Likewise.
5274 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5276 PR tree-optimization/82816
5277 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
5278 if the modes of the two types are the same.
5279 (convert_plusminus_to_widen): Likewise.
5281 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5283 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
5285 (usadv16qi): New define_expand.
5286 (usadv8hi): New define_expand.
5288 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
5291 * ipa-inline-transform.c (update_noncloned_frequencies): Always
5293 (inline_transform): Likewise.
5294 * predict.c (counts_to_freqs): Remove useless conditional.
5295 * profile-count.h (profile_count::apply_scale): Move sanity check.
5296 * tree-inline.c (copy_bb): Always scale.
5297 (copy_cfg_body): Likewise.
5299 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
5302 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
5304 (*cmp_ite0): Add enabled_for_depr_it attribute.
5305 (*cmp_ite1): Likewise.
5307 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5309 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
5312 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5314 * tree-vrp.c (vrp_int_const_binop): Return true on success and
5315 return the value by pointer.
5316 (extract_range_from_multiplicative_op_1): Update accordingly.
5317 Return as soon as an operation fails.
5319 2017-11-05 Tom de Vries <tom@codesourcery.com>
5322 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
5323 (DEF_SANITIZER_BUILTIN): ... here.
5324 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
5325 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
5327 2017-11-05 Tom de Vries <tom@codesourcery.com>
5330 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
5332 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
5333 ASM_OUTPUT_BEFORE_CASE_LABEL call.
5334 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
5336 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5337 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5338 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5340 2017-11-05 Tom de Vries <tom@codesourcery.com>
5343 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
5346 2017-11-04 Michael Clark <michaeljclark@mac.com>
5348 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
5349 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
5351 (*addsi3_extended): Likewise.
5352 (*addsi3_extended2): Likewise.
5353 (<optab>si3): Likewise.
5354 (<optab>di3): Likewise.
5355 (<optab><mode>3): Likewise.
5356 (<*optabe>si3_internal): Likewise.
5357 (zero_extendqi<SUPERQI:mode>2): Likewise.
5358 (*add<mode>hi3): Likewise.
5359 (*xor<mode>hi3): Likewise.
5360 (<optab>di3): Likewise.
5361 (*<optab>si3_extend): Likewise.
5362 (*sge<u>_<X:mode><GPR:mode>): Likewise.
5363 (*slt<u>_<X:mode><GPR:mode>): Likewise.
5364 (*sle<u>_<X:mode><GPR:mode>): Likewise.
5366 2017-11-04 Andrew Waterman <andrew@sifive.com>
5368 * config/riscv/riscv.c (riscv_option_override): Conditionally set
5369 TARGET_STRICT_ALIGN based upon -mtune argument.
5371 2017-11-04 Andrew Waterman <andrew@sifive.com>
5373 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
5375 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
5377 * config/i386/i386.c (choose_basereg): Use optional scratch
5378 register and add assertion.
5379 (x86_emit_outlined_ms2sysv_save): Use scratch register when
5380 needed, and don't allocate stack.
5381 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
5382 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
5383 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
5385 2017-11-03 Jeff Law <law@redhat.com>
5387 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
5388 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
5389 to probe at the start of a noreturn function.
5391 2017-11-03 Jakub Jelinek <jakub@redhat.com>
5393 PR tree-optimization/78821
5394 * gimple-ssa-store-merging.c: Update the file comment.
5395 (MAX_STORE_ALIAS_CHECKS): Define.
5396 (struct store_operand_info): New type.
5397 (store_operand_info::store_operand_info): New constructor.
5398 (struct store_immediate_info): Add rhs_code and ops data members.
5399 (store_immediate_info::store_immediate_info): Add rhscode, op0r
5400 and op1r arguments to the ctor, initialize corresponding data members.
5401 (struct merged_store_group): Add load_align_base and load_align
5403 (merged_store_group::merged_store_group): Initialize them.
5404 (merged_store_group::do_merge): Update them.
5405 (merged_store_group::apply_stores): Pick the constant for
5406 encode_tree_to_bitpos from one of the two operands, or skip
5407 encode_tree_to_bitpos if neither operand is a constant.
5408 (class pass_store_merging): Add process_store method decl. Remove
5409 bool argument from terminate_all_aliasing_chains method decl.
5410 (pass_store_merging::terminate_all_aliasing_chains): Remove
5411 var_offset_p argument and corresponding handling.
5412 (stmts_may_clobber_ref_p): New function.
5413 (compatible_load_p): New function.
5414 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
5415 if there is overlap and rhs_code is not INTEGER_CST. For
5416 non-overlapping stores terminate group if rhs is not mergeable.
5417 (get_alias_type_for_stmts): Change first argument from
5418 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
5419 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
5420 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
5422 (get_location_for_stmts): Change first argument from
5423 auto_vec<gimple *> & to vec<gimple *> &.
5424 (struct split_store): Remove orig_stmts data member, add orig_stores.
5425 (split_store::split_store): Create orig_stores rather than orig_stmts.
5426 (find_constituent_stmts): Renamed to ...
5427 (find_constituent_stores): ... this. Change second argument from
5428 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
5429 to info structures rather than the statements.
5430 (split_group): Rename ALLOW_UNALIGNED argument to
5431 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
5432 it. Adjust find_constituent_stores caller.
5433 (imm_store_chain_info::output_merged_store): Handle rhs_code other
5434 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
5435 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
5436 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
5437 (mem_valid_for_store_merging): New function.
5438 (handled_load): New function.
5439 (pass_store_merging::process_store): New method.
5440 (pass_store_merging::execute): Use process_store method. Adjust
5441 terminate_all_aliasing_chains caller.
5443 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5445 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
5446 Return true for more constants, symbols and label references.
5447 (aarch64_valid_floating_const): Remove unused function.
5449 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5452 * config/aarch64/aarch64.c (aarch64_layout_frame):
5453 Undo forcing of LR at bottom of frame.
5455 2017-11-03 Jeff Law <law@redhat.com>
5458 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
5459 for int_registers_saved.
5461 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
5462 extracted from tree-ssa-dom.c.
5463 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
5464 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
5465 (record_equivalences_from_incoming_edge): Add additional argument
5466 to single_pred_edge_ignoring_loop_edges call.
5467 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
5468 (uncprop_dom_walker::before_dom_children): Add additional argument
5469 to single_pred_edge_ignoring_loop_edges call.
5470 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
5471 single_pred_edge_ignoring_loop_edges rather than open coding.
5472 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
5474 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
5476 * match.pd (-(-A)): Rewrite.
5478 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5480 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
5481 (rs6000_emit_int_cmove): New declaration.
5482 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
5483 (rs6000_emit_sISEL): Delete.
5484 (rs6000_emit_int_cmove): Make non-static.
5485 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
5486 instead of rs6000_emit_sISEL.
5488 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
5490 * asan.c (create_cond_insert_point): Maintain profile.
5491 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
5493 * basic-block.h (struct basic_block_def): Remove frequency.
5494 (EDGE_FREQUENCY): Use to_frequency
5495 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
5497 (find_traces): Update to use to_frequency.
5498 (find_traces_1_round): Likewise; use only IPA counts.
5499 (bb_to_key): Likewise.
5500 (connect_traces): Use IPA counts only.
5501 (copy_bb_p): Update to use to_frequency.
5502 (fix_up_crossing_landing_pad): Likewise.
5503 (sanitize_hot_paths): Likewise.
5504 * bt-load.c (basic_block_freq): Likewise.
5505 * cfg.c (init_flow): Set count_max to uninitialized.
5506 (check_bb_profile): Remove frequencies; check counts.
5507 (dump_bb_info): Do not dump frequencies.
5508 (update_bb_profile_for_threading): Update counts only.
5509 (scale_bbs_frequencies_int): Likewise.
5510 (MAX_SAFE_MULTIPLIER): Remove.
5511 (scale_bbs_frequencies_gcov_type): Update counts only.
5512 (scale_bbs_frequencies_profile_count): Update counts only.
5513 (scale_bbs_frequencies): Update counts only.
5514 * cfg.h (struct control_flow_graph): Add count-max.
5515 (update_bb_profile_for_threading): Update prototype.
5516 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
5517 (find_many_sub_basic_blocks): Likewise.
5518 * cfgcleanup.c (try_forward_edges): Likewise.
5519 (try_crossjump_to_edge): Likewise.
5520 * cfgexpand.c (expand_gimple_cond): Likewise.
5521 (expand_gimple_tailcall): Likewise.
5522 (construct_init_block): Likewise.
5523 (construct_exit_block): Likewise.
5524 * cfghooks.c (verify_flow_info): Check consistency of counts.
5525 (dump_bb_for_graph): Do not dump frequencies.
5526 (split_block_1): Do not update frequencies.
5527 (split_edge): Do not update frequencies.
5528 (make_forwarder_block): Do not update frequencies.
5529 (duplicate_block): Do not update frequencies.
5530 (account_profile_record): Do not update frequencies.
5531 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
5532 for global heuristics.
5533 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
5534 (expected_loop_iterations_unbounded): Use counts only.
5535 * cfgloopmanip.c (scale_loop_profile): Simplify.
5536 (create_empty_loop_on_edge): Simplify
5538 (duplicate_loop_to_header_edge): Simplify
5539 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
5540 (update_br_prob_note): Take care of removing note when profile
5542 (relink_block_chain): Do not dump frequency.
5543 (rtl_account_profile_record): Use to_frequency.
5544 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
5545 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
5546 (cgraph_update_edges_for_call_stmt_node): Likewise.
5547 (cgraph_edge::verify_count_and_frequency): Update.
5548 (cgraph_node::verify_node): Temporarily disable frequency verification.
5549 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
5550 to_cgraph_frequency.
5551 (cgraph_edge::rebuild_edges): Convert to ipa counts.
5552 * cgraphunit.c (init_lowered_empty_function): Do not initialize
5554 (cgraph_node::expand_thunk): Update profile.
5555 * except.c (dw2_build_landing_pads): Do not update frequency.
5556 * final.c (compute_alignments): Use to_frequency.
5557 (dump_basic_block_info): Do not dump frequency.
5558 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
5559 (dump_gimple_bb_header): Do not dump frequency.
5560 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
5562 * gimple-streamer-in.c (input_bb): Do not stream frequency.
5563 * gimple-streamer-out.c (output_bb): Do not stream frequency.
5564 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
5565 (init_before_recovery): Do not update frequency.
5566 (sched_create_recovery_edges): Do not update frequency.
5567 * hsa-gen.c (convert_switch_statements): Do not update frequency.
5568 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
5569 (ipa_cp_c_finalize): Set max_count to uninitialized.
5570 * ipa-fnsummary.c (get_minimal_bb): Use counts.
5571 (param_change_prob): Use counts.
5572 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
5574 * ipa-split.c (consider_split): Use to_frequency.
5575 (split_function): Use to_frequency.
5576 * ira-build.c (loop_compare_func): Likewise.
5577 (mark_loops_for_removal): Likewise.
5578 (mark_all_loops_for_removal): Likewise.
5579 * loop-doloop.c (doloop_modify): Do not update frequency.
5580 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
5582 * lto-streamer-in.c (input_function): Update count_max.
5583 * omp-expand.c (expand_omp_taskreg): Update count_max.
5584 * omp-simd-clone.c (simd_clone_adjust): Update profile.
5585 * predict.c (maybe_hot_frequency_p): Use to_frequency.
5586 (maybe_hot_count_p): Use ipa counts only.
5587 (maybe_hot_bb_p): Simplify.
5588 (maybe_hot_edge_p): Simplify.
5589 (probably_never_executed): Do not take frequency argument.
5590 (probably_never_executed_bb_p): Do not pass frequency.
5591 (probably_never_executed_edge_p): Likewise.
5592 (combine_predictions_for_bb): Check that profile is nonzero.
5593 (propagate_freq): Do not set frequency.
5594 (drop_profile): Simplify.
5595 (counts_to_freqs): Simplify.
5596 (expensive_function_p): Use to_frequency.
5597 (propagate_unlikely_bbs_forward): Simplify.
5598 (determine_unlikely_bbs): Simplify.
5599 (estimate_bb_frequencies): Add hack to silence graphite issues.
5600 (compute_function_frequency): Use ipa counts.
5601 (pass_profile::execute): Update.
5602 (rebuild_frequencies): Use counts only.
5603 (force_edge_cold): Use counts only.
5604 * profile-count.c (profile_count::dump): Dump new count types.
5605 (profile_count::differs_from_p): Check compatiblity.
5606 (profile_count::to_frequency): New function.
5607 (profile_count::to_cgraph_frequency): New function.
5608 * profile-count.h (struct function): Declare.
5609 (enum profile_quality): Add profile_guessed_local and
5610 profile_guessed_global0.
5611 (class profile_proability): Decrease number of bits to 29;
5612 update from_reg_br_prob_note and to_reg_br_prob_note.
5613 (class profile_count: Update comment; decrease number of bits
5614 to 61. Check compatibility.
5615 (profile_count::compatible_p): New private member function.
5616 (profile_count::ipa_p): New member function.
5617 (profile_count::operator<): Handle global zero correctly.
5618 (profile_count::operator>): Handle global zero correctly.
5619 (profile_count::operator<=): Handle global zero correctly.
5620 (profile_count::operator>=): Handle global zero correctly.
5621 (profile_count::nonzero_p): New member function.
5622 (profile_count::force_nonzero): New member function.
5623 (profile_count::max): New member function.
5624 (profile_count::apply_scale): Handle IPA scalling.
5625 (profile_count::guessed_local): New member function.
5626 (profile_count::global0): New member function.
5627 (profile_count::ipa): New member function.
5628 (profile_count::to_frequency): Declare.
5629 (profile_count::to_cgraph_frequency): Declare.
5630 * profile.c (OVERLAP_BASE): Delete.
5631 (compute_frequency_overlap): Delete.
5632 (compute_branch_probabilities): Do not use compute_frequency_overlap.
5633 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
5634 * sched-ebb.c (rank): Use counts only.
5635 * shrink-wrap.c (handle_simple_exit): Use counts only.
5636 (try_shrink_wrapping): Use counts only.
5637 (place_prologue_for_one_component): Use counts only.
5638 * tracer.c (find_best_predecessor): Use to_frequency.
5639 (find_trace): Use to_frequency.
5640 (tail_duplicate): Use to_frequency.
5641 * trans-mem.c (expand_transaction): Do not update frequency.
5642 * tree-call-cdce.c: Do not update frequency.
5643 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
5644 (gimple_merge_blocks): Likewise.
5645 (gimple_split_edge): Likewise.
5646 (gimple_duplicate_sese_region): Likewise.
5647 (gimple_duplicate_sese_tail): Likewise.
5648 (move_sese_region_to_fn): Likewise.
5649 (gimple_account_profile_record): Likewise.
5650 (insert_cond_bb): Likewise.
5651 * tree-complex.c (expand_complex_div_wide): Likewise.
5652 * tree-eh.c (lower_resx): Update profile.
5653 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
5655 (initialize_cfun): Do not initialize frequencies
5656 (freqs_to_counts): Delete.
5657 (copy_cfg_body): Ignore count parameter.
5658 (copy_body): Update.
5659 (expand_call_inline): Update count_max.
5660 (optimize_inline_calls): Update count_max.
5661 (tree_function_versioning): Update count_max.
5662 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
5663 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
5665 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
5666 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
5667 (try_peel_loop): Likewise.
5668 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
5670 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
5671 (tree_transform_and_unroll_loop): Do not use frequencies
5672 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
5673 Use reliable prediction only.
5674 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
5675 * tree-ssa-sink.c (select_best_block): Use to_frequency.
5676 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
5677 probability scaling.
5678 * tree-ssa-threadupdate.c (create_block_for_threading): Do
5679 not update frequency
5680 (any_remaining_duplicated_blocks): Likewise.
5681 (update_profile): Likewise.
5682 (estimated_freqs_path): Delete.
5683 (freqs_to_counts_path): Delete.
5684 (clear_counts_path): Delete.
5685 (ssa_fix_duplicate_block_edges): Likewise.
5686 (duplicate_thread_path): Likewise.
5687 * tree-switch-conversion.c (gen_inbound_check): Use counts.
5688 * tree-tailcall.c (decrease_profile): Do not update frequency.
5689 (eliminate_tail_call): Likewise.
5690 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
5691 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
5692 (optimize_mask_stores): Likewise.
5693 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
5694 * ubsan.c (ubsan_expand_null_ifn): Update profile.
5695 (ubsan_expand_ptr_ifn): Update profile.
5696 * value-prof.c (gimple_ic): Simplify.
5697 * value-prof.h (gimple_ic): Update prototype.
5698 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
5699 * ipa-inline.c (compute_uninlined_call_time): Be sure that
5701 (want_inline_self_recursive_call_p): Likewise.
5702 (resolve_noninline_speculation): Only cummulate defined counts.
5703 (inline_small_functions): Use nonzero_p.
5704 (ipa_inline): Do not access freed node.
5706 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5708 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5709 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
5711 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
5713 * config/riscv/riscv.c (riscv_legitimize_move): Handle
5714 non-legitimate address.
5716 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5718 * config/rs6000/rs6000.md (*lt0_disi): Delete.
5719 (*lt0_<mode>di, *lt0_<mode>si): New.
5721 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5723 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
5724 TARGET_PAIRED_FLOAT.
5726 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
5727 Jim Wilson <jim.wilson@linaro.org>
5729 * config/aarch64/aarch64-cores.def (saphira): New CPU.
5730 * config/aarch64/aarch64-tune.md: Regenerated.
5731 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
5732 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
5734 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
5736 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
5737 (arc_expand_prologue): Restore blink for millicode.
5738 * config/arc/linux.h (LINK_EH_SPEC): Defined.
5740 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
5743 * config/i386/i386.c (ix86_vector_duplicate_value): Use
5744 gen_vec_duplicate after forcing the scalar into a register.
5746 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
5748 * combine (try_combine): Print the insns input to try_combine to the
5751 2017-11-02 Steve Ellcey <sellcey@cavium.com>
5754 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
5755 Remove second argument from aarch64_process_target_attr call.
5756 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
5758 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
5760 (aarch64_handle_attr_arch): Remove second argument.
5761 (aarch64_handle_attr_cpu): Ditto.
5762 (aarch64_handle_attr_tune): Ditto.
5763 (aarch64_handle_attr_isa_flags): Ditto.
5764 (aarch64_process_one_target_attr): Ditto.
5765 (aarch64_process_target_attr): Ditto.
5766 (aarch64_option_valid_attribute_p): Remove second argument.
5767 on aarch64_process_target_attr call.
5769 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5771 * diagnostic.c: Include "selftest-diagnostic.h".
5772 (selftest::assert_location_text): New function.
5773 (selftest::test_diagnostic_get_location_text): New function.
5774 (selftest::diagnostic_c_tests): Call it.
5776 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5778 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
5779 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
5780 (class selftest::test_diagnostic_context): Move to...
5781 * selftest-diagnostic.c: New file.
5782 * selftest-diagnostic.h: New file.
5784 2017-11-02 James Bowman <james.bowman@ftdichip.com>
5786 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
5787 offset range for FT32B.
5788 * config/ft32/ft32.h: option "mcompress" enables relaxation.
5789 * config/ft32/ft32.md: Add TARGET_NOPM.
5790 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
5791 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
5793 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
5795 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
5797 2017-11-02 Jeff Law <law@redhat.com>
5799 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
5800 virtual keyword on FINAL OVERRIDE members.
5802 * tree-ssa-propagate.h (ssa_propagation_engine): Group
5803 virtuals together. Add virtual destructor.
5804 (substitute_and_fold_engine): Similarly.
5806 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
5808 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
5810 2017-11-02 Richard Biener <rguenther@suse.de>
5812 PR tree-optimization/82795
5813 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
5815 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5817 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
5819 (gcc_GAS_CHECK_FEATURE): Remove.
5820 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
5821 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
5822 * configure: Regenerate.
5824 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
5826 * config/arc/arc.c (hwloop_optimize): Account for empty
5829 2017-11-02 Richard Biener <rguenther@suse.de>
5832 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
5833 Truncate ARRAY_REF index and element size.
5835 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
5837 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
5839 2017-11-01 Jeff Law <law@redhat.com>
5841 * tree-ssa-ccp.c (ccp_folder): New class derived from
5842 substitute_and_fold_engine.
5843 (ccp_folder::get_value): New member function.
5844 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
5845 (ccp_fold_stmt): Remove prototype.
5846 (ccp_finalize): Call substitute_and_fold from the ccp_class.
5847 * tree-ssa-copy.c (copy_folder): New class derived from
5848 substitute_and_fold_engine.
5849 (copy_folder::get_value): Renamed from get_value.
5850 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
5851 * tree-vrp.c (vrp_folder): New class derived from
5852 substitute_and_fold_engine.
5853 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
5854 (vrp_folder::get_value): New member function.
5855 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
5856 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
5857 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
5858 provide a class interface to folder/substitute routines.
5859 (ssa_prop_fold_stmt_fn): Remove typedef.
5860 (ssa_prop_get_value_fn): Likewise.
5861 (subsitute_and_fold): Remove prototype.
5862 (replace_uses_in): Likewise.
5863 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
5864 Renamed from replace_uses_in. Call the virtual member function
5865 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
5866 (substitute_and_fold_dom_walker): Remove initialization of
5867 data member entries for calbacks. Add substitute_and_fold_engine
5868 member and initialize it.
5869 (substitute_and_fold_dom_walker::before_dom_children0: Use the
5870 member functions for get_value, replace_phi_args_in c
5871 replace_uses_in, and fold_stmt calls.
5872 (substitute_and_fold_engine::substitute_and_fold): Renamed from
5873 substitute_and_fold. Remove assert. Update ctor call.
5875 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
5876 (ssa_prop_visit_phi_fn): Likewise.
5877 (class ssa_propagation_engine): New class to provide an interface
5879 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
5881 (ssa_prop_visit_phi): Likewise.
5882 (ssa_propagation_engine::simulate_stmt): Moved into class.
5883 Call visit_phi/visit_stmt from the class rather than via
5884 file scoped static variables.
5885 (ssa_propagation_engine::simulate_block): Moved into class.
5886 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
5887 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
5888 set file scoped statics for the visit_stmt/visit_phi callbacks.
5889 * tree-complex.c (complex_propagate): New class derived from
5890 ssa_propagation_engine.
5891 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
5892 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
5893 (tree_lower_complex): Call ssa_propagate via the complex_propagate
5895 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
5896 ssa_propagation_engine.
5897 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
5898 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
5899 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
5900 * tree-ssa-copy.c (copy_prop): New class derived from
5901 ssa_propagation_engine.
5902 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
5903 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
5904 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
5905 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
5906 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
5907 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
5908 (execute_vrp): Call ssa_propagate from the vrp_prop class.
5910 2017-11-01 Jakub Jelinek <jakub@redhat.com>
5912 PR rtl-optimization/82778
5913 PR rtl-optimization/82597
5914 * compare-elim.c (struct comparison): Add in_a_setter field.
5915 (find_comparison_dom_walker::before_dom_children): Remove killed
5916 bitmap and df_simulate_find_defs call, instead walk the defs.
5917 Compute last_setter and initialize in_a_setter. Merge definitions
5918 with first initialization for a few variables.
5919 (try_validate_parallel): Use insn_invalid_p instead of
5920 recog_memoized. Return insn rather than just the pattern.
5921 (try_merge_compare): Fix up comment. Don't uselessly test if
5922 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
5924 (execute_compare_elim_after_reload): Remove df_chain_add_problem
5927 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5928 Alan Hayward <alan.hayward@arm.com>
5929 David Sherwood <david.sherwood@arm.com>
5931 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
5932 aarch64_hard_regno_nregs to get the number of registers
5935 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5936 Alan Hayward <alan.hayward@arm.com>
5937 David Sherwood <david.sherwood@arm.com>
5939 * config/aarch64/constraints.md (Upl): Rename to...
5941 * config/aarch64/aarch64.md
5942 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
5945 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5946 Alan Hayward <alan.hayward@arm.com>
5947 David Sherwood <david.sherwood@arm.com>
5949 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
5950 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
5953 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5954 Alan Hayward <alan.hayward@arm.com>
5955 David Sherwood <david.sherwood@arm.com>
5957 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
5958 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
5959 (aarch64_evpc_dup): Generate rtl direcly, rather than using
5961 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
5962 of a single element.
5963 * config/aarch64/iterators.md: Add a comment above the permute
5964 unspecs to say that they are generated directly by
5965 aarch64_expand_vec_perm_const.
5966 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
5968 2017-11-01 Nathan Sidwell <nathan@acm.org>
5970 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
5972 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
5974 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
5975 and medany code models, and describe what they do.
5977 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5979 Revert accidental duplicate:
5981 * combine.c (can_change_dest_mode): Reject changes in
5982 REGMODE_NATURAL_SIZE.
5984 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
5986 PR rtl-optimization/64682
5987 PR rtl-optimization/69567
5988 PR rtl-optimization/69737
5989 PR rtl-optimization/82683
5990 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
5991 register mentioned in the note, drop the note, unless it came from I3,
5992 in which case it should go to I3 again.
5994 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5996 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
5997 and return false if not.
5998 (clear_bytes_written_by, live_bytes_read): Update accordingly.
6000 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6002 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
6003 range is known to be empty.
6005 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6006 Alan Hayward <alan.hayward@arm.com>
6007 David Sherwood <david.sherwood@arm.com>
6009 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
6010 and CONST_VECTOR_NUNITS instead of computing the number of units from
6011 the byte sizes of the vector and element.
6012 (simplify_binary_operation_1): Likewise.
6013 (simplify_const_binary_operation): Likewise.
6014 (simplify_ternary_operation): Likewise.
6016 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6017 Alan Hayward <alan.hayward@arm.com>
6018 David Sherwood <david.sherwood@arm.com>
6020 * var-tracking.c (INT_MEM_OFFSET): Replace with...
6021 (int_mem_offset): ...this new function.
6022 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
6023 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
6024 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
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 * lower-subreg.c (interesting_mode_p): New function.
6032 (compute_costs, find_decomposable_subregs, decompose_register)
6033 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
6034 (resolve_clobber, dump_choices): Use it.
6036 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6037 Alan Hayward <alan.hayward@arm.com>
6038 David Sherwood <david.sherwood@arm.com>
6040 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
6042 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6043 Alan Hayward <alan.hayward@arm.com>
6044 David Sherwood <david.sherwood@arm.com>
6046 * alias.c (find_base_value, find_base_term): Only process integer
6047 truncations. Check the precision rather than the size.
6049 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6050 Alan Hayward <alan.hayward@arm.com>
6051 David Sherwood <david.sherwood@arm.com>
6053 * machmode.h (is_narrower_int_mode): New function
6054 * optabs.c (expand_float, expand_fix): Use it.
6055 * dwarf2out.c (rotate_loc_descriptor): Likewise.
6057 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6058 Alan Hayward <alan.hayward@arm.com>
6059 David Sherwood <david.sherwood@arm.com>
6061 * rtl.h (narrower_subreg_mode): New function.
6062 * ira-color.c (update_costs_from_allocno): Use it.
6064 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6065 Alan Hayward <alan.hayward@arm.com>
6066 David Sherwood <david.sherwood@arm.com>
6068 * optabs-query.h (convert_optab_p): New function, split out from...
6069 (convert_optab_handler): ...here.
6070 (widening_optab_handler): Delete.
6071 (find_widening_optab_handler): Remove permit_non_widening parameter.
6072 (find_widening_optab_handler_and_mode): Likewise. Provide an
6073 override that operates on mode class wrappers.
6074 * optabs-query.c (widening_optab_handler): Delete.
6075 (find_widening_optab_handler_and_mode): Remove permit_non_widening
6076 parameter. Assert that the two modes are the same class and that
6077 the "from" mode is narrower than the "to" mode. Use
6078 convert_optab_handler instead of widening_optab_handler.
6079 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
6080 instead of widening_optab_handler.
6081 * expr.c (expand_expr_real_2): Update calls to
6082 find_widening_optab_handler.
6083 * optabs.c (expand_widen_pattern_expr): Likewise.
6084 (expand_binop_directly): Take the insn_code as a parameter.
6085 (expand_binop): Only call find_widening_optab_handler for
6086 conversion optabs; use optab_handler otherwise. Update calls
6087 to find_widening_optab_handler and expand_binop_directly.
6088 Use convert_optab_handler instead of widening_optab_handler.
6089 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
6090 find_widening_optab_handler and use scalar_mode rather than
6092 (convert_plusminus_to_widen): Likewise.
6094 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6095 Alan Hayward <alan.hayward@arm.com>
6096 David Sherwood <david.sherwood@arm.com>
6098 * machmode.h (fixed_size_mode): New class.
6099 * rtl.h (get_pool_mode): Return fixed_size_mode.
6100 * gengtype.c (main): Add fixed_size_mode.
6101 * target.def (get_raw_result_mode): Return a fixed_size_mode.
6102 (get_raw_arg_mode): Likewise.
6103 * doc/tm.texi: Regenerate.
6104 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
6105 * targhooks.c (default_get_reg_raw_mode): Likewise.
6106 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
6107 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
6108 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
6109 (msp430_get_raw_result_mode): Likewise.
6110 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
6111 * dbxout.c (dbxout_parms): Require fixed-size modes.
6112 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
6113 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
6114 * omp-low.c (lower_oacc_reductions): Likewise.
6115 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
6116 (simplify_subreg): Update accordingly.
6117 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
6118 (force_const_mem): Update accordingly. Return NULL_RTX for modes
6119 that aren't fixed-size.
6120 (get_pool_mode): Return a fixed_size_mode.
6121 (output_constant_pool_2): Take a fixed_size_mode.
6123 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6124 Alan Hayward <alan.hayward@arm.com>
6125 David Sherwood <david.sherwood@arm.com>
6127 * doc/rtl.texi (vec_series): Document.
6128 (const): Say that the operand can be a vec_series.
6129 * rtl.def (VEC_SERIES): New rtx code.
6130 * rtl.h (const_vec_series_p_1): Declare.
6131 (const_vec_series_p): New function.
6132 * emit-rtl.h (gen_const_vec_series): Declare.
6133 (gen_vec_series): Likewise.
6134 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
6135 (gen_vec_series): Likewise.
6136 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
6137 * simplify-rtx.c (simplify_unary_operation): Handle negations
6139 (simplify_binary_operation_series): New function.
6140 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
6141 (test_vector_ops_series): New function.
6142 (test_vector_ops): Call it.
6143 * config/powerpcspe/altivec.md (altivec_lvsl): Use
6144 gen_const_vec_series.
6145 (altivec_lvsr): Likewise.
6146 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
6148 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6149 Alan Hayward <alan.hayward@arm.com>
6150 David Sherwood <david.sherwood@arm.com>
6152 * doc/rtl.texi (const): Update description of address constants.
6153 Say that vector constants are allowed too.
6154 * common.md (E, F): Use CONSTANT_P instead of checking for
6156 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
6157 checking for CONST_VECTOR.
6158 * expmed.c (make_tree): Use build_vector_from_val for a CONST
6160 * expr.c (expand_expr_real_2): Check for vector modes instead
6161 of checking for CONST_VECTOR.
6162 * rtl.h (const_vec_p): New function.
6163 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
6164 (unwrap_const_vec_duplicate): Handle them here too.
6166 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6167 David Malcolm <dmalcolm@redhat.com>
6168 Alan Hayward <alan.hayward@arm.com>
6169 David Sherwood <david.sherwood@arm.com>
6171 * rtl.h (vec_duplicate_p): New function.
6172 * selftest-rtl.c (assert_rtx_eq_at): New function.
6173 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
6174 (assert_rtx_eq_at): Declare.
6175 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
6176 * selftest-run-tests.c (selftest::run_tests): Call it.
6177 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
6178 (simplify_unary_operation_1): Recursively handle vector duplicates.
6179 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
6181 (simplify_subreg): Handle subregs of vector duplicates.
6182 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
6183 (selftest::simplify_rtx_c_tests): New functions.
6185 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6186 Alan Hayward <alan.hayward@arm.com>
6187 David Sherwood <david.sherwood@arm.com>
6189 * emit-rtl.h (gen_const_vec_duplicate): Declare.
6190 (gen_vec_duplicate): Likewise.
6191 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
6193 (gen_const_vector): ...here.
6194 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
6195 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
6196 whose elements are all equal.
6197 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
6198 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
6199 (simplify_relational_operation): Likewise.
6200 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
6202 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
6203 (aarch64_expand_vector_init): Likewise.
6204 * config/arm/arm.c (neon_vdup_constant): Likewise.
6205 (neon_expand_vector_init): Likewise.
6206 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
6207 (arm_block_set_unaligned_vect): Likewise.
6208 (arm_block_set_aligned_vect): Likewise.
6209 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
6210 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
6211 (expand_vec_perm_even_odd_pack): Likewise.
6212 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
6213 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
6214 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
6215 gen_const_vec_duplicate.
6216 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
6217 * config/mips/mips.c (mips_gen_const_int_vector): Use
6218 gen_const_vec_duplicate.
6219 (mips_expand_vector_init): Use CONST0_RTX.
6220 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
6221 (define_split): Use gen_const_vec_duplicate.
6222 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
6223 (define_split): Use gen_const_vec_duplicate.
6224 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
6225 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
6226 * config/spu/spu.c (spu_const): Likewise.
6228 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6229 Alan Hayward <alan.hayward@arm.com>
6230 David Sherwood <david.sherwood@arm.com>
6232 * combine.c (can_change_dest_mode): Reject changes in
6233 REGMODE_NATURAL_SIZE.
6235 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
6237 * configure.ac (--enable-libssp): New.
6238 (gcc_cv_libc_provides_ssp): Check for explicit setting before
6239 trying to determine target-specific default. Adjust indentation.
6240 * configure: Regenerated.
6241 * doc/install.texi (Configuration): Expand --disable-libssp
6244 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
6246 config/i386/i386.c (ix86_expand_epilogue): Correct stack
6249 2017-10-31 Martin Jambor <mjambor@suse.cz>
6252 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
6254 2017-10-31 David Malcolm <dmalcolm@redhat.com>
6256 * auto-profile.c (autofdo_source_profile::read): Use
6257 UNKNOWN_LOCATION rather than 0.
6258 * diagnostic-core.h (warning_at_rich_loc): Rename to...
6259 (warning_at): ...this overload.
6260 (warning_at_rich_loc_n): Rename to...
6261 (warning_n): ...this overload.
6262 (error_at_rich_loc): Rename to...
6263 (error_at): ...this overload.
6264 (pedwarn_at_rich_loc): Rename to...
6265 (pedwarn): ...this overload.
6266 (permerror_at_rich_loc): Rename to...
6267 (permerror): ...this overload.
6268 (inform_at_rich_loc): Rename to...
6269 (inform): ...this overload.
6270 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
6271 (diagnostic_n_impl_richloc): Rename to...
6272 (diagnostic_n_impl): ...this rich_location *-based decl.
6273 (inform_at_rich_loc): Rename to...
6274 (inform): ...this, and add an assertion.
6275 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
6276 (warning_at_rich_loc): Rename to...
6277 (warning_at): ...this, and add an assertion.
6278 (warning_at_rich_loc_n): Rename to...
6279 (warning_n): ...this, and add an assertion.
6280 (warning_n): Update location_t-based implementation for removal of
6281 location_t-based diagnostic_n_impl.
6282 (pedwarn_at_rich_loc): Rename to...
6283 (pedwarn): ...this, and add an assertion.
6284 (permerror_at_rich_loc): Rename to...
6285 (permerror): ...this, and add an assertion.
6286 (error_n): Update for removal of location_t-based diagnostic_n_impl.
6287 (error_at_rich_loc): Rename to...
6288 (error_at): ...this, and add an assertion.
6289 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
6290 (driver::do_spec_on_infiles): Likewise.
6291 * substring-locations.c (format_warning_va): Update for renaming
6292 of inform_at_rich_loc.
6294 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
6296 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
6297 _Float<N>X built-in functions so that the variant without the
6298 "__builtin_" prefix is only enabled for the GNU C and Objective C
6299 languages when they are in non-strict ANSI/ISO mode.
6300 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
6301 * target.def (floatn_builtin_p): Add a target hook to control
6302 whether _Float<N> and _Float<N>X built-in functions without the
6303 "__builtin_" prefix are enabled, and return true for C and
6304 Objective C in the default hook. Include langhooks.h in
6306 * targhooks.h (default_floatn_builtin_p): Likewise.
6307 * targhooks.c (default_floatn_builtin_p): Likewise.
6308 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
6309 floatn_builtin_p target hook.
6310 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
6312 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
6313 Eric Botcazou <ebotcazou@adacore.com>
6315 PR rtl-optimization/81803
6316 * lra-constraints.c (curr_insn_transform): Also reload the whole
6317 register for a strict subreg no wider than a word if this is for
6318 a WORD_REGISTER_OPERATIONS target.
6320 2017-10-31 Jason Merrill <jason@redhat.com>
6322 * gdbinit.in: Skip over inlines from timevar.h.
6324 2017-10-31 Martin Liska <mliska@suse.cz>
6326 * doc/gcov.texi: Document new option.
6327 * gcov.c (print_usage): Likewise print it.
6328 (process_args): Support the argument.
6329 (format_count): New function.
6330 (format_gcov): Use the function.
6332 2017-10-31 Martin Liska <mliska@suse.cz>
6334 * gcov.c (struct name_map): do not use typedef.
6335 Define operator== and operator<.
6336 (name_search): Remove.
6337 (name_sort): Remove.
6338 (main): Do not allocate names.
6339 (process_file): Add vertical space.
6340 (generate_results): Use std::find.
6341 (release_structures): Do not release memory.
6342 (find_source): Use std::find.
6344 2017-10-31 Martin Liska <mliska@suse.cz>
6346 * gcov.c (struct line_info): Remove it's typedef.
6347 (line_info::line_info): Add proper ctor.
6348 (line_info::has_block): Do not use a typedef.
6349 (struct source_info): Do not use typedef.
6350 (circuit): Likewise.
6351 (get_cycles_count): Likewise.
6352 (output_intermediate_file): Iterate via vector iterator.
6353 (add_line_counts): Use std::vector methods.
6354 (accumulate_line_counts): Likewise.
6355 (output_lines): Likewise.
6357 2017-10-31 Martin Liska <mliska@suse.cz>
6359 * gcov.c (struct source_info): Remove typedef.
6360 (source_info::source_info): Add proper ctor.
6361 (accumulate_line_counts): Use struct, not it's typedef.
6362 (output_gcov_file): Likewise.
6363 (output_lines): Likewise.
6364 (main): Do not allocate an array.
6365 (output_intermediate_file): Use size of vector container.
6366 (process_file): Resize the vector.
6367 (generate_results): Do not preallocate, use newly added vector
6369 (release_structures): Do not release sources.
6370 (find_source): Use vector methods.
6371 (add_line_counts): Do not use typedef.
6373 2017-10-31 Martin Liska <mliska@suse.cz>
6375 * doc/gcov.texi: Document that.
6376 * gcov.c (add_line_counts): Mark lines with a non-executed
6378 (output_line_beginning): Handle such lines.
6379 (output_lines): Pass new argument.
6380 (output_intermediate_file): Print it in intermediate format.
6382 2017-10-31 Martin Liska <mliska@suse.cz>
6384 * color-macros.h: New file.
6385 * diagnostic-color.c: Factor out color related to macros to
6387 * doc/gcov.texi: Document -k option.
6388 * gcov.c (INCLUDE_STRING): Include string.h.
6389 (print_usage): Add -k option.
6390 (process_args): Parse it.
6391 (pad_count_string): New function.
6392 (output_line_beginning): Likewise.
6393 (DEFAULT_LINE_START): New macro.
6394 (output_lines): Support color output.
6396 2017-10-31 Martin Liska <mliska@suse.cz>
6398 PR gcov-profile/82633
6399 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
6400 their interaction with GCOV infrastructure.
6401 * configure.ac: Add -fkeep-{inline,static}-functions to
6403 * configure: Regenerate.
6405 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
6408 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
6410 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
6413 * config/rs6000/rs6000.md (allocate_stack): Force update interval
6414 into a register if it does not fit into an immediate offset field.
6416 2017-10-31 Olivier Hainque <hainque@adacore.com>
6418 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
6420 2017-10-31 Julia Koval <julia.koval@intel.com>
6422 * config.gcc: Add gfniintrin.h.
6423 * config/i386/gfniintrin.h: New.
6424 * config/i386/i386-builtin-types.def
6425 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
6426 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
6427 __builtin_ia32_vgf2p8affineinvqb_v32qi,
6428 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
6429 __builtin_ia32_vgf2p8affineinvqb_v16qi,
6430 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
6431 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
6432 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
6433 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
6434 V64QI_FTYPE_V64QI_V64QI_INT): New types.
6435 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
6436 * config/i386/immintrin.h: Include gfniintrin.h.
6437 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
6439 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
6441 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
6443 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
6445 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
6446 (arm_ashldi3_1bit): Remove pattern.
6447 (ashrdi3): Remove shift by 1 expansion.
6448 (arm_ashrdi3_1bit): Remove pattern.
6449 (lshrdi3): Remove shift by 1 expansion.
6450 (arm_lshrdi3_1bit): Remove pattern.
6451 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
6452 cost of ashldi3 by 1.
6453 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
6454 (<shift>di3_neon): Likewise.
6456 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6458 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
6459 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
6460 and (*aarch64_simd_mov<VQ:mode>).
6461 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
6462 pattern alternative.
6463 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
6464 attributes to match pattern alternative.
6466 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
6468 * config.gcc (powerpc*-*-*): Add emmintrin.h.
6469 * config/rs6000/emmintrin.h: New file.
6470 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
6472 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
6474 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
6475 * (movdi_vfp_cortexa8): Remove pattern.
6477 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6479 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
6481 (Specific, alpha*-dec-osf5.1): Remove.
6482 (Specific, mips-sgi-irix5): Remove.
6483 (Specific, mips-sgi-irix6): Remove.
6485 2017-10-30 Jakub Jelinek <jakub@redhat.com>
6488 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
6489 arguments to clear_bit_region_be.
6491 2017-10-30 Jim Wilson <wilson@tuliptree.org>
6493 * gimplify.c: Include memmodel.h.
6495 2017-10-30 Martin Jambor <mjambor@suse.cz>
6497 * omp-grid.c (grid_attempt_target_gridification): Also insert a
6498 condition whether loop should be executed at all.
6500 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
6502 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6503 gimple folding of vec_madd() intrinsics.
6504 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
6505 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
6506 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
6508 2017-10-30 Richard Biener <rguenther@suse.de>
6510 PR tree-optimization/82762
6512 2017-10-23 Richard Biener <rguenther@suse.de>
6514 PR tree-optimization/82129
6516 2017-08-01 Richard Biener <rguenther@suse.de>
6518 PR tree-optimization/81181
6519 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6520 (compute_antic): ... end of iteration here.
6522 2017-10-30 Joseph Myers <joseph@codesourcery.com>
6524 * doc/invoke.texi (C Dialect Options): Document -std=c17,
6525 -std=iso9899:2017 and -std=gnu17.
6526 * doc/standards.texi (C Language): Document C17 support.
6527 * doc/cpp.texi (Overview): Mention -std=c17.
6528 (Standard Predefined Macros): Document C11 and C17 values of
6529 __STDC_VERSION__. Do not refer to C99 support as incomplete.
6530 * doc/extend.texi (Inline): Do not list individual options for
6531 standards newer than C99.
6532 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
6534 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
6537 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
6539 * asan.c (asan_finish_file): Align asan globals array by shadow
6542 2017-10-30 Jakub Jelinek <jakub@redhat.com>
6545 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
6546 (struct store_immediate_info): Add bitregion_start and bitregion_end
6548 (store_immediate_info::store_immediate_info): Add brs and bre
6549 arguments and initialize bitregion_{start,end} from those.
6550 (struct merged_store_group): Add bitregion_start, bitregion_end,
6551 align_base and mask fields. Drop unnecessary struct keyword from
6552 struct store_immediate_info. Add do_merge method.
6553 (clear_bit_region_be): Use memset instead of loop storing zeros.
6554 (merged_store_group::do_merge): New method.
6555 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
6556 stores as long as the surrounding bitregions have no gaps.
6557 (merged_store_group::merge_overlapping): Use do_merge.
6558 (merged_store_group::apply_stores): Test that bitregion_{start,end}
6559 is byte aligned, rather than requiring that start and width are
6560 byte aligned. Drop unnecessary struct keyword from
6561 struct store_immediate_info. Allocate and populate also mask array.
6562 Make start of the arrays relative to bitregion_start rather than
6563 start and size them according to bitregion_{end,start} difference.
6564 (struct imm_store_chain_info): Drop unnecessary struct keyword from
6565 struct store_immediate_info.
6566 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
6567 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
6568 struct keyword from struct store_immediate_info.
6569 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
6570 between stores as long as the surrounding bitregions have no gaps.
6572 (struct split_store): Add orig non-static data member.
6573 (split_store::split_store): Initialize orig to false.
6574 (find_constituent_stmts): Return store_immediate_info *, non-NULL
6575 if there is exactly a single original stmt. Change stmts argument
6576 to pointer from reference, if NULL, don't push anything to it. Add
6577 first argument, use it to optimize skipping over orig stmts that
6578 are known to be before bitpos already. Simplify.
6579 (split_group): Return unsigned int count how many stores are or
6580 would be needed rather than a bool. Add allow_unaligned argument.
6581 Change split_stores argument from reference to pointer, if NULL,
6582 only do a dry run computing how many stores would be produced.
6583 Rewritten algorithm to use both alignment and misalign if
6584 !allow_unaligned and handle bitfield stores with gaps.
6585 (imm_store_chain_info::output_merged_store): Set start_byte_pos
6586 from bitregion_start instead of start. Compute allow_unaligned
6587 here, if true, do 2 split_group dry runs to compute which one
6588 produces fewer stores and prefer aligned if equal. Punt if
6589 new count is bigger or equal than original before emitting any
6590 statements, rather than during that. Remove no longer needed
6591 new_ssa_names tracking. Replace num_stmts with
6592 split_stores.length (). Use 32-bit stack allocated entries
6593 in split_stores auto_vec. Try to reuse original store lhs/rhs1
6594 if possible. Handle bitfields with gaps.
6595 (pass_store_merging::execute): Ignore bitsize == 0 stores.
6596 Compute bitregion_{start,end} for the stores and construct
6597 store_immediate_info with that. Formatting fixes.
6599 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
6602 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
6603 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
6605 2017-10-29 Jim Wilson <wilson@tuliptree.org>
6607 * gimplify.c: Include tm_p.h.
6609 * common.opt (gcoff): Re-add as ignored option.
6610 (gcoff1, gcoff2, gcoff3): Likewise.
6612 * Makefile.in (OBJS): Delete sdbout.o.
6613 (GTFILES): Delete $(srcdir)/sdbout.c.
6614 * debug.h: Delete sdb_debug_hooks.
6615 * final.c: Delete sdbout.h include.
6616 (final_scan_insn): Delete SDB_DEBUG check.
6617 (rest_of_clean_state): Likewise.
6618 * output.h: Delete sdb_begin_function_line.
6621 * toplev.c: Delete sdbout.h include.
6622 (process_options): Delete SDB_DEBUG check.
6623 * tree-core.h (tree_type_common): Delete pointer field of
6625 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
6626 TYPE_SYMTAB_POINTER.
6627 * tree.h (TYPE_SYMTAB_POINTER): Delete.
6628 (TYPE_SYMTAB_IS_POINTER): Delete.
6629 (TYPE_SYMTAB_IS_DIE): Renumber.
6630 * xcoffout.c: Refer to former sdbout.c file.
6631 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
6633 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
6634 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
6635 (Debugging Options): Delete -gcoff.
6636 (-gstabs): Delete SDB reference.
6638 (-gcoff@var{level}): Delete.
6639 * doc/passes.texi (Debugging information output): Delete SDB and
6640 sdbout.c references.
6641 * doc/tm.texi: Regenerate.
6642 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
6643 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
6645 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
6646 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
6647 (SDB_DEBUGGING_INFO): Delete.
6648 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
6649 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
6650 * target.def (output_source_filename): Delete COFF reference.
6652 * common.opt (gcoff): Delete.
6653 (gxcoff+): Update Negative chain.
6654 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
6656 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
6657 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
6658 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
6660 (expand_function_start): Change sdb reference to past tense.
6661 (expand_function_end): Change sdb reference to past tense.
6662 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
6663 * opts.c (debug_type_names): Delete coff entry.
6664 (common_handle_option): Delete OPT_gcoff case.
6665 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
6667 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
6668 * config/cris/cris.h: Delete SDB reference in comment.
6669 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
6670 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
6671 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
6672 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
6674 (ix86_expand_prologue): Likewise.
6675 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
6676 * config/ia64/ia64.h: Likewise.
6677 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
6678 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
6680 * config/mmix/mmix.h: Likewise.
6681 * config/nds32/nds32.c: Likewise.
6682 * config/stormy/storym16.h: Likewise.
6683 * config/visium/visium.h: Likewise.
6684 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
6686 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
6688 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
6689 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
6690 FRAME_POINTER_REGNUM point at high end of local var area.
6692 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
6694 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
6695 Move comment around. Do not reset best_edge for a copiable
6696 destination if the copy would cause a partition change.
6697 (better_edge_p): Remove redundant check.
6699 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6701 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
6703 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
6705 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
6706 for math functions that have _Float<N> and _Float<N>X variants.
6707 (mathfn_built_in_2): Add support for math functions that have
6708 _Float<N> and _Float<N>X variants.
6709 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
6710 (expand_builtin_mathfn_ternary): Add support for fma with
6711 _Float<N> and _Float<N>X variants.
6712 (expand_builtin): Likewise.
6713 (fold_builtin_3): Likewise.
6714 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
6715 create math function _Float<N> and _Float<N>X variants as external
6717 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
6718 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
6719 the __builtin_ prefix and if not strict ansi, without the prefix.
6720 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
6721 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
6722 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
6723 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
6724 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
6725 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
6726 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
6727 function signatures for fma _Float<N> and _Float<N>X variants.
6728 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
6729 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
6730 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
6731 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
6732 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
6733 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
6734 * gencfn-macros.c (print_case_cfn): Add support for math functions
6735 that have _Float<N> and _Float<N>X variants.
6736 (print_define_operator_list): Likewise.
6737 (fltfn_suffixes): Likewise.
6739 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
6740 for math functions that have _Float<N> and _Float<N>X variants.
6741 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
6742 and _Float<N>X variants.
6743 (COPYSIGN): Likewise.
6746 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
6747 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
6749 (integer_valued_read_call_p): Likewise.
6750 * fold-const-call.c (fold_const_call_ss): Likewise.
6751 (fold_const_call_sss): Add support for copysign, fmin, and fmax
6752 _Float<N> and _Float<N>X variants.
6753 (fold_const_call_ssss): Add support for fma _Float<N> and
6754 _Float<N>X variants.
6755 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
6756 support for copysign and fma _Float<N> and _Float<N>X variants.
6757 (backprop::process_builtin_call_use): Likewise.
6758 * tree-call-cdce.c (can_test_argument_range); Add support for
6759 sqrt _Float<N> and _Float<N>X variants.
6760 (edom_only_function): Likewise.
6761 (get_no_error_domain): Likewise.
6762 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
6763 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
6764 copysign _Float<N> and _Float<N>X variants.
6765 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
6766 handled by machine independent code.
6767 (FMAF128): Likewise.
6768 * doc/cpp.texi (Common Predefined Macros): Document defining
6769 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
6770 fma _Float<N> and _Float<N>X variants.
6772 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6775 * config/i386/i386-modes.def (CCFPU): Remove definition.
6776 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
6777 (ix86_cc_modes_compatible): Ditto.
6778 (ix86_expand_carry_flag_compare): Ditto.
6779 (ix86_expand_int_movcc): Ditto.
6780 (ix86_expand_int_addcc): Ditto.
6781 (ix86_reverse_condition): Ditto.
6782 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
6783 Return true/false for unordered/ordered fp comparisons.
6784 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
6785 (ix86_prepare_fp_compare_args): Update for rename.
6786 (ix86_expand_fp_compare): Update for rename. Generate unordered
6787 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
6788 (ix86_expand_sse_compare_and_jump): Ditto.
6789 * config/i386/predicates.md (fcmov_comparison_operator):
6790 Remove CCFPU mode handling.
6791 (ix86_comparison_operator): Ditto.
6792 (ix86_carry_flag_operator): Ditto.
6793 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
6794 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
6795 (*cmpu<mode>_cc_i387): Ditto.
6796 (FPCMP): Remove mode iterator.
6797 (unord): Remove mode attribute.
6798 (unord_subst): New define_subst transformation
6799 (unord): New define_subst attribute.
6801 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
6802 (*cmpi<unord>xf_i387): Ditto.
6803 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
6804 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
6805 using unord_subst transformation.
6806 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
6807 (round_saeonly): Also handle CCFP mode.
6808 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
6809 Remove UNSPEC_SAHF unspec handling.
6811 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6813 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
6815 2017-10-27 Jeff Law <law@redhat.com>
6817 * gimple-ssa-sprintf.c: Include domwalk.h.
6818 (class sprintf_dom_walker): New class, derived from dom_walker.
6819 (sprintf_dom_walker::before_dom_children): New function.
6820 (struct call_info): Moved into sprintf_dom_walker class
6821 (compute_formath_length, handle_gimple_call): Likewise.
6822 (sprintf_length::execute): Call the dominator walker rather
6823 than walking the statements.
6825 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
6826 gimple statement locations.
6827 (check_array_bounds): Corresponding changes. Get the statement's
6828 location directly from wi->stmt.
6830 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
6833 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
6835 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6837 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
6838 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
6840 2017-10-27 Jakub Jelinek <jakub@redhat.com>
6843 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
6844 * config/i386/i386.c (maybe_get_pool_constant): Removed.
6845 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
6846 maybe_get_pool_constant.
6847 * config/i386/predicates.md (zero_extended_scalar_load_operand):
6850 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6852 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
6853 2.26 caveat. Update gas and gld versions.
6854 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
6857 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6859 * cgraph.h (set_malloc_flag): Declare.
6860 * cgraph.c (set_malloc_flag_1): New function.
6861 (set_malloc_flag): Likewise.
6862 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
6863 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
6865 (read_ipa_call_summary): Add support for reading is_return_callee.
6866 (write_ipa_call_summary): Stream is_return_callee.
6867 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
6868 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
6869 ipa-prop.h, ipa-fnsummary.h.
6870 (pure_const_names): Change to static.
6871 (malloc_state_e): Define.
6872 (malloc_state_names): Define.
6873 (funct_state_d): Add field malloc_state.
6874 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
6875 (check_retval_uses): New function.
6876 (malloc_candidate_p): Likewise.
6877 (analyze_function): Add support for malloc attribute.
6878 (pure_const_write_summary): Stream malloc_state.
6879 (pure_const_read_summary): Add support for reading malloc_state.
6880 (dump_malloc_lattice): New function.
6881 (propagate_malloc): New function.
6882 (warn_function_malloc): New function.
6883 (ipa_pure_const::execute): Call propagate_malloc and
6884 ipa_free_fn_summary.
6885 (pass_local_pure_const::execute): Add support for malloc attribute.
6886 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
6887 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
6889 2017-10-27 Martin Liska <mliska@suse.cz>
6891 PR gcov-profile/82457
6892 * doc/invoke.texi: Document that one needs a non-strict ISO mode
6893 for fork-like functions to be properly instrumented.
6895 2017-10-27 Richard Biener <rguenther@suse.de>
6898 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
6899 info when we redirected EH.
6901 2017-10-26 Michael Collison <michael.collison@arm.com>
6903 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
6905 (<optab>_trunchf<GPI:mode>2: New pattern.
6906 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
6907 * config/aarch64/iterators.md (wv): New mode attribute.
6908 (vf, VF): New mode attributes.
6909 (vgp, VGP): New mode attributes.
6910 (s): Update attribute with SImode and DImode prefixes.
6912 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
6914 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
6915 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
6916 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
6917 (nios2_option_overide): Initialize it. Don't allow R0-relative
6918 addressing with PIC.
6919 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
6920 (nios2_symbolic_constant_p): Likewise.
6921 (nios2_legitimate_address_p): Likewise.
6922 (nios2_r0rel_section_name_p): New.
6923 (nios2_symbol_ref_in_r0rel_data_p): New.
6924 (nios2_emit_move_sequence): Handle r0rel_constant_p.
6925 (r0rel_constant_p): New.
6926 (nios2_print_operand_address): Handle r0rel_constant_p.
6927 (nios2_cdx_narrow_form_p): Likewise.
6928 * config/nios2/nios2.opt (mr0rel-sec=): New option.
6929 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
6930 (Nios II Options): Document -mr0rel-sec.
6932 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
6934 * config/nios2/nios2.c: Include xregex.h.
6935 (nios2_gprel_sec_regex): New.
6936 (nios2_option_overide): Initialize it. Don't allow GP-relative
6937 addressing with PIC.
6938 (nios2_small_section_name_p): Check for regex match.
6939 * config/nios2/nios2.opt (mgprel-sec=): New option.
6940 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
6941 (Nios II Options): Document -mgprel-sec.
6943 2017-10-26 Jim Wilson <wilson@tuliptree.org>
6945 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
6947 2017-10-26 Tom de Vries <tom@codesourcery.com>
6949 PR tree-optimization/82707
6950 * gimple.c (gimple_copy): Fix unsharing of
6951 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
6953 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
6955 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
6956 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
6957 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
6958 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
6959 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
6960 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
6961 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
6962 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
6963 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
6964 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
6965 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
6966 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
6967 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
6968 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
6969 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
6970 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
6971 _mm512_mask_cmpunord_ps_mask): New intrinsics.
6973 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6975 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
6977 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
6978 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
6979 warning to rs6000.c. Remove the Undocumented flag, since it has
6981 (-mabi=ibmlongdouble): Likewise.
6982 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
6983 already set, set the default format for long double.
6984 (rs6000_debug_reg_global): Print whether long double is IBM or
6986 (rs6000_option_override_internal): Rework setting long double
6987 format. Only warn if the user is changing the long double default
6988 and they did not use -Wno-psabi.
6989 * doc/invoke.texi (PowerPC options): Update the documentation for
6990 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
6992 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6993 Alan Hayward <alan.hayward@arm.com>
6994 David Sherwood <david.sherwood@arm.com>
6996 * rtl.h (wider_subreg_mode): New function.
6997 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
6998 rather than an unsigned int *.
6999 * ira-color.c (regno_max_ref_width): Replace with...
7000 (regno_max_ref_mode): ...this new variable.
7001 (coalesced_pseudo_reg_slot_compare): Update accordingly.
7002 Use wider_subreg_mode.
7003 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
7004 rather than an unsigned int *.
7005 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
7006 (process_alt_operands): Likewise.
7007 (invariant_p): Likewise.
7008 * lra-spills.c (assign_mem_slot): Likewise.
7009 (add_pseudo_to_slot): Likewise.
7010 * lra.c (collect_non_operand_hard_regs): Likewise.
7011 (add_regs_to_insn_regno_info): Likewise.
7012 * reload1.c (regno_max_ref_width): Replace with...
7013 (regno_max_ref_mode): ...this new variable.
7014 (reload): Update accordingly. Update call to
7015 ira_sort_regnos_for_alter_reg.
7016 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
7017 (init_eliminable_invariants): Update to use regno_max_ref_mode.
7018 (scan_paradoxical_subregs): Likewise.
7020 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7022 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
7023 (aarch64_frame): Add emit_frame_chain boolean.
7024 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7025 Move eh_return case to aarch64_layout_frame.
7026 (aarch64_layout_frame): Initialize emit_frame_chain.
7027 (aarch64_expand_prologue): Use emit_frame_chain.
7029 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7031 * config/aarch64/aarch64.c (aarch64_layout_frame):
7032 Ensure LR is always stored at the bottom of the callee-saves.
7033 Remove rarely used frame layout which saves callee-saves at top of
7034 frame, so the store of LR can be used as a valid probe in all cases.
7036 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7038 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
7039 Improve unaligned TImode/TFmode base/offset split.
7041 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7042 Alan Hayward <alan.hayward@arm.com>
7043 David Sherwood <david.sherwood@arm.com>
7045 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
7046 * combine.c (find_single_use_1): Likewise.
7047 (expand_field_assignment): Likewise.
7048 (move_deaths): Likewise.
7049 * lra-constraints.c (simplify_operand_subreg): Likewise.
7050 (curr_insn_transform): Likewise.
7051 * lra.c (collect_non_operand_hard_regs): Likewise.
7052 (add_regs_to_insn_regno_info): Likewise.
7053 * rtlanal.c (reg_referenced_p): Likewise.
7054 (covers_regno_no_parallel_p): Likewise.
7056 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7058 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
7059 Don't print any bits outside the precision of the value.
7060 * wide-int.cc (test_printing): Add some new tests.
7062 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7064 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
7065 supports -xbrace_comment option.
7066 * configure: Regenerate.
7067 * config.in: Regenerate.
7068 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
7069 (ASM_CPU_SPEC): Use it.
7071 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7073 * target.def (static_rtx_alignment): New hook.
7074 * targhooks.h (default_static_rtx_alignment): Declare.
7075 * targhooks.c (default_static_rtx_alignment): New function.
7076 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
7077 * doc/tm.texi: Regenerate.
7078 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
7079 instead of targetm.constant_alignment. Remove call to
7081 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7082 (cris_preferred_mininum_alignment): New function, split out from...
7083 (cris_constant_alignment): ...here.
7084 (cris_static_rtx_alignment): New function.
7085 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
7087 (ix86_constant_alignment): ...here.
7088 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7089 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7090 (mmix_static_rtx_alignment): New function.
7091 * config/spu/spu.c (spu_static_rtx_alignment): New function.
7092 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7094 2017-10-26 Tamar Christina <tamar.christina@arm.com>
7097 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
7098 Add flag_trapping_math and flag_fp_int_builtin_inexact.
7100 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
7102 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
7103 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
7104 mark as a sign-extending load.
7105 (local_pic_load_u): Define.
7107 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
7110 * fold-const.c (operand_equal_for_comparison_p): Also return true
7111 if ARG0 is a simple variant of ARG1 with narrower precision.
7112 (fold_ternary_loc): Always pass unstripped operands to the predicate.
7114 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
7116 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
7118 * i386.h (processor_costs): Add gather_static, gather_per_elt,
7119 scatter_static, scatter_per_elt.
7120 * x86-tune-costs.h: Add new cost entries.
7122 2017-10-25 Richard Biener <rguenther@suse.de>
7124 * tree-ssa-sccvn.h (vn_eliminate): Declare.
7125 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
7126 class pass_fre): Move to ...
7127 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
7128 class pass_fre): ... here and adjust for statistics.
7130 2017-10-25 Jakub Jelinek <jakub@redhat.com>
7133 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
7134 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
7135 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
7136 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
7139 2017-10-25 Richard Biener <rguenther@suse.de>
7141 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
7142 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
7143 eliminate_push_avail, eliminate_insert): Move inside...
7144 (class eliminate_dom_walker): ... this class in preparation
7146 (fini_eliminate): Remove by merging with ...
7147 (eliminate): ... this function. Adjust for class changes.
7148 (pass_pre::execute): Remove fini_eliminate call.
7149 (pass_fre::execute): Likewise.
7151 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7154 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
7155 (VPERMI2, VPERMI2I): New mode iterators.
7156 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
7157 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
7159 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
7160 mode iterator. Remove 3 old define_insn patterns.
7161 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
7162 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
7163 VPERMI2 mode iterator, remove the other two expanders.
7164 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
7165 to use VPERMI2 mode iterator, add another alternative for vpermi2*
7166 instructions, remove the other two patterns.
7167 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
7168 mode iterator, remove the other two patterns.
7169 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
7170 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
7171 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
7172 and adjust argument order accordingly.
7173 (ix86_expand_vec_perm): Adjust caller.
7174 (expand_vec_perm_1): Likewise.
7175 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
7176 (expand_vec_perm_vpermt2_vpshub2): ... this.
7177 (ix86_expand_vec_perm_const_1): Adjust caller.
7178 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
7181 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
7182 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
7183 (vec_shl_<mode>): Remove unused expander.
7184 (avx512bw_<shift_insn><mode>3): New define_insn.
7185 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
7186 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
7188 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
7191 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
7194 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7196 PR rtl-optimization/82396
7197 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
7198 (autopref_multipass_init): Simplify initialization.
7199 (autopref_rank_data): Simplify sort order.
7200 * gcc/sched-int.h (autopref_multipass_data_): Remove
7201 multi_mem_insn_p, min_offset and max_offset.
7203 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7206 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7207 Check special value of flag_omit_frame_pointer.
7208 (aarch64_can_eliminate): Likewise.
7209 (aarch64_override_options_after_change_1): Simplify handling of
7210 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
7212 2017-10-24 Richard Biener <rguenther@suse.de>
7214 PR tree-optimization/82697
7215 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
7216 zero for conditional load and unconditional store.
7218 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7220 * doc/install.texi: Document bootstrap-cet.
7222 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7225 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
7226 ENDBR instruction at function entrance if function is only
7229 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7232 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
7233 patterns to better describe from which operation the CF is computed.
7234 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
7235 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
7236 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
7237 is 0, use _0 suffixed expanders instead of emitting a comparison
7240 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
7242 * config/i386/i386.md(*movsf_internal, *movdf_internal):
7243 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
7245 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
7248 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
7249 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
7250 * loop-iv.c (iv_get_reaching_def): Likewise.
7251 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
7252 variable is promoted and the partition contains undefined values.
7254 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7256 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
7258 (nios2_address_cost): Define.
7259 (nios2_legitimize_address): Recognize (exp + constant) directly.
7260 (TARGET_ADDRESS_COST): Define.
7262 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7264 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
7265 (nios2_symbolic_memory_operand_p): Declare.
7266 (nios2_split_large_constant): Declare.
7267 (nios2_split_symbolic_memory_operand): Declare.
7268 * config/nios2/nios2.c: Adjust includes.
7269 (nios2_symbolic_constant_allowed): New.
7270 (nios2_symbolic_constant_p): New.
7271 (nios2_plus_symbolic_constant_p): New.
7272 (nios2_valid_addr_expr_p): Recognize addresses involving
7274 (nios2_legitimate_address_p): Likewise, also LO_SUM.
7275 (nios2_symbolic_memory_operand_p): New.
7276 (nios2_large_constant_p): New.
7277 (nios2_split_large_constant): New.
7278 (nios2_split_plus_large_constant): New.
7279 (nios2_split_symbolic_memory_operand): New.
7280 (nios2_legitimize_address): Code refactoring. Handle addresses
7281 involving symbolic constants.
7282 (nios2_emit_move_sequence): Likewise.
7283 (nios2_print_operand): Improve error output.
7284 (nios2_print_operand_address): Handle LO_SUM.
7285 (nios2_cdx_narrow_form_p): Likewise.
7286 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
7287 operands involving symbolic constants.
7288 (movhi_internal, movsi_internal): Likewise.
7289 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
7290 (extendhisi2, extendqi<mode>2): Likewise.
7292 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7294 * tree-pass.h (PROP_rtl_split_insns): Define.
7295 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
7297 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7299 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
7301 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7304 * target.def (const_not_ok_for_debug_p): Default to
7305 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
7306 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
7307 * targhooks.c (default_const_not_ok_for_debug_p): New function.
7308 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
7309 which targetm.const_not_ok_for_debug_p returned true.
7310 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
7312 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
7314 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
7315 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
7316 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
7318 (ix86_const_not_ok_for_debug_p): New function.
7319 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
7320 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
7322 2017-10-23 David Malcolm <dmalcolm@redhat.com>
7325 * system.h: Conditionally include "unique-ptr.h" if
7326 INCLUDE_UNIQUE_PTR is defined.
7327 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
7328 of defining INCLUDE_UNIQUE_PTR before including "system.h".
7330 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
7332 * config/rl78/rl78.md: New define_expand "subdi3".
7334 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
7337 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
7338 DF_REF_INSN if DF_REF_INSN_INFO is false.
7340 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
7342 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
7343 xmm_move instead of sse_move.
7344 (sse_store_index): New function.
7345 (ix86_register_move_cost): Be more sensible about mismatch stall;
7346 model AVX moves correctly; make difference between sse->integer and
7348 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
7349 moves; make difference between SSE and AVX.
7350 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
7351 and zmm_move. Increase size of sse load and store tables;
7352 add unaligned load and store tables; add ssemmx_to_integer.
7353 * x86-tune-costs.h: Update all entries according to real
7354 move latencies from Agner Fog's manual and chip documentation.
7356 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7359 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
7360 * config/i386/constraints.md (Wf): New constraint.
7361 * config/i386/i386.md (UNSPEC_SBB): New unspec.
7362 (cmp<dwi>_doubleword): Removed.
7363 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
7364 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
7365 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
7366 expand with cmp<dwi>_doubleword. For LTU and GEU use
7367 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
7369 * common.opt (gcolumn-info): Enable by default.
7370 * doc/invoke.texi (gcolumn-info): Document new default.
7372 2017-10-23 Richard Biener <rguenther@suse.de>
7374 PR tree-optimization/82672
7375 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
7376 Fold the stmt if we propagated into it.
7378 2017-10-23 Richard Biener <rguenther@suse.de>
7380 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
7381 (bitmap_remove_expr_from_set): ... this. All callers call this
7382 for non-constant values.
7383 (bitmap_set_subtract): Rename to...
7384 (bitmap_set_subtract_expressions): ... this. Adjust and
7386 (bitmap_set_contains_value): Remove superfluous check.
7387 (bitmap_set_replace_value): Inline into single caller ...
7388 (bitmap_value_replace_in_set): ... here and simplify.
7389 (dependent_clean): Merge into ...
7390 (clean): ... this using an overload. Adjust.
7391 (prune_clobbered_mems): Adjust.
7392 (compute_antic_aux): Likewise.
7393 (compute_partial_antic_aux): Likewise.
7395 2017-10-23 Richard Biener <rguenther@suse.de>
7397 PR tree-optimization/82129
7399 2017-08-01 Richard Biener <rguenther@suse.de>
7401 PR tree-optimization/81181
7402 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7403 (compute_antic): ... end of iteration here.
7405 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7407 * target.def (starting_frame_offset): New hook.
7408 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
7409 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
7410 * doc/tm.texi.in: Regenerate.
7411 * hooks.h (hook_hwi_void_0): Declare.
7412 * hooks.c (hook_hwi_void_0): New function.
7413 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
7414 STARTING_FRAME_OFFSET.
7415 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
7416 * reload1.c (reload): Likewise.
7417 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
7418 instead of STARTING_FRAME_OFFSET.
7419 * function.c (try_fit_stack_local): Likewise.
7420 (assign_stack_local_1): Likewise
7421 (instantiate_virtual_regs): Likewise.
7422 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
7423 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
7424 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
7425 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
7426 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
7427 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
7428 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
7429 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
7430 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
7431 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
7432 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
7433 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
7434 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
7435 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
7436 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
7437 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
7438 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
7439 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
7440 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
7441 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
7442 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
7443 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
7444 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
7445 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
7446 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
7447 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
7448 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
7449 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
7450 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
7451 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
7452 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
7453 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
7454 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
7455 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
7456 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
7457 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
7458 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
7459 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
7460 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
7461 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
7462 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
7463 * config/avr/avr.c (avr_starting_frame_offset): Make static and
7464 return a HOST_WIDE_INT.
7465 (avr_builtin_setjmp_frame_value): Use it instead of
7466 STARTING_FRAME_OFFSET.
7467 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7468 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
7469 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
7471 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7472 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
7473 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
7474 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7475 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
7476 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
7477 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7478 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
7479 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
7480 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7481 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
7482 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
7484 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7485 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
7486 * config/mips/mips.c (mips_compute_frame_info): Refer to
7487 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
7488 (mips_starting_frame_offset): New function.
7489 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7490 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
7491 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
7492 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
7493 and return a HOST_WIDE_INT.
7494 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7495 (mmix_initial_elimination_offset): Refer to
7496 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
7497 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
7498 * config/pa/pa.c (pa_starting_frame_offset): New function.
7499 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
7500 (pa_expand_prologue): Likewise.
7501 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7502 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
7503 !FRAME_GROWS_DOWNWARD handling to...
7504 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7505 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
7506 !FRAME_GROWS_DOWNWARD handling to...
7507 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7508 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
7509 !FRAME_GROWS_DOWNWARD handling to...
7510 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7511 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
7513 (rs6000_starting_frame_offset): New function.
7514 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
7515 !FRAME_GROWS_DOWNWARD handling to...
7516 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7517 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
7518 !FRAME_GROWS_DOWNWARD handling to...
7519 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7520 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
7521 !FRAME_GROWS_DOWNWARD handling to...
7522 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7523 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
7524 (rs6000_starting_frame_offset): New function.
7525 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
7526 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
7527 * config/vax/vax.c (vax_starting_frame_offset): New function.
7528 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
7529 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7530 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
7531 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
7532 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7533 * system.h (STARTING_FRAME_OFFSET): Poison.
7535 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7537 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
7538 SCALAR_TYPE_MODE instead of TYPE_MODE.
7540 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7541 Alan Hayward <alan.hayward@arm.com>
7542 David Sherwood <david.sherwood@arm.com>
7544 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
7546 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7547 Alan Hayward <alan.hayward@arm.com>
7548 David Sherwood <david.sherwood@arm.com>
7550 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
7552 2017-10-23 Richard Biener <rguenther@suse.de>
7554 PR tree-optimization/82129
7555 * tree-ssa-pre.c (bitmap_set_and): Remove.
7556 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
7557 canonicalizing expressions in the set to those with lowest
7558 ID rather than taking that from the first edge.
7560 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7562 * combine.c (rtx_equal_for_field_assignment_p): Use
7563 byte_lowpart_offset.
7565 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7566 Alan Hayward <alan.hayward@arm.com>
7567 David Sherwood <david.sherwood@arm.com>
7569 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
7570 to a SUBREG_PROMOTED_VAR.
7572 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7573 Alan Hayward <alan.hayward@arm.com>
7574 David Sherwood <david.sherwood@arm.com>
7576 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
7577 (expand_debug_source_expr): Likewise.
7578 * combine.c (combine_simplify_rtx): Likewise.
7579 * cse.c (fold_rtx): Likewise.
7580 * optabs.c (expand_float): Likewise.
7581 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7582 (simplify_binary_operation_1): Likewise.
7584 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7585 Alan Hayward <alan.hayward@arm.com>
7586 David Sherwood <david.sherwood@arm.com>
7588 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
7589 (record_promoted_value): Likewise.
7590 * expr.c (expand_expr_real_2): Likewise.
7591 * ree.c (update_reg_equal_equiv_notes): Likewise.
7592 (combine_set_extension): Likewise.
7593 * rtlanal.c (low_bitmask_len): Likewise.
7594 * simplify-rtx.c (neg_const_int): Likewise.
7595 (simplify_binary_operation_1): Likewise.
7597 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7598 Alan Hayward <alan.hayward@arm.com>
7599 David Sherwood <david.sherwood@arm.com>
7601 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
7602 * regcprop.c (maybe_mode_change): Likewise.
7603 * reload1.c (alter_reg): Likewise.
7605 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7607 * inchash.h (inchash::hash::add_wide_int): New function.
7608 * lto-streamer-out.c (hash_tree): Use it.
7610 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7612 * inchash.h (inchash::hash::add_wide_int): Rename to...
7613 (inchash::hash::add_hwi): ...this.
7614 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
7615 (polymorphic_call_target_hasher::hash): Likewise.
7616 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
7617 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
7618 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
7619 * lto-streamer-out.c (hash_tree): Likewise.
7620 * optc-save-gen.awk: Likewise.
7621 * tree.c (add_expr): Likewise.
7623 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
7626 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
7627 for ordered inequality comparisons even with TARGET_IEEE_FP.
7629 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
7632 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
7633 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
7634 Expand with cmp<dwi>_doubleword.
7636 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7638 * extend.texi: Add x86 specific to 'nocf_check' attribute.
7639 List CET intrinsics.
7640 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
7641 specific to -fcf-protection option.
7643 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7645 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
7646 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
7647 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
7648 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
7649 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
7650 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
7651 (extra_objs): Add cet.o for Linux/x86 targets.
7652 (tmake_file): Add i386/t-cet for Linux/x86 targets.
7653 * config/i386/cet.c: New file.
7654 * config/i386/cetintrin.h: Likewise.
7655 * config/i386/t-cet: Likewise.
7656 * config/i386/cpuid.h (bit_SHSTK): New.
7657 (bit_IBT): Likewise.
7658 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
7659 pass IBT and SHSTK bits.
7660 * config/i386/i386-builtin-types.def
7661 (VOID_FTYPE_UNSIGNED_PVOID): New.
7662 (VOID_FTYPE_UINT64_PVOID): Likewise.
7663 * config/i386/i386-builtin.def: Add CET intrinsics.
7664 * config/i386/i386-c.c (ix86_target_macros_internal): Add
7665 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
7666 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
7667 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
7669 * config/i386/i386.c (rest_of_insert_endbranch): New.
7670 (pass_data_insert_endbranch): Likewise.
7671 (pass_insert_endbranch): Likewise.
7672 (make_pass_insert_endbranch): Likewise.
7673 (ix86_notrack_prefixed_insn_p): Likewise.
7674 (ix86_target_string): Add -mibt, -mshstk flags.
7675 (ix86_option_override_internal): Add flag_cf_protection
7677 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
7678 (ix86_print_operand): Add 'notrack' prefix output.
7679 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
7680 (ix86_expand_builtin): Expand CET intrinsics.
7681 (x86_output_mi_thunk): Add 'endbranch' instruction.
7682 * config/i386/i386.h (TARGET_IBT): New.
7683 (TARGET_IBT_P): Likewise.
7684 (TARGET_SHSTK): Likewise.
7685 (TARGET_SHSTK_P): Likewise.
7686 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
7687 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
7688 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
7689 (builtin_setjmp_setup): New pattern.
7690 (builtin_longjmp): Likewise.
7691 (rdssp<mode>): Likewise.
7692 (incssp<mode>): Likewise.
7693 (saveprevssp): Likewise.
7694 (rstorssp): Likewise.
7695 (wrss<mode>): Likewise.
7696 (wruss<mode>): Likewise.
7697 (setssbsy): Likewise.
7698 (clrssbsy): Likewise.
7699 (nop_endbr): Likewise.
7700 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
7702 * config/i386/immintrin.h: Include <cetintrin.h>.
7703 * config/i386/linux-common.h
7704 (file_end_indicate_exec_stack_and_cet): New prototype.
7705 (TARGET_ASM_FILE_END): New.
7707 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7709 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
7710 latencies instead of having separate table; make difference between
7711 integer and float costs.
7712 * i386.h (processor_costs): Remove scalar_stmt_cost,
7713 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
7714 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
7716 * x86-tune-costs.h: Remove entries which has been removed in
7717 procesor_costs from all tables; make cond_taken_branch_cost
7718 and cond_not_taken_branch_cost COST_N_INSNS based.
7720 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7722 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
7724 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7726 * config/i386/i386.md (isa): Remove fma_avx512f.
7727 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
7728 <avx512>_fmadd_<mode>_mask3<round_name>,
7729 <avx512>_fmsub_<mode>_mask<round_name>,
7730 <avx512>_fmsub_<mode>_mask3<round_name>,
7731 <avx512>_fnmadd_<mode>_mask<round_name>,
7732 <avx512>_fnmadd_<mode>_mask3<round_name>,
7733 <avx512>_fnmsub_<mode>_mask<round_name>,
7734 <avx512>_fnmsub_<mode>_mask3<round_name>,
7735 <avx512>_fmaddsub_<mode>_mask<round_name>,
7736 <avx512>_fmaddsub_<mode>_mask3<round_name>,
7737 <avx512>_fmsubadd_<mode>_mask<round_name>,
7738 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
7739 (*vec_widen_umult_even_v16si<mask_name>,
7740 *vec_widen_smult_even_v16si<mask_name>): Likewise.
7741 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
7743 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7745 * extend.texi: Add 'nocf_check' documentation.
7746 * gimple.texi: Add second parameter to
7747 gimple_build_call_from_tree.
7748 * invoke.texi: Add -fcf-protection documentation.
7749 * rtl.texi: Add REG_CALL_NOTRACK documenation.
7751 2017-10-20 Richard Biener <rguenther@suse.de>
7753 PR tree-optimization/82473
7754 * tree-vect-loop.c (vectorizable_reduction): Properly get at
7755 the largest input type.
7757 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7759 * c-attribs.c (handle_nocf_check_attribute): New function.
7760 (c_common_attribute_table): Add 'nocf_check' handling.
7761 * gimple-parser.c: Add second argument NULL to
7762 gimple_build_call_from_tree.
7763 * attrib.c (comp_type_attributes): Check nocf_check attribute.
7764 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
7766 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
7767 * common.opt: Add fcf-protection flag.
7768 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
7769 * flag-types.h: Add enum cf_protection_level.
7770 * gimple.c (gimple_build_call_from_tree): Add second parameter.
7771 Add 'nocf_check' attribute propagation to gimple call.
7772 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
7773 (gimple_build_call_from_tree): Update prototype.
7774 (gimple_call_nocf_check_p): New function.
7775 (gimple_call_set_nocf_check): Likewise.
7776 * gimplify.c: Add second argument to gimple_build_call_from_tree.
7777 * ipa-icf.c: Add nocf_check attribute in statement hash.
7778 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
7779 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
7780 * toplev.c (process_options): Add flag_cf_protection handling.
7782 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7784 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
7786 2017-10-20 Richard Biener <rguenther@suse.de>
7788 PR tree-optimization/82603
7789 * tree-if-conv.c (predicate_mem_writes): Make sure to only
7790 remove false predicated stores.
7792 2017-10-20 Richard Biener <rguenther@suse.de>
7794 * graphite-isl-ast-to-gimple.c
7795 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
7796 Remove return value and simplify, dump copied stmt after lhs
7798 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
7799 Reduce dump verbosity.
7800 (gsi_insert_earliest): Likewise.
7801 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
7802 * graphite.c (print_global_statistics): Adjust dumping.
7803 (print_graphite_scop_statistics): Likewise.
7804 (print_graphite_statistics): Do not dump loops here.
7805 (graphite_transform_loops): But here.
7807 2017-10-20 Nicolas Roche <roche@adacore.com>
7809 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
7810 * configure: Regenerate.
7812 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7815 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
7816 functions when optimizing replace GIMPLE_RETURN stmts with
7817 calls to __builtin_unreachable ().
7820 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
7821 for -fsanitize=thread link of executables.
7822 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
7823 link of executables.
7826 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
7828 (<shift_insn><mode>3<mask_name>): Change the last of the 3
7829 define_insns for logical vector shifts to use VI248_AVX512BW
7830 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
7831 condition, useless isa and prefix attributes. Change the first
7832 2 of these define_insns to ...
7833 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
7834 define_insn for avx512vl.
7835 (<shift_insn><mode>3): ... and this, new define_insn without
7836 masking for non-avx512vl.
7839 * config/i386/sse.md (*andnot<mode>3,
7840 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
7841 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
7842 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
7843 not applied use empty suffix even for TARGET_AVX512VL.
7844 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
7845 is applied, supply evex,evex or evex,evex,evex instead of just
7848 2017-10-20 Julia Koval <julia.koval@intel.com>
7850 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
7851 (OPTION_MASK_ISA_GFNI_UNSET): New.
7852 (ix86_handle_option): Handle OPT_mgfni.
7853 * config/i386/cpuid.h (bit_GFNI): New.
7854 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
7855 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
7856 * config/i386/i386.c (ix86_target_string): Add -mgfni.
7857 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
7858 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
7859 * config/i386/i386.opt: Add mgfni.
7861 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
7863 * config/msp430/msp430.c (msp430_option_override): Disable
7864 -fdelete-null-pointer-checks.
7865 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
7867 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7869 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
7870 of x87 and SSE instructions.
7872 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7874 * asan.c (create_cond_insert_point): Do not update edge count.
7875 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
7876 (afdo_propagate_circuit): Likewise.
7877 (afdo_calculate_branch_prob): Likewise.
7878 (afdo_annotate_cfg): Likewise.
7879 * basic-block.h (struct edge_def): Remove count.
7880 (edge_def::count): New accessor.
7881 * bb-reorder.c (rotate_loop): Update.
7882 (find_traces_1_round): Update.
7883 (connect_traces): Update.
7884 (sanitize_hot_paths): Update.
7885 * cfg.c (unchecked_make_edge): Update.
7886 (make_single_succ_edge): Update.
7887 (check_bb_profile): Update.
7888 (dump_edge_info): Update.
7889 (update_bb_profile_for_threading): Update.
7890 (scale_bbs_frequencies_int): Update.
7891 (scale_bbs_frequencies_gcov_type): Update.
7892 (scale_bbs_frequencies_profile_count): Update.
7893 (scale_bbs_frequencies): Update.
7894 * cfganal.c (connect_infinite_loops_to_exit): Update.
7895 * cfgbuild.c (compute_outgoing_frequencies): Update.
7896 (find_many_sub_basic_blocks): Update.
7897 * cfgcleanup.c (try_forward_edges): Update.
7898 (try_crossjump_to_edge): Update
7899 * cfgexpand.c (expand_gimple_cond): Update
7900 (expand_gimple_tailcall): Update
7901 (construct_exit_block): Update
7902 * cfghooks.c (verify_flow_info): Update
7903 (redirect_edge_succ_nodup): Update
7904 (split_edge): Update
7905 (make_forwarder_block): Update
7906 (duplicate_block): Update
7907 (account_profile_record): Update
7908 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
7909 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
7910 * cfgloopmanip.c (scale_loop_profile): Update.
7912 (lv_adjust_loop_entry_edge): Update.
7913 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
7914 (force_nonfallthru_and_redirect): Update.
7915 (purge_dead_edges): Update.
7916 (rtl_flow_call_edges_add): Update.
7917 * cgraphunit.c (init_lowered_empty_function): Update.
7918 (cgraph_node::expand_thunk): Update.
7919 * gimple-pretty-print.c (dump_probability): Update.
7920 (dump_edge_probability): Update.
7921 * gimple-ssa-isolate-paths.c (isolate_path): Update.
7922 * haifa-sched.c (sched_create_recovery_edges): Update.
7923 * hsa-gen.c (convert_switch_statements): Update.
7924 * ifcvt.c (dead_or_predicable): Update.
7925 * ipa-inline-transform.c (inline_transform): Update.
7926 * ipa-split.c (split_function): Update.
7927 * ipa-utils.c (ipa_merge_profiles): Update.
7928 * loop-doloop.c (add_test): Update.
7929 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
7930 * lto-streamer-in.c (input_cfg): Update.
7931 (input_function): Update.
7932 * lto-streamer-out.c (output_cfg): Update.
7933 * modulo-sched.c (sms_schedule): Update.
7934 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
7935 * predict.c (maybe_hot_edge_p): Update.
7936 (unlikely_executed_edge_p): Update.
7937 (probably_never_executed_edge_p): Update.
7938 (dump_prediction): Update.
7939 (drop_profile): Update.
7940 (propagate_unlikely_bbs_forward): Update.
7941 (determine_unlikely_bbs): Update.
7942 (force_edge_cold): Update.
7943 * profile.c (compute_branch_probabilities): Update.
7944 * reg-stack.c (better_edge): Update.
7945 * shrink-wrap.c (handle_simple_exit): Update.
7946 * tracer.c (better_p): Update.
7947 * trans-mem.c (expand_transaction): Update.
7948 (split_bb_make_tm_edge): Update.
7949 * tree-call-cdce.c: Update.
7950 * tree-cfg.c (gimple_find_sub_bbs): Update.
7951 (gimple_split_edge): Update.
7952 (gimple_duplicate_sese_region): Update.
7953 (gimple_duplicate_sese_tail): Update.
7954 (gimple_flow_call_edges_add): Update.
7955 (insert_cond_bb): Update.
7956 (execute_fixup_cfg): Update.
7957 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
7958 * tree-complex.c (expand_complex_div_wide): Update.
7959 * tree-eh.c (lower_resx): Update.
7960 (unsplit_eh): Update.
7961 (cleanup_empty_eh_move_lp): Update.
7962 * tree-inline.c (copy_edges_for_bb): Update.
7963 (freqs_to_counts): Update.
7964 (copy_cfg_body): Update.
7965 * tree-ssa-dce.c (remove_dead_stmt): Update.
7966 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
7967 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
7968 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
7969 (unloop_loops): Update.
7970 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
7971 * tree-ssa-loop-split.c (connect_loops): Update.
7972 (split_loop): Update.
7973 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
7974 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
7975 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
7976 * tree-ssa-reassoc.c (branch_fixup): Update.
7977 * tree-ssa-tail-merge.c (replace_block_by): Update.
7978 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
7979 (compute_path_counts): Update.
7980 (update_profile): Update.
7981 (recompute_probabilities): Update.
7982 (update_joiner_offpath_counts): Update.
7983 (estimated_freqs_path): Update.
7984 (freqs_to_counts_path): Update.
7985 (clear_counts_path): Update.
7986 (ssa_fix_duplicate_block_edges): Update.
7987 (duplicate_thread_path): Update.
7988 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
7989 (case_bit_test_cmp): Update.
7990 (collect_switch_conv_info): Update.
7991 (gen_inbound_check): Update.
7992 (do_jump_if_equal): Update.
7993 (emit_cmp_and_jump_insns): Update.
7994 * tree-tailcall.c (decrease_profile): Update.
7995 (eliminate_tail_call): Update.
7996 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
7997 (vect_do_peeling): Update.
7998 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
7999 * ubsan.c (ubsan_expand_null_ifn): Update.
8000 (ubsan_expand_ptr_ifn): Update.
8001 * value-prof.c (gimple_divmod_fixed_value): Update.
8002 (gimple_mod_pow2): Update.
8003 (gimple_mod_subtract): Update.
8004 (gimple_ic): Update.
8005 (gimple_stringop_fixed_value): Update.
8007 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8010 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
8012 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
8014 PR rtl-optimization/82395
8015 * ira-color.c (allocno_priority_compare_func): Fix comparison step
8016 based on non_spilled_static_chain_regno_p.
8018 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8020 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
8021 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
8022 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
8024 2017-10-19 Martin Sebor <msebor@redhat.com>
8026 PR tree-optimization/82596
8027 * tree.c (array_at_struct_end_p): Handle STRING_CST.
8029 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8031 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
8032 (get_mem_refs_of_builtin_call): Likewise.
8033 * builtins.c (expand_builtin_apply): Adjust call to
8034 allocate_dynamic_stack_space.
8035 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
8036 the third argument to allocate_dynamic_stack_space, otherwise -1.
8037 (expand_builtin): Deal with all alloca variants.
8038 (is_inexpensive_builtin): Likewise.
8039 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
8040 * calls.c (special_function_p): Deal with all alloca variants.
8041 (initialize_argument_information): Adjust call to
8042 allocate_dynamic_stack_space.
8043 (expand_call): Likewise.
8044 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
8045 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
8046 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
8047 use it for the stack usage computation.
8048 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
8049 * function.c (gimplify_parameters): Call build_alloca_call_expr.
8050 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
8051 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
8052 (in_loop_p): Remove first argument and useless check.
8053 (pass_walloca::execute): Remove useless test and adjust call to above.
8054 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
8055 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
8056 (gimplify_call_expr): Deal with all alloca variants.
8057 * hsa-gen.c (gen_hsa_alloca): Likewise.
8058 (gen_hsa_insns_for_call): Likewise.
8059 * ipa-pure-const.c (special_builtin_state): Likewise.
8060 * tree-chkp.c (chkp_build_returned_bound): Likewise.
8061 * tree-object-size.c (alloc_object_size): Likewise.
8062 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
8063 (call_may_clobber_ref_p_1): Likewise.
8064 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
8065 (ccp_fold_stmt): Likewise.
8066 (optimize_stack_restore): Likewise.
8067 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
8068 (mark_all_reaching_defs_necessary_1): Likewise.
8069 (propagate_necessity): Likewise.
8070 (eliminate_unnecessary_stmts): Likewise.
8071 * tree.c (build_common_builtin_nodes): Build
8072 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
8073 (build_alloca_call_expr): New function.
8074 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
8075 (CASE_BUILT_IN_ALLOCA): Likewise.
8076 (build_alloca_call_expr): Declare.
8077 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
8079 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8082 * dwarf2out.c (new_die_raw): New static inline function.
8083 (new_die): Use it to create the DIE.
8084 (add_AT_external_die_ref): Likewise.
8085 (clone_die): Likewise.
8086 (clone_as_declaration): Likewise.
8087 (dwarf2out_vms_debug_main_pointer): Likewise.
8088 (base_type_die): Likewise. Remove early return for corner cases.
8089 Do not call add_pubtype on the DIE here.
8090 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
8091 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
8092 typedefs for base types with DW_AT_endianity. Make sure a DIE with
8093 native order exists for base types, attach the DIE manually and call
8094 add_pubtype on it. Do not equate a reverse order DIE to the type.
8096 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
8098 * config/arm/arm.c (align_ok_ldrd_strd): New function.
8099 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
8101 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
8103 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8105 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
8106 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
8107 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
8108 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
8109 * opts.c (sanitizer_opts): Add builtin.
8110 * ubsan.c (instrument_builtin): New function.
8111 (pass_ubsan::execute): Call it.
8112 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
8113 * doc/invoke.texi: Document -fsanitize=builtin.
8115 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
8116 builtins, store max (log2 (align), 0) into uchar field instead of
8117 align into uptr field.
8118 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
8119 store uchar 0 field instead of uptr 0 field.
8120 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
8121 instead of passing one address of struct with 2 locations pass
8122 two addresses of structs with 1 location each.
8123 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
8124 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
8125 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
8126 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
8127 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
8128 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
8129 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
8130 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
8132 2017-10-19 Martin Liska <mliska@suse.cz>
8135 * file-find.c (remove_prefix): Remove.
8136 * file-find.h (remove_prefix): Likewise.
8137 * gcc-ar.c: Remove smartness of lookup.
8139 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
8141 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
8142 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
8143 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
8145 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8148 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
8149 (setcc + and to xor + setcc): New peephole2.
8151 2017-10-19 Tom de Vries <tom@codesourcery.com>
8153 * doc/sourcebuild.texi (Test Directives, Variants of
8154 dg-require-support): Add dg-require-stack-size.
8156 2017-10-19 Martin Liska <mliska@suse.cz>
8159 * gimplify.c (gimplify_decl_expr): Do not instrument variables
8160 that have a large alignment.
8161 (gimplify_target_expr): Likewise.
8163 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
8165 PR rtl-optimization/82602
8166 * ira.c (rtx_moveable_p): Return false for volatile asm.
8168 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
8171 * config/i386/i386-modes.def (CCGZ): New CC mode.
8172 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
8173 * config/i386/predicates.md (ix86_comparison_operator):
8175 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
8176 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
8177 with double-word subtraction.
8178 (put_condition_code): Handle CCGZmode.
8180 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
8182 * wide-int.cc (debug (const wide_int &)): New.
8183 (debug (const wide_int *)): New.
8184 (debug (const widest_int &)): New.
8185 (debug (const widest_int *)): New.
8187 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
8190 * lra-constraints.c (curr_insn_transform): Use non-input operand
8191 instead of output one for matched reload.
8193 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8195 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
8196 (tree-ssa-loop-ivopts.h): New header file.
8197 (struct builtin_info): New fields.
8198 (classify_builtin_1): Compute and record base and offset parts for
8199 memset builtin partition by calling strip_offset.
8200 (offset_cmp, fuse_memset_builtins): New functions.
8201 (finalize_partitions): Fuse adjacent memset partitions by calling
8203 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
8204 Expose the interface.
8205 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
8207 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8209 PR tree-optimization/82574
8210 * tree-loop-distribution.c (find_single_drs): New parameter. Check
8211 that data reference must be executed exactly once per iteration
8212 against the outermost loop in nest.
8213 (classify_partition): Update call to above function.
8215 2017-10-18 Richard Biener <rguenther@suse.de>
8217 PR tree-optimization/82591
8218 * graphite.c (graphite_transform_loops): Move code gen message
8220 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8221 Here. Handle scop_to_isl_ast failing.
8222 (scop_to_isl_ast): Limit the number of ISL operations.
8224 2017-10-18 Richard Biener <rguenther@suse.de>
8226 * graphite-isl-ast-to-gimple.c
8227 (translate_isl_ast_to_gimple::set_rename): Simplify.
8228 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
8229 (graphite_copy_stmts_from_block): ... here.
8230 (copy_bb_and_scalar_dependences): Simplify.
8231 (add_parameters_to_ivs_params): Canonicalize.
8232 (generate_entry_out_of_ssa_copies): Simplify.
8233 * graphite-sese-to-poly.c (extract_affine_name): Simplify
8234 by passing in ISL dimension.
8235 (parameter_index_in_region_1): Rename to ...
8236 (parameter_index_in_region): ... this.
8237 (extract_affine): Adjust assert, pass down parameter index.
8238 (add_param_constraints): Use range-info when available.
8239 (build_scop_context): Adjust.
8240 * sese.c (new_sese_info): Adjust.
8241 (free_sese_info): Likewise.
8242 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
8243 Remove unused typedefs.
8244 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
8246 2017-10-18 Martin Liska <mliska@suse.cz>
8248 * combine.c (simplify_compare_const): Add gcc_fallthrough.
8250 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8252 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
8253 (s390_sched_init): Do not reset s390_sched_state if we entered the
8254 current basic block via a fallthru edge and all others are unlikely.
8256 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8258 * config/s390/s390.c (NUM_SIDES): New variable.
8259 (LONGRUNNING_THRESHOLD): New variable.
8260 (LATENCY_FACTOR): New variable.
8261 (s390_sched_score): Decrease score for long-running instructions on
8263 (s390_sched_variable_issue): Perform bookkeeping for long-running
8266 2017-10-18 Richard Biener <rguenther@suse.de>
8268 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
8269 Simplify with removal of the parameter rename map.
8270 (set_rename): Likewise.
8271 (should_copy_to_new_region): Likewise.
8272 (graphite_copy_stmts_from_block): Likewise.
8273 (copy_bb_and_scalar_dependences): Remove initialization of
8274 unused copied_bb_map.
8276 (copy_internal_parameters): Likewise.
8277 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
8278 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
8279 Use INTEGRAL_TYPE_P.
8280 (parameter_index_in_region_1): Rename to ...
8281 (assign_parameter_index_in_region): ... this. Assert we have
8282 a parameter we handle.
8283 (scan_tree_for_params): Adjust.
8284 * sese.h (parameter_rename_map_t): Remove.
8285 (struct sese_info_t): Remove unused parameter_rename_map and
8286 copied_bb_map members.
8287 * sese.c (new_sese_info): Adjust.
8288 (free_sese_info): Likewise.
8290 2017-10-18 Martin Liska <mliska@suse.cz>
8293 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
8294 on an abnormal edge.
8296 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
8298 * doc/invoke.texi (ffunction-sections and fdata-sections):
8301 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8303 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
8304 the use statement can throw internally.
8306 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8308 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
8309 any RTX present on the RHS of a SET.
8310 * compare-elim.c (try_eliminate_compare): Restore comment.
8312 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8314 * langhooks.h (struct lang_hooks): Document that tree_size langhook
8315 may be also called on tcc_type nodes.
8316 * langhooks.c (lhd_tree_size): Likewise.
8318 2017-10-17 David Malcolm <dmalcolm@redhat.com>
8320 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
8321 format_warning_at_substring.
8322 (maybe_warn): Convert source_range * param to a location_t. Pass
8323 UNKNOWN_LOCATION rather than NULL to fmtwarn.
8324 (format_directive): Remove code to extract source_ranges and
8325 source_range * in favor of just a location_t.
8326 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
8328 * substring-locations.c (format_warning_va): Convert
8329 source_range * param to a location_t.
8330 (format_warning_at_substring): Likewise.
8331 * substring-locations.h (format_warning_va): Likewise.
8332 (format_warning_at_substring): Likewise.
8334 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
8336 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
8338 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
8339 and scatter/gather ops.
8341 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
8342 vec_gather_load and vec_scatter_store.
8343 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
8344 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
8345 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
8346 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
8347 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
8348 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
8350 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
8352 * reg-stack.c (compare_for_stack_reg): Add bool argument.
8353 Detect FTST instruction and handle its register pops. Only pop
8354 second operand if can_pop_second_op is true.
8355 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
8356 set can_pop_second_op to false in the compare_for_stack_reg call.
8358 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
8359 output_fp_compare for stack register operands.
8360 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
8361 instructions here. Do not emit stack register pops here. Assert
8362 that FCOMPP pops next to top stack register. Rewrite function.
8364 2017-10-17 Nathan Sidwell <nathan@acm.org>
8367 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
8368 use DECL_ASSEMBLER_NAME_RAW.
8371 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
8374 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
8375 Wilco Dijkstra <wilco.dijkstra@arm.com>
8377 * builtins.c (expand_builtin_update_setjmp_buf): Add a
8378 converstion to Pmode from the buf_addr.
8380 2017-10-17 Richard Biener <rguenther@suse.de>
8382 * graphite-dependences.c (scop_get_reads_and_writes): Change
8383 output parameters to references.
8385 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
8387 PR 71026/tree-optimization
8388 * fold-const.c (distribute_real_division): Removed.
8389 (fold_binary_loc): Remove calls to distribute_real_divison.
8391 2017-10-17 Richard Biener <rguenther@suse.de>
8393 * graphite-scop-detection.c
8394 (scop_detection::stmt_has_simple_data_refs_p): Always use
8395 the full nest as region.
8396 (try_generate_gimple_bb): Likewise.
8397 * sese.c (scalar_evolution_in_region): Simplify now that
8398 SCEV can handle instantiation in regions.
8399 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
8400 in the non-loop part of a function if requested.
8402 2017-10-17 Richard Biener <rguenther@suse.de>
8404 PR tree-optimization/82563
8405 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
8407 (graphite_regenerate_ast_isl): Call it.
8408 * graphite-scop-detection.c (build_scops): Remove entry edge split.
8410 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8412 PR tree-optimization/82549
8413 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
8414 Formatting fixes. Instead of calling make_bit_field_ref with negative
8417 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
8419 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
8420 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
8421 _mm_maskz_reduce_ss): New.
8422 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
8423 __builtin_ia32_reducess_mask): Ditto..
8424 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
8425 * config/i386/sse.md (reduces<mode>): Renamed to ...
8426 (reduces<mode><mask_scalar_name>): ... this.
8427 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
8429 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
8430 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
8432 2017-10-16 David Malcolm <dmalcolm@redhat.com>
8434 * Makefile.in (OBJS): Add unique-ptr-tests.o.
8435 * selftest-run-tests.c (selftest::run_tests): Call
8436 selftest::unique_ptr_tests_cc_tests.
8437 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
8438 * unique-ptr-tests.cc: New file.
8440 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
8443 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8445 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8446 (make_hard_regno_born, make_hard_regno_dead): Update
8447 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
8449 2017-10-16 Jeff Law <law@redhat.com>
8451 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
8453 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
8455 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
8457 2017-10-16 Olivier Hainque <hainque@adacore.com>
8459 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
8460 with_cpu if we were configured for an e500v2 target cpu name.
8462 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
8464 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
8465 * doc/invoke.texi: Document +nodsp as a valid extension for
8468 2017-10-16 Martin Liska <mliska@suse.cz>
8470 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
8471 (test_set_range): Likewise.
8472 (test_range_functions): Rename to ...
8473 (test_bit_in_range): ... this.
8474 (sbitmap_c_tests): Add new test.
8476 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8478 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
8480 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
8481 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
8483 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8485 * config/aarch64/aarch64-builtins.c
8486 (aarch64_types_quadopu_lane_qualifiers): New.
8487 (TYPES_QUADOPU_LANE): New.
8488 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
8489 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
8490 (aarch64_<sur>dot_laneq<vsi2qi>): New.
8491 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
8492 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
8493 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
8494 (Vdottype, DOTPROD): New.
8495 (sur): Add SDOT and UDOT.
8497 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8499 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
8500 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
8501 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
8503 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
8504 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
8505 Enable TARGET_DOTPROD.
8506 (cortex-a75.cortex-a55): Likewise.
8507 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
8509 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8511 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
8512 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
8514 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
8516 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
8517 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
8518 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
8519 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
8520 * config/arm/types.md (neon_dot, neon_dot_q): New.
8521 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
8523 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8525 * config/arm/arm.h (TARGET_DOTPROD): New.
8526 * config/arm/arm.c (arm_arch_dotprod): New.
8527 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
8528 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
8529 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
8530 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
8531 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
8532 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
8533 * doc/invoke.texi (armv8.2-a): Document dotprod
8535 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
8537 * i386.c (ix86_vec_cost): New function.
8538 (ix86_rtx_costs): Handle vector operations better.
8539 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
8540 * x86-tune-costs.h: Add new costs to all tables.
8542 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
8544 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
8546 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
8547 divsd, sqrtss and sqrtsd
8548 * x86-tune-costs.h: Add new entries to all costs.
8549 (znver1_cost): Fix to match real instruction latencies.
8551 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8552 Michael Collison <michael.collison@arm.com>
8554 * compare-elim.c: Include emit-rtl.h.
8555 (can_merge_compare_into_arith): New function.
8556 (try_validate_parallel): Likewise.
8557 (try_merge_compare): Likewise.
8558 (try_eliminate_compare): Call the above when no previous clobber
8560 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
8563 2017-10-14 Jakub Jelinek <jakub@redhat.com>
8567 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
8568 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
8572 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
8573 to be any operand_equal_p operands. For & (B - 1) require
8574 B to be power of 2. Recognize
8575 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
8577 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8580 * optabs.c (expand_memory_blockage): Fix call of
8581 targetm.have_memory_blockage.
8583 2017-10-14 Jakub Jelinek <jakub@redhat.com>
8586 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
8587 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
8588 objects to extra_objs instead of overwriting it.
8590 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8592 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
8593 Use any_fp_register_operand as operand[3] predicate. Simplify
8594 equality test for operands[2] and operands[4] memory location.
8595 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
8596 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
8597 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
8598 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
8599 any_fp_register_operand as operand[1] predicate. Simplify
8600 equality test for operands[0] and operands[3] memory location.
8601 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
8602 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
8603 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
8605 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8607 * target-insns.def: Add memory_blockage.
8608 * optabs.c (expand_memory_blockage): New function.
8609 (expand_asm_memory_barrier): Rename ...
8610 (expand_asm_memory_blockage): ... to this.
8611 (expand_mem_thread_fence): Call expand_memory_blockage
8612 instead of expand_asm_memory_barrier.
8613 (expand_mem_singnal_fence): Ditto.
8614 (expand_atomic_load): Ditto.
8615 (expand_atomic_store): Ditto.
8616 * doc/md.texi (Standard Pattern Names For Generation):
8617 Document memory_blockage instruction pattern.
8619 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
8621 * config/rl78/rl78.c (rl78_emit_libcall): New function.
8622 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
8623 * config/rl78/rl78.md: New define_expand "adddi3".
8625 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
8627 * cfghooks.c (verify_flow_info): Disable check that all probabilities
8630 2017-10-13 Jeff Law <law@redhat.com>
8632 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
8634 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8637 * internal-fn.c (expand_mul_overflow): If both operands have
8638 the same highpart of -1 or 0 and the topmost bit of lowpart
8639 is different, overflow is if res <= 0 rather than res < 0.
8641 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
8643 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
8644 TARGET_P9_VECTOR code for unaligned_load case.
8646 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
8648 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
8650 2017-10-13 Nathan Sidwell <nathan@acm.org>
8652 * tree-core.h (tree_contains_struct): Make bool.
8653 * tree.c (tree_contains_struct): Likewise.
8654 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
8655 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
8656 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
8657 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
8658 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
8660 2017-10-13 Richard Biener <rguenther@suse.de>
8662 * graphite-isl-ast-to-gimple.c
8663 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
8664 parameters and dominance check.
8665 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
8666 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
8667 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
8668 Do not update SSA form here or do intermediate IL verification.
8669 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
8670 (graphite_initialize): Remove check on the number of loops in
8671 the function and inline into graphite_transform_loops.
8672 (graphite_finalize): Inline into graphite_transform_loops.
8673 (graphite_transform_loops): Perform SSA update and IL verification
8675 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
8677 2017-10-13 Richard Biener <rguenther@suse.de>
8679 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
8680 graphite_expression_type_precision): Avoid global constructor
8682 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
8683 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
8684 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
8685 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
8686 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
8687 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
8689 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
8692 * config/i386/i386.h (ix86_red_zone_size): New.
8693 * config/i386/i386.md (push peephole2s): Replace
8694 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
8696 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8697 Alan Hayward <alan.hayward@arm.com>
8698 David Sherwood <david.sherwood@arm.com>
8700 * combine.c (can_change_dest_mode): Reject changes in
8701 REGMODE_NATURAL_SIZE.
8703 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8704 Alan Hayward <alan.hayward@arm.com>
8705 David Sherwood <david.sherwood@arm.com>
8707 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
8708 (expand_debug_source_expr): Likewise.
8709 * combine.c (combine_simplify_rtx): Likewise.
8710 * cse.c (fold_rtx): Likewise.
8711 * fwprop.c (canonicalize_address): Likewise.
8712 * targhooks.c (default_shift_truncation_mask): Likewise.
8714 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8715 Alan Hayward <alan.hayward@arm.com>
8716 David Sherwood <david.sherwood@arm.com>
8718 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
8719 (widened_mode): Likewise.
8720 (expand_unop): Likewise.
8721 * ree.c (transform_ifelse): Likewise.
8722 (merge_def_and_ext): Likewise.
8723 (combine_reaching_defs): Likewise.
8724 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8726 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8727 Alan Hayward <alan.hayward@arm.com>
8728 David Sherwood <david.sherwood@arm.com>
8730 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
8731 * combine.c (gen_lowpart_for_combine): Likewise.
8732 * dwarf2out.c (rtl_for_decl_location): Likewise.
8733 * final.c (alter_subreg): Likewise.
8734 * rtlhooks.c (gen_lowpart_general): Likewise.
8735 (gen_lowpart_if_possible): Likewise.
8737 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8738 Alan Hayward <alan.hayward@arm.com>
8739 David Sherwood <david.sherwood@arm.com>
8741 * calls.c (expand_call): Use subreg_lowpart_offset.
8742 * cse.c (cse_insn): Likewise.
8743 * regcprop.c (copy_value): Likewise.
8744 (copyprop_hardreg_forward_1): Likewise.
8746 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8749 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
8750 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
8751 =Q constraints to +Q and into insn condition add check
8752 that operands[0] and operands[1] are equal.
8753 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
8754 =Q constraints to +Q and into insn condition add check
8755 that operands[0] is equal to either operands[1] or operands[2].
8758 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
8759 instead of handling MINUS_EXPR twice (once for each argument),
8760 canonicalize operand order and handle just once, use rtype where
8761 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
8764 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
8765 any values of __C while still being pattern recognizable as a simple
8768 2017-10-13 Richard Biener <rguenther@suse.de>
8770 PR tree-optimization/82451
8772 2017-10-02 Richard Biener <rguenther@suse.de>
8774 PR tree-optimization/82355
8775 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
8776 a mapping for the enclosing loop but avoid generating one for
8778 (copy_bb_and_scalar_dependences): Remove premature codegen
8779 error on PHIs in blocks duplicated into multiple places.
8780 * graphite-scop-detection.c
8781 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
8782 in the region use it as loop and nest to analyze the DR in.
8783 (try_generate_gimple_bb): Likewise.
8784 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
8785 (add_loop_constraints): For blocks in a loop not in the region
8786 create a dimension with a single iteration.
8787 * sese.h (gbb_loop_at_index): Remove assert.
8789 * cfgloop.c (loop_preheader_edge): For the loop tree root
8790 return the single successor of the entry block.
8791 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8792 Reset the SCEV hashtable and niters.
8793 * graphite-scop-detection.c
8794 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
8795 assert that we only have POLYNOMIAL_CHREC that vary in loops
8796 contained in the region.
8797 (scop_detection::graphite_can_represent_expr): Adjust.
8798 (scop_detection::stmt_has_simple_data_refs_p): For loops
8799 not in the region set loop to NULL. The nest is now the
8800 entry edge to the region.
8801 (try_generate_gimple_bb): Likewise.
8802 * sese.c (scalar_evolution_in_region): Adjust for
8803 instantiate_scev change.
8804 * tree-data-ref.h (graphite_find_data_references_in_stmt):
8805 Make nest parameter the edge into the region.
8806 (create_data_ref): Likewise.
8807 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
8808 entry edge into a region and adjust instantiate_scev calls.
8809 (create_data_ref): Likewise.
8810 (graphite_find_data_references_in_stmt): Likewise.
8811 (find_data_references_in_stmt): Pass the loop preheader edge
8812 from the nest argument.
8813 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
8814 parameter the edge into the region.
8815 (instantiate_parameters): Use the loop preheader edge as entry.
8816 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
8818 (get_instantiated_value_entry): Make instantiate_below parameter
8819 the edge into the region.
8820 (instantiate_scev_name): Likewise. Adjust dominance checks,
8821 when we cannot use loop-based instantiation instantiate by
8822 walking use-def chains.
8823 (instantiate_scev_poly): Adjust.
8824 (instantiate_scev_binary): Likewise.
8825 (instantiate_scev_convert): Likewise.
8826 (instantiate_scev_not): Likewise.
8827 (instantiate_array_ref): Remove.
8828 (instantiate_scev_3): Likewise.
8829 (instantiate_scev_2): Likewise.
8830 (instantiate_scev_1): Likewise.
8831 (instantiate_scev_r): Do not blindly handle N-operand trees.
8832 Do not instantiate array-refs. Handle all constants and invariants.
8833 (instantiate_scev): Make instantiate_below parameter
8834 the edge into the region.
8835 (resolve_mixers): Use the loop preheader edge for the region
8836 parameter to instantiate_scev_r.
8837 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
8839 2017-10-13 Richard Biener <rguenther@suse.de>
8841 PR tree-optimization/82525
8842 * graphite-isl-ast-to-gimple.c
8843 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
8845 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
8846 Fail code generation when we cannot represent the isl integer.
8847 (binary_op_to_tree): Elide modulo operations that are no-ops
8848 in the type we code generate. Remove now superfluous code
8851 2017-10-13 Richard Biener <rguenther@suse.de>
8853 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
8854 (scop_detection::harmful_loop_in_region): Remove premature
8855 IV type restriction.
8856 (scop_detection::graphite_can_represent_scev): We can handle
8857 pointer IVs just fine.
8859 2017-10-13 Alan Modra <amodra@gmail.com>
8861 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
8862 "Clobbers and Scratch Registers". Add paragraph on
8863 alternative to clobbers for scratch registers and OpenBLAS
8866 2017-10-13 Alan Modra <amodra@gmail.com>
8868 * doc/extend.texi (Clobbers): Correct vax example. Delete old
8869 example of a memory input for a string of known length. Move
8870 commentary out of table. Add a number of new examples
8871 covering array memory inputs.
8873 2017-10-12 Martin Liska <mliska@suse.cz>
8875 PR tree-optimization/82493
8876 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
8877 (test_range_functions): New function.
8878 (sbitmap_c_tests): Likewise.
8879 * selftest-run-tests.c (selftest::run_tests): Run new tests.
8880 * selftest.h (sbitmap_c_tests): New function.
8882 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
8884 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
8886 * config/rs6000/amo.h: Fix spacing issue.
8888 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8891 * config/i386/i386.md (*ashl<mode>3_mask_1,
8892 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
8893 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
8896 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
8898 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
8899 (profile_probability): Set max_probability
8900 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
8902 * profile-count.c (profile_probability::differs_from_p): Do not
8903 rely on max_probaiblity == 10000
8905 2017-10-12 Jeff Law <law@redhat.com>
8907 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
8910 2017-10-12 Martin Sebor <msebor@redhat.com>
8914 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
8915 (handle_alias_pairs): Call it.
8916 * common.opt (-Wattribute-alias): New option.
8917 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
8918 * doc/invoke.texi (-Wattribute-alias): Document.
8920 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
8923 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
8925 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8927 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8928 (make_hard_regno_born, make_hard_regno_dead): Update
8929 bb_killed_pseudos and bb_gen_pseudos.
8931 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
8933 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
8935 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
8937 * config/alpha/alpha.c (alpha_split_conditional_move):
8938 Use std::swap instead of manually swapping.
8939 (alpha_stdarg_optimize_hook): Ditto.
8940 (alpha_canonicalize_comparison): Ditto.
8942 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8944 * tree-loop-distribution.c (struct builtin_info): New struct.
8945 (struct partition): Refactor fields into struct builtin_info.
8946 (partition_free): Free struct builtin_info.
8947 (build_size_arg_loc, build_addr_arg_loc): Delete.
8948 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
8949 information from struct builtin_info.
8950 (find_single_drs): New function refactored from classify_partition.
8951 Also moved builtin validity checks to this function.
8952 (compute_access_range, alloc_builtin): New functions.
8953 (classify_builtin_st, classify_builtin_ldst): New functions.
8954 (classify_partition): Refactor code into functions find_single_drs,
8955 classify_builtin_st and classify_builtin_ldst.
8956 (distribute_loop): Don't do runtime alias check when distributing
8958 (find_seed_stmts_for_distribution): New function.
8959 (pass_loop_distribution::execute): Refactor code finding seed
8960 stmts into above function. Support distribution for the innermost
8961 two-level loop nest. Adjust dump information.
8963 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8965 * tree-loop-distribution.c: Adjust the general comment.
8966 (NUM_PARTITION_THRESHOLD): New macro.
8967 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
8968 (classify_partition): Skip builtin pattern of loop nest's inner loop.
8969 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
8970 in call to build_partition_graph.
8971 (finalize_partitions): New parameter. Make loop distribution more
8972 conservative by fusing more partitions.
8973 (distribute_loop): Don't do runtime alias check in case of loop nest
8975 (find_seed_stmts_for_distribution): New function.
8976 (prepare_perfect_loop_nest): New function.
8977 (pass_loop_distribution::execute): Refactor code finding seed stmts
8978 and loop nest into above functions. Support loop nest distribution.
8979 Adjust dump information accordingly.
8981 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8983 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
8984 and set PTYPE_SEQUENTIAL for merged partition.
8986 2017-10-12 Richard Biener <rguenther@suse.de>
8988 PR tree-optimization/69728
8990 2017-09-19 Richard Biener <rguenther@suse.de>
8992 PR tree-optimization/69728
8993 * graphite-sese-to-poly.c (schedule_error): New global.
8994 (add_loop_schedule): Handle empty domain by failing the
8996 (build_original_schedule): Handle schedule_error.
8998 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
8999 domain by returning an unchanged schedule.
9001 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9003 * genrecog.c (validate_pattern): For VEC_SELECT verify that
9004 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
9006 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
9008 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
9011 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9014 * expr.c (store_field): Don't optimize away bitsize == 0 store
9015 from CALL_EXPR with addressable return type.
9017 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9019 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
9020 * config/rs6000/rs6000.md (sel): Delete mode attribute.
9021 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
9022 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
9023 TARGET_ISEL instead of TARGET_ISEL<sel>.
9025 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
9027 * config/rs6000/rs6000.c
9028 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
9030 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9032 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
9034 (reg_or_cint_operand): Fix comment.
9035 (reg_or_zero_operand): New predicate.
9036 * config/rs6000/rs6000-protos.h (output_isel): Delete.
9037 * config/rs6000/rs6000.c (output_isel): Delete.
9038 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
9039 instead of reg_or_cint_operand. Output instruction directly (not via
9041 (isel_unsigned_<mode>): Ditto.
9042 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
9043 gpc_reg_operand. Add an instruction alternative for this. Output
9044 instruction directly.
9045 (*isel_reversed_unsigned_<mode>): Ditto.
9047 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9049 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
9050 (TARGET_CANONICALIZE_COMPARISON): Define.
9052 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
9055 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
9056 Check whether the dest is REG before adding REG_EQUIV note.
9058 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9061 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9063 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9064 (make_hard_regno_born, make_hard_regno_dead): Update
9065 bb_killed_pseudos and bb_gen_pseudos.
9067 2017-10-11 Nathan Sidwell <nathan@acm.org>
9069 * incpath.h (enum incpath_kind): Name enum, prefix values.
9070 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
9071 * incpath.c (heads, tails): Use INC_MAX.
9072 (add_env_var_paths, add_standard_paths): Use incpath_kind.
9073 (merge_include_chains, split_quote_chain,
9074 register_include_chains): Update incpath_kind names.
9075 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
9076 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
9078 (add_framework_path, darwin_register_objc_includes): Likewise.
9079 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
9081 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9083 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
9084 Do not use float_operator operator predicate.
9085 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
9086 * config/i386/predicates.md (float_operator): Remove predicate.
9088 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9090 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
9091 (*jccxf_i387): Ditto.
9092 (*jcc<mode>_i387): Ditto.
9093 (*jccu<mode>_i387): Ditto.
9094 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
9095 (*jcc_*_i387 splitters): Remove.
9096 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
9097 * config/i386/i386.c (ix86_split_fp_branch): Remove.
9098 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
9101 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9103 * profile-count.h (slow_safe_scale_64bit): New function.
9104 (safe_scale_64bit): New inline.
9105 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
9106 * profile-count.c: Include wide-int.h
9107 (slow_safe_scale_64bit): New.
9109 2017-10-11 Nathan Sidwell <nathan@acm.org>
9111 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
9112 HAS_DECL_ASSEMBLER_NAME_P.
9113 * gimple-expr.c (gimple_decl_printable_name: Check
9114 HAS_DECL_ASSEMBLER_NAME_P too.
9115 * ipa-utils.h (type_in_anonymous_namespace_p): Check
9116 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
9117 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
9118 * passes.c (rest_of_decl_compilation): Check
9119 HAS_DECL_ASSEMBLER_NAME_P too.
9120 * recog.c (verify_changes): Likewise.
9121 * tree-pretty-print.c (dump_decl_name): Likewise.
9122 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
9124 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
9125 (SET_DECL_ASSEMBLER_NAME): Use it.
9126 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
9127 (COPY_DECL_ASSEMBLER_NAME): Likewise.
9128 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
9130 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9132 * config.gcc (i386, x86_64): Add extra objects.
9133 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
9134 (ix86_min_insn_size): Declare.
9135 (ix86_issue_rate): Declare.
9136 (ix86_adjust_cost): Declare.
9137 (ia32_multipass_dfa_lookahead): Declare.
9138 (ix86_macro_fusion_p): Declare.
9139 (ix86_macro_fusion_pair_p): Declare.
9140 (ix86_bd_has_dispatch): Declare.
9141 (ix86_bd_do_dispatch): Declare.
9142 (ix86_core2i7_init_hooks): Declare.
9143 (ix86_atom_sched_reorder): Declare.
9144 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
9145 (COSTS_N_BYTES): Move to x86-tune-costs.h.
9146 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
9147 (rip_relative_addr_p): Rename to ...
9148 (ix86_rip_relative_addr_p): ... this one; export.
9149 (memory_address_length): Update.
9150 (ix86_issue_rate): Move to x86-tune-sched.c.
9151 (ix86_flags_dependent): Move to x86-tune-sched.c.
9152 (ix86_agi_dependent): Move to x86-tune-sched.c.
9153 (exact_dependency_1): Move to x86-tune-sched.c.
9154 (exact_store_load_dependency): Move to x86-tune-sched.c.
9155 (ix86_adjust_cost): Move to x86-tune-sched.c.
9156 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
9157 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
9158 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
9159 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
9160 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
9161 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
9162 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
9163 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
9164 (min_insn_size): Rename to ...
9165 (ix86_min_insn_size): ... this one; export.
9166 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
9167 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
9168 (core2i7_first_cycle_multipass_backtrack): Move to
9169 x86-tune-sched-core.c.
9170 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
9171 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
9172 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
9173 (ix86_avoid_jump_mispredicts): Update.
9174 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
9175 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
9176 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
9177 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
9178 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
9179 (MAX_INSN): Move to ix86-tune-sched-bd.c.
9180 (MAX_IMM): Move to ix86-tune-sched-bd.c.
9181 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
9182 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
9183 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
9184 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
9185 (MAX_STORE): Move to ix86-tune-sched-bd.c.
9186 (BIG): Move to ix86-tune-sched-bd.c.
9187 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
9188 (enum insn_path): Move to ix86-tune-sched-bd.c.
9189 (get_mem_group): Move to ix86-tune-sched-bd.c.
9190 (is_cmp): Move to ix86-tune-sched-bd.c.
9191 (dispatch_violation): Move to ix86-tune-sched-bd.c.
9192 (is_branch): Move to ix86-tune-sched-bd.c.
9193 (is_prefetch): Move to ix86-tune-sched-bd.c.
9194 (init_window): Move to ix86-tune-sched-bd.c.
9195 (allocate_window): Move to ix86-tune-sched-bd.c.
9196 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
9197 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
9198 (process_end_window): Move to ix86-tune-sched-bd.c.
9199 (allocate_next_window): Move to ix86-tune-sched-bd.c.
9200 (find_constant): Move to ix86-tune-sched-bd.c.
9201 (get_num_immediates): Move to ix86-tune-sched-bd.c.
9202 (has_immediate): Move to ix86-tune-sched-bd.c.
9203 (get_insn_path): Move to ix86-tune-sched-bd.c.
9204 (get_insn_group): Move to ix86-tune-sched-bd.c.
9205 (count_num_restricted): Move to ix86-tune-sched-bd.c.
9206 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
9207 (add_insn_window): Move to ix86-tune-sched-bd.c.
9208 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
9209 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
9210 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
9211 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
9212 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
9213 (do_dispatch): Move to ix86-tune-sched-bd.c.
9214 (has_dispatch): Move to ix86-tune-sched-bd.c.
9215 * config/i386/t-i386: Add new object files.
9216 * config/i386/x86-tune-costs.h: New file.
9217 * config/i386/x86-tune-sched-atom.c: New file.
9218 * config/i386/x86-tune-sched-bd.c: New file.
9219 * config/i386/x86-tune-sched-core.c: New file.
9220 * config/i386/x86-tune-sched.c: New file.
9222 2017-10-11 Liu Hao <lh_mouse@126.com>
9224 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
9225 the generic version below instead.
9226 (should_colorize): Recognize Windows consoles as terminals
9228 * pretty-print.c [__MINGW32__] (write_all): New function.
9229 [__MINGW32__] (find_esc_head): Likewise.
9230 [__MINGW32__] (find_esc_terminator): Likewise.
9231 [__MINGW32__] (eat_esc_sequence): Likewise.
9232 [__MINGW32__] (mingw_ansi_fputs): New function that handles
9234 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
9237 2017-10-11 Richard Biener <rguenther@suse.de>
9239 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
9240 Properly call analyze_scalar_evolution with the loop of the stmt.
9242 2017-10-11 Richard Biener <rguenther@suse.de>
9244 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
9245 * tree-core.h (tree_base): Add chrec_var union member.
9246 * tree.h (CHREC_VAR): Remove.
9247 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
9248 * tree-chrec.h (build_polynomial_chrec): Adjust.
9249 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
9250 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
9252 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
9254 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
9255 * match.pd: ... here.
9256 ((T) X == (T) Y): Relax condition.
9258 2017-10-11 Bin Cheng <bin.cheng@arm.com>
9260 PR tree-optimization/82472
9261 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
9263 (break_alias_scc_partitions): Update postorder number.
9265 2017-10-11 Martin Liska <mliska@suse.cz>
9268 * opts.c (parse_no_sanitize_attribute): Do not use error_value
9270 * opts.h (parse_no_sanitize_attribute): Remove last argument.
9272 2017-10-11 Martin Liska <mliska@suse.cz>
9274 * print-rtl.c (print_insn): Move declaration of idbuf
9275 to same scope as name.
9277 2017-10-11 Martin Liska <mliska@suse.cz>
9282 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9285 2017-10-11 Martin Liska <mliska@suse.cz>
9288 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9291 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
9293 * config.gcc (default_gnu_indirect_function): Default to yes for
9294 arm*-*-linux* with glibc.
9296 2017-10-11 Richard Biener <rguenther@suse.de>
9298 * tree-scalar-evolution.c (get_scalar_evolution): Handle
9299 default-defs and types we do not want to analyze.
9300 (interpret_loop_phi): Replace unreachable code with an assert.
9301 (compute_scalar_evolution_in_loop): Remove and inline ...
9302 (analyze_scalar_evolution_1): ... here, replacing condition with
9303 what makes the intent clearer. Remove handling of cases
9304 get_scalar_evolution now handles.
9306 2017-10-10 Jim Wilson <wilson@tuliptree.org>
9308 PR rtl-optimization/81434
9309 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
9310 comment for main loop. In sched_group_found if, also add checks for
9311 pass and min_cost_group.
9313 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
9315 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
9316 (rs6000_insn_cost): New function.
9317 * config/rs6000/rs6000.md (cost): New attribute.
9319 2017-10-10 Jakub Jelinek <jakub@redhat.com>
9320 H.J. Lu <hongjiu.lu@intel.com>
9324 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
9325 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
9326 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
9327 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
9328 to OPTION_MASK_ISA_AVX512VL - builtins that have both
9329 OPTION_MASK_ISA_MMX and some other bit set require both
9330 mmx and the ISAs without the mmx bit.
9331 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
9332 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
9333 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
9334 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
9335 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
9336 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
9337 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
9338 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
9339 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
9340 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
9341 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
9342 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
9343 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
9344 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
9345 Add OPTION_MASK_ISA_MMX.
9347 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
9349 * config.gcc (armv7*-*-freebsd*): New target.
9350 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
9352 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
9354 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
9355 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
9358 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
9360 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
9361 a template parameter.
9362 (WIDE_INT_REF_FOR): Update accordingly.
9363 * tree.h (wi::int_traits <const_tree>): Delete.
9364 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
9365 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
9366 (wi::tree_to_wide_ref): New typedef.
9367 (wi::to_wide): New function.
9368 * calls.c (get_size_range): Use wi::to_wide when operating on
9370 * cgraph.c (cgraph_node::create_thunk): Likewise.
9371 * config/i386/i386.c (ix86_data_alignment): Likewise.
9372 (ix86_local_alignment): Likewise.
9373 * dbxout.c (stabstr_O): Likewise.
9374 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
9375 * expr.c (const_vector_from_tree): Likewise.
9376 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
9377 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
9378 (fold_negate_expr_1, int_const_binop_1, const_binop)
9379 (fold_convert_const_int_from_real, optimize_bit_field_compare)
9380 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
9381 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
9382 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
9383 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
9384 (fold_not_const, round_up_loc): Likewise.
9385 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
9386 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
9387 (alloca_call_type): Likewise.
9388 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
9389 * godump.c (go_output_typedef): Likewise.
9390 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
9391 * internal-fn.c (get_min_precision): Likewise.
9392 * ipa-cp.c (ipcp_store_vr_results): Likewise.
9393 * ipa-polymorphic-call.c
9394 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
9395 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
9396 (ipa_modify_call_arguments): Likewise.
9397 * match.pd: Likewise.
9398 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
9399 * print-tree.c (print_node_brief, print_node): Likewise.
9400 * stmt.c (expand_case): Likewise.
9401 * stor-layout.c (layout_type): Likewise.
9402 * tree-affine.c (tree_to_aff_combination): Likewise.
9403 * tree-cfg.c (group_case_labels_stmt): Likewise.
9404 * tree-data-ref.c (dr_analyze_indices): Likewise.
9405 (prune_runtime_alias_test_list): Likewise.
9406 * tree-dump.c (dequeue_and_dump): Likewise.
9407 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
9408 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
9409 * tree-pretty-print.c (dump_generic_node): Likewise.
9410 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
9411 (simple_iv_with_niters): Likewise.
9412 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
9413 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
9414 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
9415 * tree-ssa-loop-niter.c (split_to_var_and_offset)
9416 (refine_value_range_using_guard, number_of_iterations_ne_max)
9417 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
9418 (get_cst_init_from_scev, record_nonwrapping_iv)
9419 (scev_var_range_cant_overflow): Likewise.
9420 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
9421 * tree-ssa-pre.c (compute_avail): Likewise.
9422 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
9423 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
9424 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
9425 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
9426 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
9427 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
9428 (dump_case_nodes, try_switch_expansion): Likewise.
9429 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
9430 (vect_do_peeling): Likewise.
9431 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
9432 * tree-vect-stmts.c (vectorizable_load): Likewise.
9433 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
9434 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
9435 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
9436 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
9437 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
9438 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
9439 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
9440 (evrp_dom_walker::before_dom_children): Likewise.
9441 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
9442 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
9443 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
9444 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
9445 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
9446 (get_range_pos_neg): Likewise.
9447 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
9448 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
9449 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
9450 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
9451 * config/avr/avr.c (avr_fold_builtin): Likewise.
9452 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
9453 * config/msp430/msp430.c (msp430_attr): Likewise.
9454 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
9455 * config/powerpcspe/powerpcspe-c.c
9456 (altivec_resolve_overloaded_builtin): Likewise.
9457 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
9458 (rs6000_expand_ternop_builtin): Likewise.
9459 * config/rs6000/rs6000-c.c
9460 (altivec_resolve_overloaded_builtin): Likewise.
9461 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
9462 (rs6000_expand_ternop_builtin): Likewise.
9463 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
9465 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9467 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
9468 when copying loop nest with only one inner loop.
9470 2017-10-10 Richard Biener <rguenther@suse.de>
9472 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
9473 blocks if SCEV is active.
9474 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
9476 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
9477 (scev_initialize): Assert we are not yet initialized.
9479 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9481 * tree-loop-distribution.c (generate_loops_for_partition): Remove
9482 inner loop's exit stmt by making it always exit the loop, otherwise
9483 we would generate an infinite empty loop.
9485 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9487 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
9488 renaming variables in new preheader if it's deleted.
9490 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9492 * tree-loop-distribution.c (struct partition): Remove unused field
9493 loops of the structure.
9494 (partition_alloc, partition_free): Ditto.
9495 (build_rdg_partition_for_vertex): Ditto.
9497 2017-10-09 Jeff Law <law@redhat.com>
9499 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
9500 return type to match prototype and documentation.
9502 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9504 * config/rs6000/rs6000.c (processor_costs): Move to ...
9505 * config/rs6000/rs6000.h: ... here.
9506 (rs6000_cost): Declare.
9508 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
9510 * except.c (setjmp_fn): New global variable.
9511 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
9512 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
9513 if DONT_USE_BUILTIN_SETJMP is defined.
9515 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9517 * target.def (insn_cost): New hook.
9518 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
9519 * doc/tm.texi: Regenerate.
9520 * rtlanal.c (insn_cost): Use the new hook.
9522 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9524 * combine.c (combine_validate_cost): Compute the new insn_cost,
9525 not just pattern_cost.
9526 (try_combine): Adjust comment.
9528 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9530 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
9532 * combine.c (uid_insn_cost): Adjust comment.
9533 (combine_validate_cost): Adjust comment. Use pattern_cost instead
9535 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
9536 * dse.c (find_shift_sequence): Ditto.
9537 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
9538 (bb_valid_for_noce_process_p): Use pattern_cost.
9539 * rtl.h (insn_rtx_cost): Delete.
9540 (pattern_cost): New prototype.
9541 (insn_cost): New prototype.
9542 * rtlanal.c (insn_rtx_cost): Rename to...
9543 (pattern_cost): ... this.
9546 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
9548 * config/i386/i386.md (*jcc_2): Remove insn pattern.
9549 (*jcc<mode>_0_r_i387): Ditto.
9550 (*jccxf_r_i387): Ditto.
9551 (*jcc<mode>_r_i387): Ditto.
9552 (*jccu<mode>_r_i387): Ditto.
9553 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
9554 (*jcc): Rename from *jcc_1.
9556 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9558 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
9559 deferred rescans after the lvx/stvx recombination pre-pass.
9561 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
9563 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
9564 memory operation instruction support.
9565 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
9566 (rs6000-ibm-aix[789]*): Likewise.
9567 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
9568 Document new functions.
9570 2017-10-09 Richard Biener <rguenther@suse.de>
9572 PR tree-optimization/82397
9573 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
9574 equality only for semantically equal trees.
9576 2017-10-09 Richard Biener <rguenther@suse.de>
9578 PR tree-optimization/82449
9579 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
9580 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
9581 allow constant addresses.
9582 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
9585 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9587 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
9590 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9593 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
9596 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9599 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
9601 2017-10-09 Jakub Jelinek <jakub@redhat.com>
9604 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
9605 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
9607 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
9609 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
9610 (WI_BINARY_PREDICATE_RESULT): Likewise.
9611 (wi::binary_traits::operator_result): New type.
9612 (wi::binary_traits::predicate_result): Likewise.
9613 (generic_wide_int::operator~, unary generic_wide_int::operator-)
9614 (generic_wide_int::operator==, generic_wide_int::operator!=)
9615 (generic_wide_int::operator&, generic_wide_int::and_not)
9616 (generic_wide_int::operator|, generic_wide_int::or_not)
9617 (generic_wide_int::operator^, generic_wide_int::operator+
9618 (binary generic_wide_int::operator-, generic_wide_int::operator*):
9620 (operator~, unary operator-, operator==, operator!=, operator&)
9621 (operator|, operator^, operator+, binary operator-, operator*): New
9623 * expr.c (get_inner_reference): Use wi::bit_and_not.
9624 * fold-const.c (fold_binary_loc): Likewise.
9625 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
9626 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
9627 (bit_value_binop): Likewise.
9628 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
9629 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
9630 (extract_range_from_binary_expr_1): Likewise.
9631 (masked_increment): Likewise.
9632 (simplify_bit_ops_using_ranges): Likewise.
9634 2017-10-09 Martin Jambor <mjambor@suse.cz>
9637 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
9638 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
9639 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
9641 (hsa_fixup_mov_insn_type): New function.
9642 (hsa_op_with_type::get_in_type): Use it.
9643 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
9644 immediates in an assert.
9645 (hsa_op_with_type::extend_int_to_32bit): New method.
9646 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
9647 types. Convert to dest type if necessary.
9648 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
9649 (reg_for_gimple_ssa): Pass false as min32int to
9650 hsa_type_for_scalar_tree_type.
9651 (gen_hsa_addr): Fixup type when creating addresable temporary.
9652 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
9653 (gen_hsa_unary_operation): Extend operands and convert to dest type if
9654 necessary. Call hsa_fixup_mov_insn_type.
9655 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
9656 extend operands and convert to dest type if necessary.
9657 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
9658 to dest type if necessary.
9659 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
9660 if conversion nt necessary and size matches.
9661 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
9662 to dest type if necessary.
9663 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
9664 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
9666 (gen_hsa_clrsb): Likewise.
9667 (gen_hsa_ffs): Likewise.
9668 (gen_hsa_divmod): Extend operands and convert to dest type if
9670 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
9672 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
9674 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
9675 Remove empty default arguments. Use a brace block as output
9677 (conditional return): Ditto.
9679 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
9680 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
9682 (group_ending_nop): Ditto.
9683 (doloop_end): Ditto.
9684 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
9685 (splitters for those): Ditto.
9687 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
9689 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
9690 a conditional jump (and the compare for it) so that pc_rtx is the
9692 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
9693 for the deleted and renamed ctr<mode>_internal[234] patterns.
9694 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
9695 Delete second conditional return pattern.
9696 (ctr<mode>_internal2): Delete this second bdnz pattern.
9697 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
9698 (ctr<mode>_internal4): Delete this second bdz pattern.
9700 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9702 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
9703 (always_initialized_rtx_for_ssa_name_p): New predicate.
9704 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
9705 (finish_out_of_ssa): Free new field of SA.
9706 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
9707 * tree-ssa-coalesce.c: Include tree-ssa.h.
9708 (get_parm_default_def_partitions): Remove extern keyword.
9709 (get_undefined_value_partitions): New function.
9710 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
9711 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
9713 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
9715 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9717 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
9719 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9721 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
9722 for targets that preffer 128bit.
9724 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9726 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
9728 2017-10-08 Olivier Hainque <hainque@adacore.com>
9730 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
9731 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
9733 (thumb_set_return_address): Likewise.
9735 2017-10-08 Olivier Hainque <hainque@adacore.com>
9737 * common/config/arm/arm-common.c (arm_except_unwind_info):
9738 Handle DWARF2_UNWIND_INFO.
9740 2017-10-07 Michael Collison <michael.collison@arm.com>
9742 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
9745 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
9747 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
9748 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
9749 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
9750 defined, force the creation of a new block for a dispatch label.
9752 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
9754 * invoke.texi (Wsuggest-attribute=cold): Document.
9755 * common.opt (Wsuggest-attribute=cold): New
9756 * ipa-pure-const.c (warn_function_cold): New function.
9757 * predict.c (compute_function_frequency): Use it.
9758 * predict.h (warn_function_cold): Declare.
9760 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
9762 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
9765 2017-10-06 Martin Liska <mliska@suse.cz>
9767 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
9768 keyword for member functions.
9769 (struct sanopt_tree_couple): New struct.
9770 (struct sanopt_tree_couple_hash): New function.
9771 (struct sanopt_ctx): Add new hash_map.
9772 (has_dominating_ubsan_ptr_check): New function.
9773 (record_ubsan_ptr_check_stmt): Likewise.
9774 (maybe_optimize_ubsan_ptr_ifn): Likewise.
9775 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
9776 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
9778 2017-10-06 Sudakshina Das <sudi.das@arm.com>
9781 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
9782 aarch64_simd_valid_immediate on CONST_VECTORs.
9783 (aarch64_reg_or_bic_imm): Likewise.
9785 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
9787 PR rtl-optimization/82396
9788 * haifa-sched.c (ready_sort_real): Disable qsort checking.
9790 2017-10-06 Sebastian Pop <sebpop@gmail.com>
9792 * graphite-dependences.c (scop_get_reads): Move code to...
9793 (scop_get_must_writes): Move code to...
9794 (scop_get_may_writes): Move code to...
9795 (scop_get_reads_and_writes): ... here.
9796 (scop_get_dependences): Call scop_get_reads_and_writes.
9798 2017-10-06 Jakub Jelinek <jakub@redhat.com>
9800 PR tree-optimization/82434
9801 * fold-const.h (can_native_encode_type_p,
9802 can_native_encode_string_p): Remove.
9803 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
9804 don't encode anything, just return what would be otherwise returned.
9805 (native_encode_fixed, native_encode_complex, native_encode_vector):
9807 (native_encode_string): Likewise. Inline by hand
9808 can_native_encode_string_p.
9809 (can_native_encode_type_p): Remove.
9810 (can_native_encode_string_p): Remove.
9811 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
9812 STRING_CSTs using can_native_encode_string_p, test all
9813 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
9814 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
9815 argument from native_encode_expr.
9816 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
9817 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
9820 2017-10-06 Richard Biener <rguenther@suse.de>
9822 PR tree-optimization/82397
9823 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
9824 operand_equal_p but rely on data_ref_compare_tree for detecting
9826 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
9827 to match up with dr_group_sort_cmp.
9829 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9832 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
9834 * config/s390/s390-builtin-types.def: Regenerate.
9836 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9839 * config/s390/s390-builtin-types.def: Regenerate.
9840 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
9841 Change flag from B_VXE to B_VX.
9842 (s390_vec_min_dbl): Remove B_VXE flag.
9844 2017-10-06 Richard Biener <rguenther@suse.de>
9846 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
9847 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
9848 translate_isl_ast_to_gimple::is_valid_rename,
9849 translate_isl_ast_to_gimple::get_rename,
9850 translate_isl_ast_to_gimple::get_def_bb_for_const,
9851 translate_isl_ast_to_gimple::get_new_name,
9852 translate_isl_ast_to_gimple::collect_all_ssa_names,
9853 translate_isl_ast_to_gimple::copy_loop_phi_args,
9854 translate_isl_ast_to_gimple::collect_all_ssa_names,
9855 translate_isl_ast_to_gimple::copy_loop_phi_args,
9856 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
9857 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
9858 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
9859 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
9860 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
9861 translate_isl_ast_to_gimple::copy_cond_phi_args,
9862 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
9863 translate_isl_ast_to_gimple::edge_for_new_close_phis,
9864 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
9865 translate_isl_ast_to_gimple::rename_uses,
9866 translate_isl_ast_to_gimple::rename_all_uses): Remove.
9867 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
9868 (set_rename_for_each_def): Likewise.
9869 (graphite_copy_stmts_from_block): Handle debug stmt resetting
9870 here. Handle rewriting SCEV analyzable uses here.
9871 (copy_bb_and_scalar_dependences): Generate code for PHI
9873 (graphite_regenerate_ast_isl): Adjust.
9874 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
9875 (add_write, add_read): New functions.
9876 (build_cross_bb_scalars_def): Use it and simplify.
9877 (build_cross_bb_scalars_use): Likewise.
9878 (graphite_find_cross_bb_scalar_vars): Inline into...
9879 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
9880 simulating out-of-SSA. Compute liveout and add dependencies.
9881 (build_scops): Force an empty entry block.
9882 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
9884 (sese_build_liveouts): Declare.
9885 (sese_trivially_empty_bb_p): Likewise.
9886 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
9887 compute liveout and debug_liveout.
9888 (sese_bad_liveouts_use): Remove.
9889 (sese_reset_debug_liveouts_bb): Likewise.
9890 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
9891 (sese_build_liveouts): Build liveout and debug_liveout and store
9893 (new_sese_info): Adjust.
9894 (free_sese_info): Likewise.
9895 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
9896 do not build liveout here.
9897 (move_sese_in_condition): Adjust region entry.
9898 (scev_analyzable_p): Match up with chrec_apply requirements.
9899 (sese_trivially_empty_bb_p): New.
9900 * tree-into-ssa.c (get_reaching_def): Properly support generating
9901 default-defs for incremental rewrite of anonymous names.
9903 2017-10-06 Richard Biener <rguenther@suse.de>
9905 * graphite-sese-to-poly.c (extract_affine): For casts increasing
9906 precision do not perform modulo reduction.
9908 2017-10-06 Richard Biener <rguenther@suse.de>
9910 PR tree-optimization/82436
9911 * tree-vect-slp.c (vect_supported_load_permutation_p): More
9912 conservatively choose the vectorization factor when checking
9913 whether we can perform the required load permutation.
9914 (vect_transform_slp_perm_load): Assert when we may not fail.
9916 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
9918 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
9919 message for incompatible -msdata=* and -mcall-* options.
9921 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9923 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
9924 rate for post-reload scheduling.
9926 2017-10-05 Tamar Christina <tamar.christina@arm.com>
9928 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
9930 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9932 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
9933 to improve monte carlo in scimark.
9935 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9937 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
9938 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
9939 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
9940 pentium4_cost, nocona_cost): Set reassociation width to 1.
9941 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
9942 width to 2 for fp operations and 1 otherwise.
9943 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
9945 (atom_cost): Set reassociation width to 2.
9946 (slm_cost, generic_cost): Set fp reassociation width
9947 to 2 and 1 otherwise.
9948 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
9949 (core_cost): Set fp reassociation width to 4 and vector to 2.
9950 (ix86_reassociation_width): Rewrite using cost table; special case
9951 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
9952 and TARGET_AVX128_OPTIMAL.
9953 * config/i386/i386.h (processor_costs): Add
9954 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
9955 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
9956 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
9957 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
9958 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
9959 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
9961 2017-10-05 Nathan Sidwell <nathan@acm.org>
9963 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
9965 2017-10-05 Tamar Christina <tamar.christina@arm.com>
9967 * config/arm/arm.c (arm_test_fpu_data): New.
9968 (arm_run_selftests): Call arm_test_fpu_data.
9970 2017-10-04 Nathan Sidwell <nathan@acm.org>
9972 * toplev.c (toplev::main): Remove excess parens on pretty_printer
9974 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
9976 2017-10-04 Sudakshina Das <sudi.das@arm.com>
9978 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
9979 check type for aarch64_simd_valid_immediate.
9980 (aarch64_output_simd_mov_immediate): Update prototype.
9981 (aarch64_simd_valid_immediate): Update prototype.
9982 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
9983 support for ORR-immediate.
9984 (and<mode>3): modified pattern to add support for BIC-immediate.
9985 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
9986 now checks for valid immediate for BIC and ORR based on new enum
9988 (aarch64_output_simd_mov_immediate): Function now used to output
9989 BIC/ORR imm as well based on new enum argument.
9990 * config/aarch64/constraints.md (Do): New vector immediate constraint.
9992 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
9993 (aarch64_reg_or_bic_imm): Likewise.
9995 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9997 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
9998 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
10000 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10004 PR rtl-optimization/82396
10005 * haifa-sched.c (autopref_multipass_init): Simplify
10007 (autopref_rank_data): Simplify sort order.
10008 * sched-int.h (autopref_multipass_data_): Remove
10009 multi_mem_insn_p, min_offset and max_offset.
10011 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10013 * doc/sourcebuild.texi: Document vect_peeling_profitable.
10015 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10017 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
10018 vect_doubleint_cvt.
10020 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10022 * doc/sourcebuild.texi: Document vect_long_mult.
10024 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
10026 PR tree-optimization/82413
10027 * fold-const.c (build_range_check): Use widest_int when comparing
10028 the maximum ETYPE value with HIGH.
10030 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10032 PR rtl-optimization/82396
10033 * haifa-sched.c (autopref_multipass_init): Simplify
10035 (autopref_rank_data): Simplify sort order.
10036 * sched-int.h (autopref_multipass_data_): Remove
10037 multi_mem_insn_p, min_offset and max_offset.
10039 2017-10-04 Jakub Jelinek <jakub@redhat.com>
10041 PR tree-optimization/82381
10042 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
10043 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
10046 PR tree-optimization/82374
10047 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
10048 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
10049 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
10050 current_function_decl to the new decl.
10052 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
10054 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
10055 helper macro for IEEE float128 hardware built-in functions.
10056 (SQRTF128_ODD): Add built-in functions with the round-to-odd
10058 (TRUNCF128_ODD): Likewise.
10059 (ADDF128_ODD): Likewise.
10060 (SUBF128_ODD): Likewise.
10061 (MULF128_ODD): Likewise.
10062 (DIVF128_ODD): Likewise.
10063 (FMAF128_ODD): Likewise.
10064 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
10065 UNSPEC_TRUNC_ROUND_TO_ODD.
10066 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
10067 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
10068 floating point round to odd instructions.
10069 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
10070 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
10071 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
10072 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
10073 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
10074 (trunc<mode>sf2_hw): Change the truncate with round to odd
10075 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
10076 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
10077 to odd hardware instructions.
10078 (sub<mode>3_odd): Likewise.
10079 (mul<mode>3_odd): Likewise.
10080 (div<mode>3_odd): Likewise.
10081 (sqrt<mode>2_odd): Likewise.
10082 (fma<mode>4_odd): Likewise.
10083 (fms<mode>4_odd): Likewise.
10084 (nfma<mode>4_odd): Likewise.
10085 (nfms<mode>4_odd): Likewise.
10086 (trunc<mode>df2_odd): Change the truncate with round to odd
10087 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
10089 * doc/extend.texi (PowerPC built-in functions): Update documentation
10090 for existing IEEE float128-bit built-in functions. Add built-in
10091 functions that generate the IEEE 128-bit floating point round to
10094 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
10096 PR rtl-optimization/77729
10097 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
10098 to (X&(C1&~C2))|C2 transformations.
10100 2017-10-03 Martin Jambor <mjambor@suse.cz>
10102 PR tree-optimization/82363
10103 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
10104 mismatch, mark lacc written regardless of racc.
10106 2017-10-03 Jakub Jelinek <jakub@redhat.com>
10108 PR tree-optimization/82381
10109 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
10110 stmt_to_insert nor wheather SSA_NAMEs are default defs.
10111 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
10112 fallthrough into reassoc_stmt_dominates_stmt_p.
10115 * combine.c (combine_instructions): Don't combine in unreachable
10118 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
10121 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
10122 function to not use the have_cpu variable. Do not set cpu_index,
10123 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
10124 or the default cpu.
10125 (rs6000_valid_attribute_p): Remove duplicate initializations of
10126 old_optimize and func_optimize.
10127 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
10128 (rs6000_activate_target_options): Make global.
10129 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
10132 2017-10-02 Jakub Jelinek <jakub@redhat.com>
10134 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
10135 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
10136 Set *poffset to 0 and *psize and *pmax_size to -1 if
10137 *poffset + *psize overflows in HOST_WIDE_INT.
10139 PR tree-optimization/82387
10140 PR tree-optimization/82388
10141 PR tree-optimization/82389
10142 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
10143 instead of live_bytes non-NULL.
10145 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
10148 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
10151 2017-10-02 Richard Biener <rguenther@suse.de>
10153 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
10154 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
10155 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
10157 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10159 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
10160 requested precision matches the type's.
10161 * calls.c (alloc_max_size): Calculate the new candidate size as
10162 a widest_int and use wi::to_widest when comparing it with the
10163 current candidate size.
10164 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
10165 zero rather than integer_zero_node.
10166 * match.pd: Check for a no-op conversion before using wi::add
10167 rather than after. Use tree_to_uhwi when summing small shift
10168 counts into an unsigned int.
10170 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10171 Alan Hayward <alan.hayward@arm.com>
10172 David Sherwood <david.sherwood@arm.com>
10175 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
10176 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
10177 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
10178 POINTER_AND_FP_REGS.
10180 2017-10-02 Richard Biener <rguenther@suse.de>
10182 PR tree-optimization/82355
10183 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
10184 a mapping for the enclosing loop but avoid generating one for
10185 the loop tree root.
10186 (copy_bb_and_scalar_dependences): Remove premature codegen
10187 error on PHIs in blocks duplicated into multiple places.
10188 * graphite-scop-detection.c
10189 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
10190 in the region use it as loop and nest to analyze the DR in.
10191 (try_generate_gimple_bb): Likewise.
10192 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
10193 (add_loop_constraints): For blocks in a loop not in the region
10194 create a dimension with a single iteration.
10195 * sese.h (gbb_loop_at_index): Remove assert.
10197 2017-10-01 Kevin Buettner <kevinb@redhat.com>
10199 * omp-expand.c (adjust_context_scope): New function.
10200 (expand_parallel_call): Call adjust_context_scope.
10202 2017-10-01 Jeff Law <law@redhat.com>
10204 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
10205 dom_opt_dom_walker class with direct access to private members.
10206 Add comments. Call test_for_singularity.
10207 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
10208 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
10209 m_dummy_cond anymore.
10210 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
10212 (pass_dominator:execute): Build the dummy_cond here and pass it
10213 to the dom_opt_dom_walker ctor.
10214 (test_for_singularity): New function.
10216 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
10217 Maya Rashish <coypu@sdf.org>
10219 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
10220 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
10221 (alpha*-*-netbsd*) Use nbsd_tm_file.
10222 (arm*-*-netbsdelf*) Likewise.
10223 (i[34567]86-*-netbsdelf*) Likewise.
10224 (x86_64-*-netbsd*) Likewise.
10225 (mips*-*-netbsd*) Likewise.
10226 (powerpc-*-netbsd*) Likewise.
10227 (sh*-*-netbsd*) Likewise.
10228 (sparc-*-netbsdelf*) Likewise.
10229 (sparc64-*-netbsd*) Likewise.
10230 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
10232 (vax-*-netbsdelf*) Likewise.
10233 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
10234 (UINT_FAST8_TYPE) Likewise.
10235 (INT_FAST16_TYPE) Check CHAR_FAST16.
10236 (UINT_FAST16_TYPE) Likewise.
10238 2017-09-30 Jakub Jelinek <jakub@redhat.com>
10241 * config/i386/i386.md
10242 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
10243 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
10244 *divmodsi4_zext_2): New define_insn_and_split.
10245 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
10246 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
10247 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
10248 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
10249 New define_insn_and_split.
10250 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
10251 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
10252 operands[1] having DImode when mode is SImode.
10254 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
10255 always SImode for DIV and MOD in REG_EQUAL notes.
10257 2017-09-29 Yury Gribov <tetra2005@gmail.com>
10259 PR middle-end/82319
10260 * match.pd: Fix handling of NaNs in pattern.
10262 2017-09-29 Jeff Law <law@redhat.com>
10264 * sbitmap.c (bitmap_bit_in_range_p): New function.
10265 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
10266 * tree-ssa-dse.c (live_bytes_read): New function.
10267 (dse_classify_store): Ignore reads of dead bytes.
10269 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
10270 typos and whitespace errors.
10271 * config/i386/predicates.md (address_no_seg_operand): Likewise.
10272 * config/s390/s390.c (s390_emit_prologue): Likewise.
10274 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10277 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
10278 with a symbol for LRA.
10280 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10282 PR rtl-optimization/82338
10283 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
10285 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
10287 * genmodes.c (calc_wider_mode): Suppress qsort macro.
10288 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
10289 (qsort_chk): Declare.
10290 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
10291 (qsort_chk): New function.
10293 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10295 PR tree-optimization/82337
10296 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
10297 phi definition if the PHI result appears in an abnormal PHI.
10298 (find_basis_for_base_expr): Don't record a basis if the LHS of the
10299 basis appears in an abnormal PHI.
10301 2017-09-29 Richard Biener <rguenther@suse.de>
10303 * graphite-isl-ast-to-gimple.c
10304 (translate_isl_ast_to_gimple::set_codegen_error): New function.
10305 (binary_op_to_tree): Use it.
10306 (get_rename_from_scev): Likewise.
10307 (copy_loop_phi_nodes): Likewise.
10308 (copy_bb_and_scalar_dependences): Likewise.
10309 (translate_pending_phi_nodes): Likewise.
10311 2017-09-29 Jakub Jelinek <jakub@redhat.com>
10314 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
10315 for movabsq $(i32 << shift), r64.
10317 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
10319 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
10320 index when encoding %esp as %rsp to avoid 0x67 prefix.
10322 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
10324 * config/i386/i386.md (*movsf_internal, *movdf_internal):
10325 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
10327 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
10329 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
10330 Extensions with more than 16 double VFP registers.
10331 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
10332 to_clear_mask and all code related to it. Replace the remaining
10333 entry by a sbitmap and adapt code accordingly.
10335 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
10337 * brig-builtins.def: Change pure attributes to const.
10339 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10341 * config.gcc (default_gnu_indirect_function): Default to yes for
10342 sparc*-*-linux* with glibc.
10344 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10346 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
10347 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
10348 when creating .init_array and .fini_array sections with priority
10351 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
10354 * config/aarch64/aarch64.c
10355 (aarch64_builtin_support_vector_misalignment): Always return false
10356 when misalignment is unknown.
10358 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
10360 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
10361 this function to return false if the definition used by the swap
10362 instruction is artificial, or if the memory address from which the
10363 constant value is loaded is not represented by a base address held
10364 in a register or if the base address register is a frame or stack
10365 pointer. Additionally, return false if the base address of the
10366 loaded constant is a SYMBOL_REF but is not considered to be a
10368 (replace_swapped_load_constant): New function.
10369 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
10370 loaded constant vector with a load of a swapped constant vector.
10372 2017-09-27 Carl Love <cel@us.ibm.com>
10374 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
10375 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
10376 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
10379 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
10381 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
10382 first, always call autopref_rank_data otherwise.
10384 2017-09-27 Richard Biener <rguenther@suse.de>
10386 * graphite-scop-detection.c (find_scop_parameters): Move
10387 loop bound handling ...
10388 (gather_bbs::before_dom_children): ... here, avoiding the need
10389 to build scop_info->loop_nest.
10390 (record_loop_in_sese): Remove.
10391 * sese.h (sese_info_t::loop_nest): Remove.
10392 * sese.c (new_sese_info): Do not allocate loop_nest.
10393 (free_sese_info): Do not free loop_nest.
10395 2017-09-27 Jakub Jelinek <jakub@redhat.com>
10398 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
10399 lhs from calls if the lhs has addressable type.
10401 2017-09-27 Richard Biener <rguenther@suse.de>
10403 * graphite.h (scop::max_alias_set): New member.
10404 * graphite-scop-detection.c: Remove references to non-existing
10405 --param in comments.
10406 (build_alias_sets): Record the maximum alias set used for drs.
10407 (build_scops): Support zero as unlimited for
10408 --param graphite-max-arrays-per-scop.
10409 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
10410 and inline into ...
10411 (build_poly_sr_1): ... here. Compute alias set based on the
10412 maximum alias set used for drs rather than
10413 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
10415 2017-09-27 Richard Biener <rguenther@suse.de>
10417 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
10418 --param loop-block-tile-size=0 to disable tiling.
10420 2017-09-27 Richard Biener <rguenther@suse.de>
10422 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
10423 (graphite-max-nb-scop-params): Document special value zero.
10424 * domwalk.h (dom_walker::STOP): New symbolical constant.
10425 (dom_walker::dom_walker): Add optional parameter for bb to
10427 (dom_walker::~dom_walker): Declare.
10428 (dom_walker::before_dom_children): Document STOP return value.
10429 (dom_walker::m_user_bb_to_rpo): New member.
10430 (dom_walker::m_bb_to_rpo): Likewise.
10431 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
10432 mapping here if not provided by the user.
10433 (dom_walker::~dom_walker): Free bb to RPO mapping if not
10434 provided by the user.
10435 (dom_walker::STOP): Define.
10436 (dom_walker::walk): Do not compute bb to RPO mapping here.
10437 Support STOP return value from before_dom_children to stop
10439 * graphite-optimize-isl.c (optimize_isl): If the schedule
10440 is the same still generate code if -fgraphite-identity
10441 or -floop-parallelize-all are given.
10442 * graphite-scop-detection.c: Include cfganal.h.
10443 (gather_bbs::gather_bbs): Get and pass through bb to RPO
10445 (gather_bbs::before_dom_children): Return STOP for BBs
10447 (build_scops): Compute bb to RPO mapping and pass it to
10448 the domwalk. Treat --param graphite-max-nb-scop-params=0
10449 as not limiting the number of params.
10450 * graphite.c (graphite_initialize): Remove limit on the
10451 number of basic-blocks in a function.
10452 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
10453 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
10454 default value of 10.
10456 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10458 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
10459 Adjust code to eliminate needing to do the shift right 32-bits
10460 operation after XSCVDPSPN.
10462 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10464 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
10465 ((X / Y) != 0 -> X >= Y): Likewise.
10467 2017-09-26 Carl Love <cel@us.ibm.com>
10469 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
10470 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
10471 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
10472 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
10473 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
10474 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
10475 definitions and overloading.
10476 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
10477 statement for P9V_BUILTIN_XST_LEN_R.
10478 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
10479 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
10480 define_expand and define_insn for the instructions and builtins.
10481 * doc/extend.texi: Update the built-in documentation file for the new
10482 built-in functions.
10483 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
10484 define_insn for the instructions
10486 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
10489 * gcc/config/netbsd-protos.h: New file.
10490 * gcc/config/netbsd.c: New file.
10491 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
10492 * gcc/config/t-netbsd: New file.
10493 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
10494 (tmake_file) Add t-netbsd.
10495 (extra_objs) Add netbsd.o.
10497 2017-09-26 Janus Weil <janus@gcc.gnu.org>
10501 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
10503 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10505 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
10506 sign extension from a vector register to a GPR by doing a 32-bit
10507 direct move and then an EXTSW.
10508 (extendsi<mode>2 splitter): Likewise.
10509 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
10510 right or vector extract after doing XSCVDPSPN. Use
10511 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
10513 (movdi_from_sf_zero_ext): Likewise.
10514 (reload_gpr_from_vsxsf): Likewise.
10515 (p8_mfvsrd_4_disf): Delete, no longer used.
10516 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
10517 then needing to move the SFmode to a GPR to use the XSCVDPSP
10518 instruction instead of FRSP and XSCVDPSPN.
10519 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
10520 it is adjacent to the other XSCVSPDP insns.
10521 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
10522 SFmode to be in traditional Altivec registers.
10523 (vsx_xscvdpspn): Eliminate useless alternative constraint.
10524 (vsx_xscvspdpn): Likewise.
10525 (vsx_xscvspdpn_scalar): Likewise.
10527 2017-09-26 Martin Jambor <mjambor@suse.cz>
10529 * tree-sra.c (compare_access_positions): Put integral types first,
10530 stabilize sorting of integral types, remove conditions putting
10531 non-full-precision integers last.
10532 (sort_and_splice_var_accesses): Disable scalarization if a
10533 non-integert would be represented by a non-full-precision integer.
10535 2017-09-26 Joseph Myers <joseph@codesourcery.com>
10537 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
10538 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
10539 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
10540 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
10541 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
10542 conditionals inside the function instead of around it. Call
10543 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
10544 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
10546 2017-09-26 Richard Biener <rguenther@suse.de>
10548 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
10550 (scop_detection::build_scop_breadth): ... this. Removed.
10551 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
10552 (scop_detection::harmful_stmt_in_bb): Likewise.
10553 (scop_detection::graphite_can_represent_stmt): Likewise.
10554 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
10555 (scop_detection::can_represent_loop): Remove recursion, fold in ...
10556 (scop_detection::can_represent_loop_1): ... this. Removed.
10557 (scop_detection::harmful_loop_in_region): Simplify after inlining
10558 the above and remove more quadraticness.
10559 (build_scops): Adjust.
10560 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
10563 2017-09-26 Jakub Jelinek <jakub@redhat.com>
10566 * config/i386/i386.c (ix86_print_operand_address_as): Only test
10567 REGNO (base) == SP_REG if base is a REG.
10569 PR middle-end/35691
10570 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
10571 if it is different SSA_NAME.
10572 (optimize_range_tests_cmp_bitwise): New function.
10573 (optimize_range_tests): Call it.
10575 2017-09-26 Richard Biener <rguenther@suse.de>
10577 PR tree-optimization/82321
10578 * graphite.c (canonicalize_loop_closed_ssa): Properly check
10579 for the def being inside the loop.
10581 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10583 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
10585 * config/s390/s390-builtins.def: Fix constraint on op4.
10587 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10589 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
10590 independent expanders.
10591 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
10592 ("vec_ordered", "vec_unordered"): New expanders.
10594 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10596 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
10599 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10601 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
10602 vec_unpacks_lo_v16qi.
10603 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
10605 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10607 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
10608 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
10609 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
10611 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10613 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
10615 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
10617 ("*vec_slb<mode>"): New insn pattern.
10618 ("vec_shr_<mode>"): New expander.
10619 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
10620 and force the shift count operand to V16QImode.
10621 ("vec_srb<mode>"): Set shift count mode to V16QI.
10623 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10625 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
10626 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
10627 ("vec_widen_smult_hi_<mode>"): New expander definitions.
10629 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
10632 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
10634 2017-09-26 Richard Biener <rguenther@suse.de>
10636 PR tree-optimization/82320
10637 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
10640 2017-09-25 Jeff Law <law@redhat.com>
10642 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
10643 prototype for new argument.
10644 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
10645 mostly extracted from rs6000_emit_allocate_stack.
10646 (rs6000_emit_probe_stack_range_stack_clash): New function.
10647 (rs6000_emit_allocate_stack): Call
10648 rs6000_emit_probe_stack_range_stack_clash as needed.
10649 (rs6000_emit_probe_stack_range): Add additional argument
10650 to call to gen_probe_stack_range{si,di}.
10651 (output_probe_stack_range): New.
10652 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
10653 (output_probe_stack_range_stack_clash): New.
10654 (rs6000_emit_prologue): Emit notes into dump file as requested.
10655 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
10656 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
10657 Add additional operand and pass it to output_probe_stack_range.
10659 2017-09-25 Bin Cheng <bin.cheng@arm.com>
10661 PR tree-optimization/82163
10662 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
10663 (checking_verify_loop_closed_ssa): New parameter.
10664 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
10665 (check_loop_closed_ssa_stmt): Delete.
10666 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
10667 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
10668 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
10671 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
10673 * brig-builtins.def: Treat HSAIL barrier builtins as
10674 setjmp/longjump style functions.
10676 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10678 * target.def (constant_alignment): New hook.
10679 * defaults.h (CONSTANT_ALIGNMENT): Delete.
10680 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
10681 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
10682 * doc/tm.texi: Regenerate.
10683 * targhooks.h (default_constant_alignment): Declare.
10684 (constant_alignment_word_strings): Likewise.
10685 * targhooks.c (default_constant_alignment): New function.
10686 (constant_alignment_word_strings): Likewise.
10687 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
10688 instead of CONSTANT_ALIGNMENT.
10689 * varasm.c (align_variable, get_variable_align, build_constant_desc)
10690 (force_const_mem): Likewise.
10691 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
10692 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
10693 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
10694 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10695 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
10697 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
10698 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10699 constant_alignment_word_strings.
10700 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
10701 (CONSTANT_ALIGNMENT): Likewise.
10702 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10703 (arm_constant_alignment): New function.
10704 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
10705 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10706 constant_alignment_word_strings.
10707 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
10708 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10709 constant_alignment_word_strings.
10710 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
10711 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10712 (cris_constant_alignment): New function.
10713 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
10714 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10715 (epiphany_constant_alignment): New function.
10716 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
10717 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10718 constant_alignment_word_strings.
10719 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
10720 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10721 constant_alignment_word_strings.
10722 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
10723 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10724 constant_alignment_word_strings.
10725 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
10726 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
10727 * config/i386/i386.c (ix86_constant_alignment): Make static.
10728 Use the same interface as the target hook.
10729 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10730 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
10731 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10732 constant_alignment_word_strings.
10733 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
10734 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
10735 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10736 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
10737 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10738 constant_alignment_word_strings.
10739 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
10740 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10741 constant_alignment_word_strings.
10742 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
10743 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10744 constant_alignment_word_strings.
10745 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
10746 * config/microblaze/microblaze.c (microblaze_constant_alignment):
10748 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10749 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
10750 * config/mips/mips.c (mips_constant_alignment): New function.
10751 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10752 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
10753 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
10754 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10755 (mmix_constant_alignment): Make static. Use the same interface
10756 as the target hook.
10757 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
10758 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10759 constant_alignment_word_strings.
10760 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
10761 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10762 constant_alignment_word_strings.
10763 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
10764 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10765 constant_alignment_word_strings.
10766 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
10767 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10768 (rs6000_constant_alignment): New function.
10769 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
10770 * config/riscv/riscv.c (riscv_constant_alignment): New function.
10771 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10772 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
10773 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10774 (rs6000_constant_alignment): New function.
10775 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
10776 * config/s390/s390.c (s390_constant_alignment): New function.
10777 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10778 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
10779 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10780 constant_alignment_word_strings.
10781 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
10782 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10783 (sparc_constant_alignment): New function.
10784 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
10785 * config/spu/spu.c (spu_constant_alignment): New function.
10786 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10787 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
10788 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10789 constant_alignment_word_strings.
10790 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
10791 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10792 constant_alignment_word_strings.
10793 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
10794 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10795 constant_alignment_word_strings.
10796 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
10797 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10798 (visium_constant_alignment): New function.
10799 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
10800 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10801 (xtensa_constant_alignment): New function.
10802 * system.h (CONSTANT_ALIGNMENT): Poison.
10804 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
10806 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10807 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
10808 (rs6000_builtin_valid_without_lhs): New helper function.
10809 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10810 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
10812 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10814 * target.h (vec_perm_indices): Use unsigned short rather than
10816 (auto_vec_perm_indices): Likewise.
10817 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
10818 Use unsigned int rather than unsigned char.
10819 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
10821 2017-09-25 Richard Biener <rguenther@suse.de>
10823 * cfgloop.h (sort_sibling_loops): Declare.
10824 * cfgloop.c (sort_sibling_loops_cmp): New helper.
10825 (sort_sibling_loops): New function sorting the sibling loop list
10827 * graphite.c (graphite_transform_loops): Sort sibling loops.
10829 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
10831 * target.def (vec_perm_const_ok): Change sel parameter to
10833 * optabs-query.c (can_vec_perm_p): Update accordingly.
10834 * doc/tm.texi: Regenerate.
10835 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
10836 auto_vec_perm_indices and remove separate nelt field.
10837 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
10838 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
10839 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
10840 (aarch64_expand_vec_perm_const): Update accordingly.
10841 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
10842 to vec_perm_indices.
10843 * config/arm/arm.c (expand_vec_perm_d): Change perm to
10844 auto_vec_perm_indices and remove separate nelt field.
10845 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
10846 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
10847 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
10849 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
10850 to vec_perm_indices.
10851 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
10852 sel to vec_perm_indices.
10853 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
10854 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
10855 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
10857 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
10860 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
10863 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
10864 on the FUNCTION_DECL function context if it has a DIE that is a
10867 2017-09-25 Richard Biener <rguenther@suse.de>
10869 PR tree-optimization/82285
10870 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
10873 2017-09-25 Tom de Vries <tom@codesourcery.com>
10877 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
10880 2017-09-25 Richard Biener <rguenther@suse.de>
10882 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
10883 ISL errors other than isl_error_quota happen. Dump if the
10884 schedule is the same.
10885 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
10886 errors instead of aborting inside ISL.
10888 2017-09-25 Iain Sandoe <iain@codesourcery.com>
10891 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
10892 of libgcc_eh for m64.
10893 * config/i386/darwin64.h: Likewise.
10895 2017-09-25 Richard Biener <rguenther@suse.de>
10897 PR middle-end/82144
10898 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
10899 attribute for incomplete types nor twice for complete ones.
10901 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
10904 * config/i386/i386.c (ix86_print_operand_address_as): Encode
10905 %esp as %rsp to avoid 0x67 prefix if there is no index or base
10908 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
10911 * config/i386/i386.opt (mprefer-avx256): Use
10912 ix86_target_flags variable.
10913 * config/i386/i386.c (ix86_target_string): Move
10914 -mprefer-avx256 to flag2_opts.
10916 2017-09-22 Jakub Jelinek <jakub@redhat.com>
10918 PR middle-end/35691
10919 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
10920 and x != -1 | y != -1 into (x & y) != -1.
10922 2017-09-22 Steve Ellcey <sellcey@cavium.com>
10924 * config.gcc: Add new case statement to set
10925 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
10926 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
10927 s390x-*-linux* case statements. Added aarch64 to the list of
10928 supported architectures.
10930 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10932 PR tree-optimization/82289
10933 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
10934 STMT_VINFO_RELEVANT_P.
10936 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10937 Alan Hayward <alan.hayward@arm.com>
10938 David Sherwood <david.sherwood@arm.com>
10940 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
10941 for VR_RANGE only; don't allow VR_ANTI_RANGE.
10942 (extract_range_from_binary_expr_1): Don't call
10943 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
10945 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10946 Alan Hayward <alan.hayward@arm.com>
10947 David Sherwood <david.sherwood@arm.com>
10949 * target.def (preferred_vector_alignment): New hook.
10950 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
10952 * doc/tm.texi: Regenerate.
10953 * targhooks.h (default_preferred_vector_alignment): Declare.
10954 * targhooks.c (default_preferred_vector_alignment): New function.
10955 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
10957 (DR_TARGET_ALIGNMENT): New macro.
10958 (aligned_access_p): Update commentary.
10959 (vect_known_alignment_in_bytes): New function.
10960 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
10962 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
10963 Calculate the misalignment based on the target alignment rather than
10965 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
10966 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
10967 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
10968 the target alignment, rather than masking the element misalignment
10969 with the number of elements in a vector. Also use the target
10970 alignment when calculating the maximum number of peels.
10971 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
10972 instead of TYPE_ALIGN_UNIT.
10973 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
10974 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
10975 (vect_create_addr_base_for_vector_ref): Update call accordingly.
10976 (vect_create_data_ref_ptr): Likewise.
10977 (vect_setup_realignment): Realign by ANDing with
10978 -DR_TARGET_MISALIGNMENT.
10979 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
10980 the number of peels based on DR_TARGET_ALIGNMENT.
10981 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
10982 with the guaranteed alignment boundary when deciding whether
10984 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
10985 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
10986 (ensure_base_align): Remove stmt_info parameter. Get the
10987 target base alignment from DR_TARGET_ALIGNMENT.
10988 (vectorizable_store): Update call accordingly. Interpret
10989 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
10991 (vectorizable_load): Likewise.
10993 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10994 Alan Hayward <alan.hayward@arm.com>
10995 David Sherwood <david.sherwood@arm.com>
10997 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
10998 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
10999 (vect_enhance_data_refs_alignment): Likewise.
11001 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11003 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
11004 error. Only quit immediately if parsing is complete.
11005 (BEGIN): Initialize fatal_err and parse_done.
11006 (begin fpu, end fpu): Check number of arguments.
11007 (begin arch, end arch): Likewise.
11008 (begin cpu, end cpu): Likewise.
11009 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
11010 (optalias): Likewise.
11012 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11014 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
11015 * config/arm/arm-isa.h: Delete. Move definitions to ...
11016 * arm-cpus.in: ... here. Use new feature and fgroup values.
11017 * config/arm/arm.c (arm_option_override): Use lower case for feature
11019 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
11020 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
11021 * config/arm/parsecpu.awk (END): Add new command 'isa'.
11023 (print_isa_bits_for): New function.
11024 (gen_isa): New function.
11025 (gen_comm_data): Use print_isa_bits_for.
11026 (define feature): New keyword.
11027 (define fgroup): New keyword.
11028 * config/arm/t-arm (TM_H): Remove.
11029 (GTM_H): Add arm-isa.h.
11030 (arm-isa.h): Add rule to generate file.
11031 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
11032 case for feature bit names.
11034 2017-09-22 Richard Biener <rguenther@suse.de>
11036 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
11038 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
11039 print of no dependency loops ...
11040 * graphite.c (graphite_transform_loops): ... here.
11041 (canonicalize_loop_closed_ssa_form): Work from inner to outer
11043 (same_close_phi_node, remove_duplicate_close_phi,
11044 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
11045 (canonicalize_loop_closed_ssa): ... here and simplify.
11046 * graphite-optimize-isl.c: Include tree-vectorizer.h.
11047 (optimize_isl): Use dump_printf_loc to tell when we stopped
11048 optimizing because of an ISL timeout.
11050 2017-09-22 Richard Biener <rguenther@suse.de>
11052 PR tree-optimization/82291
11053 * tree-if-conv.c (predicate_mem_writes): Make sure to
11054 remove writes in blocks predicated with false.
11056 2017-09-22 Richard Biener <rguenther@suse.de>
11058 * sese.c: Include cfganal.h.
11059 (if_region_set_false_region): Remove.
11060 (create_if_region_on_edge): Likewise.
11061 (move_sese_in_condition): Re-implement without destroying
11064 2017-09-22 Richard Biener <rguenther@suse.de>
11066 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
11067 Verify both BBs contain loop PHI nodes before dispatching to
11068 copy_loop_phi_args.
11069 (graphite_regenerate_ast_isl): Do not recompute dominators,
11070 do not verify three times. Restructure for clarity.
11071 * graphite-scop-detection.c (same_close_phi_node,
11072 remove_duplicate_close_phi, make_close_phi_nodes_unique,
11073 defined_in_loop_p, canonicalize_loop_closed_ssa,
11074 canonicalize_loop_closed_ssa_form): Simplify, remove excess
11075 checking and SSA rewrite, move to ...
11076 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
11077 (graphite_initialize): Do not pass in ctx, do not reset the
11078 SCEV cache, compute only dominators.
11079 (graphite_transform_loops): Allocate ISL ctx after
11080 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
11081 Maintain post-dominators only around build_scops.
11082 * sese.c (if_region_set_false_region): Make static. Free
11083 and recompute dominators.
11084 (move_sese_in_condition): Assert we don't get called with
11085 post-dominators computed.
11086 * sese.h (if_region_set_false_region): Remove.
11088 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
11090 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
11091 mode attribute for TARGET_AVX512VL.
11093 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
11095 * config/i386/i386.opt (mprefer-avx256): New option.
11096 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
11098 (ix86_preferred_simd_mode): Return 256-bit AVX modes
11099 for TARGET_PREFER_AVX256.
11100 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
11102 2017-09-21 Jeff Law <law@redhat.com>
11104 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
11105 Fix dump output if the only stack space is for pushed registers.
11107 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11109 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
11112 2017-09-21 Martin Sebor <msebor@redhat.com>
11115 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
11116 code (in C++) or code that triggers warnings.
11118 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
11120 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
11122 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
11124 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
11125 * sched-rgn.c: Ditto.
11126 * sel-sched-ir.c: Ditto.
11128 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
11130 * toplev.h (set_random_seed): Adjust return type.
11131 * toplev.c (init_local_tick): Move eager initialization of random_seed
11132 to get_random_seed. Adjust comment.
11133 (init_random_seed): Inline to get_random_seed, delete.
11134 (get_random_seed): Initialize random_seed lazily.
11135 (set_random_seed): Do not return previous value.
11136 (print_switch_value): Do not call get_random_seed.
11138 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
11140 * cgraph.c (delete_function_version): New, broken out from...
11141 (cgraph_node::delete_function_version): ...here. Rename to
11142 cgraph_node::delete_function_version_by_decl. Update all uses.
11143 (cgraph_node::remove): Call delete_function_version.
11145 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11148 * tree-inline.c (expand_call_inline): Emit clobber stmts for
11149 VAR_DECLs to which addressable non-volatile parameters are mapped
11150 and for id->retvar after the return value assignment. Clear
11151 id->retval and id->retbnd after inlining.
11153 2017-09-21 Richard Biener <rguenther@suse.de>
11155 PR tree-optimization/82276
11156 PR tree-optimization/82244
11157 * tree-vrp.c (build_assert_expr_for): Set
11158 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
11160 (remove_range_assertions): Revert earlier change.
11162 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
11165 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
11167 2017-09-21 Richard Biener <rguenther@suse.de>
11169 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11170 Restore valid IL after code generation errors.
11171 * graphite.c (graphite_transform_loops): Diagnose code
11172 generation issues as MSG_MISSED_OPTIMIZATION and continue
11173 with processing SCOPs.
11175 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11176 Alan Hayward <alan.hayward@arm.com>
11177 David Sherwood <david.sherwood@arm.com>
11179 * calls.c (compute_argument_addresses): Use simplify_gen_binary
11180 rather than choosing between plus_constant and gen_rtx_<CODE>.
11181 * expr.c (emit_push_insn): Likewise.
11182 (expand_expr_real_2): Likewise.
11184 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11185 Alan Hayward <alan.hayward@arm.com>
11186 David Sherwood <david.sherwood@arm.com>
11188 * loop-unroll.c (split_iv): Call copy_rtx on the step.
11190 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11191 Alan Hayward <alan.hayward@arm.com>
11192 David Sherwood <david.sherwood@arm.com>
11194 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
11195 calling tree_to_uhwi.
11197 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11198 Alan Hayward <alan.hayward@arm.com>
11199 David Sherwood <david.sherwood@arm.com>
11201 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
11202 INTEGER_CST rather than a negative test for ADDR_EXPR.
11204 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11205 Alan Hayward <alan.hayward@arm.com>
11206 David Sherwood <david.sherwood@arm.com>
11208 * tree-vrp.c (extract_range_from_binary_expr_1): Check
11209 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
11211 2017-09-21 Richard Biener <rguenther@suse.de>
11213 PR tree-optimization/71351
11214 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
11215 graphite_create_new_loop_guard): Remove, fold remaining parts
11217 (translate_isl_ast_node_for): ... here and simplify.
11219 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11222 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
11223 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
11224 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
11225 alternative always use QI mode, for -Os imov (=R,R) alternative
11226 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
11229 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11230 Jeff Law <law@redhat.com>
11232 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
11233 (allocate_stack_space): New function, partially extracted from
11234 s390_emit_prologue.
11235 (s390_emit_prologue): Track offset to most recent stack probe.
11236 Code to allocate space moved into allocate_stack_space.
11237 Dump actions when no stack is allocated.
11238 (s390_prologue_plus_offset): New function.
11239 (s390_emit_stack_probe): Likewise.
11241 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
11243 * common.opt (Wa, Wl, Wp, g, gz=): Add
11245 (gno-column-info): Remove.
11246 (gcolumn-info): Drop RejectNegative.
11247 (gno-): New prefix.
11248 (gno-record-gcc-switches): Remove.
11249 (grecord-gcc-switches): Drop RejectNegative.
11250 (gno-split-dwarf): Remove.
11251 (gsplit-dwarf): Drop RejectNegative.
11252 (gno-strict-dwarf): Remove.
11253 (gstrict-dwarf): Drop RejectNegative.
11254 * config/darwin.opt (gfull, gused): Add RejectNegative.
11255 * dwarf2out.c (gen_producer_string): Drop
11256 gno-record-gcc-switches handler.
11257 * optc-gen.awk: Add g to prefixes with negative forms.
11258 * opts-common.c (remapping_prefix_p): New.
11259 (find_opt): Check it.
11260 (generate_canonical_option): Test g prefix.
11261 (option_map): Add -gno- mapping.
11262 (add_misspelling_candidates): Check remapping_prefix_p.
11264 2017-09-20 Jeff Law <law@redhat.com>
11266 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
11267 thinko in stack clash protection support.
11269 * explow.c (compute_stack_clash_protection_loop_data): Use
11270 CONST_INT_P instead of explicit test. Verify object is a
11271 CONST_INT_P before looking at INTVAL.
11272 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
11273 instead of explicit test.
11275 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
11278 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
11279 address instead of to r1 and r11.
11281 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
11283 * config.gcc: Support "knm".
11284 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
11285 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
11287 * config/i386/i386.c (m_KNM): Define.
11288 (processor_target_table): Add "knm".
11290 (ix86_option_override_internal): Add "knm".
11291 (ix86_issue_rate): Add PROCESSOR_KNM.
11292 (ix86_adjust_cost): Ditto.
11293 (ia32_multipass_dfa_lookahead): Ditto.
11294 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
11295 (fold_builtin_cpu): Add M_INTEL_KNM.
11296 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
11297 (processor_type): Add PROCESSOR_KNM.
11298 * config/i386/x86-tune.def: Add m_KNM.
11299 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
11301 2017-09-20 Richard Biener <rguenther@suse.de>
11303 PR tree-optimization/80213
11304 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
11305 are allowed in empty BBs as well.
11306 (canonicalize_loop_closed_ssa): Also look for other complex
11308 (scop_detection::get_sese): Include the loop-closed PHI block
11310 (scop_detection::merge_sese): Remove code adding extra blocks.
11311 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
11312 (build_scops): Assert the final returned scop is invalid.
11314 2017-09-20 Richard Biener <rguenther@suse.de>
11316 PR tree-optimization/82264
11317 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
11319 (vn_phi_lookup): Likewise.
11320 (vn_phi_insert): Likewise.
11322 2017-09-20 Jakub Jelinek <jakub@redhat.com>
11324 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
11325 that fits into uhwi or shwi, add DW_AT_const_value regardless
11326 of early_dwarf without going through RTL, using add_AT_unsigned
11329 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
11330 (DEBUG_LTO_ABBREV_SECTION): Likewise.
11331 (DEBUG_LTO_MACINFO_SECTION): Likewise.
11332 (DEBUG_MACRO_SECTION): Likewise.
11333 (DEBUG_LTO_MACRO_SECTION): Likewise.
11334 (DEBUG_STR_DWO_SECTION): Likewise.
11335 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
11336 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
11337 (DEBUG_LTO_DWO_LINE_SECTION): Define.
11338 (DEBUG_LTO_LINE_STR_SECTION): Define.
11339 (init_sections_and_labels): Initialize debug_line_str_section
11340 variable. Initialize debug_loc_section for -gdwarf-5 to
11341 DEBUG_LOCLISTS_SECTION. Formatting fixes.
11343 2017-09-20 Richard Biener <rguenther@suse.de>
11345 * graphite-sese-to-poly.c (extract_affine): Properly handle
11346 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
11348 2017-09-20 Richard Biener <rguenther@suse.de>
11350 PR tree-optimization/81373
11351 * graphite-scop-detection.c (build_cross_bb_scalars_def):
11352 Force SESE live-out defs to be handled even if they are
11355 2017-09-19 Jeff Law <law@redhat.com>
11357 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
11358 nothing for stack adjustments with REG_STACK_CHECK.
11359 * sched-deps.c (parse_add_or_inc): Reject insns with
11360 REG_STACK_CHECK from dependency breaking.
11361 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
11362 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
11363 * reg-notes.def (STACK_CHECK): New note.
11365 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
11366 (ix86_expand_prologue): Dump stack clash info as needed.
11367 Call ix86_adjust_stack_and_probe_stack_clash as needed.
11369 * function.c (dump_stack_clash_frame_info): New function.
11370 * function.h (dump_stack_clash_frame_info): Prototype.
11371 (enum stack_clash_probes): New enum.
11373 * config/alpha/alpha.c (alpha_expand_prologue): Also check
11374 flag_stack_clash_protection.
11375 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
11376 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
11377 (arm_frame_pointer_required): Likewise.
11378 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
11379 (ia64_expand_prologue): Likewise.
11380 * config/mips/mips.c (mips_expand_prologue): Likewise.
11381 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
11382 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11383 (sparc_flat_expand_prologue): Likewise.
11384 * config/spu/spu.c (spu_expand_prologue): Likewise.
11386 * explow.c: Include "params.h".
11387 (anti_adjust_stack_and_probe_stack_clash): New function.
11388 (get_stack_check_protect): Likewise.
11389 (compute_stack_clash_protection_loop_data): Likewise.
11390 (emit_stack_clash_protection_loop_start): Likewise.
11391 (emit_stack_clash_protection_loop_end): Likewise.
11392 (allocate_dynamic_stack_space): Use get_stack_check_protect.
11393 Use anti_adjust_stack_and_probe_stack_clash.
11394 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
11395 (emit_stack_clash_protection_loop_start): Likewise.
11396 (emit_stack_clash_protection_loop_end): Likewise.
11397 * rtl.h (get_stack_check_protect): Prototype.
11398 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
11399 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
11400 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
11402 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
11404 * doc/tm.texi: Rebuilt.
11405 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
11406 get_stack_check_protect.
11407 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
11408 * config/arm/arm.c (arm_expand_prologue): Likewise.
11409 (arm_frame_pointer_required): Likewise.
11410 * config/i386/i386.c (ix86_expand_prologue): Likewise.
11411 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
11412 * config/mips/mips.c (mips_expand_prologue): Likewise.
11413 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
11414 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
11415 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11416 (sparc_flat_expand_prologue): Likewise.
11418 * common.opt (-fstack-clash-protection): New option.
11419 * flag-types.h (enum stack_check_type): Note difference between
11420 -fstack-check= and -fstack-clash-protection.
11421 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
11422 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
11423 * toplev.c (process_options): Issue warnings/errors for cases
11424 not handled with -fstack-clash-protection.
11425 * doc/invoke.texi (-fstack-clash-protection): Document new option.
11426 (-fstack-check): Note additional problem with -fstack-check=generic.
11427 Note that -fstack-check is primarily for Ada and refer users
11428 to -fstack-clash-protection for stack-clash-protection.
11429 Document new params for stack clash protection.
11431 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
11433 * config/i386/i386.c (ix86_split_long_move): Do not handle
11434 address used for LEA in a special way.
11436 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
11438 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
11440 2017-09-19 Martin Sebor <msebor@redhat.com>
11443 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
11444 of incompatible types.
11446 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
11448 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11449 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
11450 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11451 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
11453 2017-09-19 Richard Biener <rguenther@suse.de>
11455 PR tree-optimization/82244
11456 * tree-vrp.c (remove_range_assertions): Do not propagate
11457 a constant to abnormals but replace the assert with a copy.
11459 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
11461 PR rtl-optimization/57878
11462 PR rtl-optimization/68988
11463 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
11464 avoidance test involving non_reload_pseudos. Move frequency test
11465 below the general fragmentation avoidance test.
11467 2017-09-19 Richard Biener <rguenther@suse.de>
11469 PR tree-optimization/69728
11470 * graphite-sese-to-poly.c (schedule_error): New global.
11471 (add_loop_schedule): Handle empty domain by failing the
11473 (build_original_schedule): Handle schedule_error.
11475 2017-09-19 Richard Biener <rguenther@suse.de>
11477 * graphite-scop-detection.c (scop_detection::can_represent_loop):
11478 Do not iterate to sibling loops but only to siblings of inner
11481 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
11484 * config/m68k/m68k.md (moveq feeding equality comparison): Check
11485 that the registers are different.
11487 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
11489 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
11490 to processor_model and "amdfam17h" to arch_names_table.
11491 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
11493 2017-09-18 Jakub Jelinek <jakub@redhat.com>
11496 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
11498 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11499 Alan Hayward <alan.hayward@arm.com>
11500 David Sherwood <david.sherwood@arm.com>
11502 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
11504 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
11506 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
11507 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
11508 vect_schedule_slp_instance.
11509 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
11510 Update call to vect_slp_analyze_node_operations. Simplify return
11512 (vect_slp_analyze_bb_1): Update call accordingly.
11513 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
11514 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
11515 (vect_schedule_slp): Update call accordingly.
11517 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11518 Alan Hayward <alan.hayward@arm.com>
11519 David Sherwood <david.sherwood@arm.com>
11521 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
11522 with types that aren't in fact scalar.
11524 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11526 * tree-vect-slp.c (vect_record_max_nunits): New function,
11528 (vect_build_slp_tree_1): ...here.
11529 (vect_build_slp_tree_2): Call it for phis too.
11531 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11533 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
11534 to vect_get_vec_def_for_operand when getting the mask operand.
11536 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11537 Alan Hayward <alan.hayward@arm.com>
11538 David Sherwood <david.sherwood@arm.com>
11540 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
11543 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11544 Alan Hayward <alan.hayward@arm.com>
11545 David Sherwood <david.sherwood@arm.com>
11547 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
11548 calculation for vector booleans.
11550 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11551 Alan Hayward <alan.hayward@arm.com>
11552 David Sherwood <david.sherwood@arm.com>
11554 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
11556 (vect_transform_stmt): ...here.
11557 (vect_analyze_stmt): Use it instead of calling
11558 vectorizable_live_operation directly.
11560 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
11562 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
11563 non-SIMT targets in acc vector loops.
11565 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
11567 * configure.ac: Add arc and check if assembler supports gdwarf2.
11568 * configure: Regenerate.
11570 2017-09-18 Richard Biener <rguenther@suse.de>
11572 PR tree-optimization/82220
11573 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
11574 epilogue niters from the min_profitable_iters compute.
11576 2017-09-18 Jakub Jelinek <jakub@redhat.com>
11579 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
11580 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
11581 (ix86_init_pic_reg): Revert 2017-09-01 changes.
11583 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
11586 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
11587 switching to a new text section.
11589 2017-09-18 Richard Biener <rguenther@suse.de>
11591 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
11593 (build_alias_set): Reject aliases with no access function.
11595 2017-09-18 Richard Biener <rguenther@suse.de>
11597 PR tree-optimization/79622
11598 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
11600 (build_cross_bb_scalars_use): Likewise.
11602 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
11604 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
11606 2017-09-18 Alan Modra <amodra@gmail.com>
11609 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
11610 stack_pointer_rtx for count 0. Update comments. Break up
11611 large rtl expression.
11613 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
11615 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
11616 Increase to 20 bytes.
11617 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
11618 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
11619 or avx version of the stub.
11621 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
11624 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
11625 compute the minimum stack alignment. Also update preferred stack
11626 boundary for leaf functions.
11628 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
11630 PR tree-optimization/82228
11631 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
11634 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
11636 * common/config/nds32/nds32-common.c
11637 (nds32_option_optimization_table): Refine formatting.
11638 (nds32_option_optimization_table): Use -fsched-pressure and
11639 -fomit-frame-pointer for specific optimization level.
11641 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
11643 * config/nds32/nds32.c: Refine formatting and comments.
11644 * config/nds32/nds32.h: Likewise.
11645 * config/nds32/nds32.md: Likewise.
11646 * config/nds32/nds32-cost.c: Likewise.
11647 * config/nds32/nds32-isr.c: Likewise.
11648 * config/nds32/nds32-md-auxiliary.c: Likewise.
11649 * config/nds32/nds32-multiple.md: Likewise.
11650 * config/nds32/nds32-predicates.c: Likewise.
11652 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
11653 Jakub Jelinek <jakub@redhat.com>
11655 Add support for -std=c++2a.
11656 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
11658 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
11660 2017-09-15 Steve Ellcey <sellcey@cavium.com>
11663 * doc/extend.texi (Common Function Attributes): Add
11664 references to ARM, AArch64, and S/390 specific attributes.
11665 (Function Specific Option Pragmas): Add AArch64 and S/390
11666 to list of back ends that support the target pragma.
11668 2017-09-15 Nathan Sidwell <nathan@acm.org>
11670 * doc/standards.texi: Fix C++17 description. Update URLs for
11673 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
11675 * common.opt (Wcast-align=strict): New warning option.
11676 * doc/invoke.texi: Document -Wcast-align=strict.
11678 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
11680 * cgraph.h (cgraph_thunk_info): Add comments.
11681 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
11682 assert for VIRTUAL_* arguments stricter.
11684 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
11686 PR tree-optimization/71026
11687 * match.pd: Move RDIV patterns from fold-const.c
11688 * fold-const.c (distribute_real_division): Removed.
11689 (fold_binary_loc): Remove calls to distribute_real_divison.
11691 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11693 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
11694 c++1z and gnu++1z as deprecated. Change other references to
11695 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
11696 Change -Wc++1z-compat to -Wc++17-compat.
11697 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
11698 * dwarf2out.c (highest_c_language): Handle C++17.
11699 (gen_compile_unit_die): Likewise.
11701 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11703 PR rtl-optimization/82192
11704 * combine.c (make_extraction): Don't look through non-paradoxical
11705 SUBREGs or TRUNCATE if pos + len is or might be bigger than
11708 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11709 Alan Hayward <alan.hayward@arm.com>
11710 David Sherwood <david.sherwood@arm.com>
11712 * target.def (function_arg_offset): New hook.
11713 * targhooks.h (default_function_arg_offset): Declare.
11714 * targhooks.c (default_function_arg_offset): New function.
11715 * function.c (locate_and_pad_parm): Use
11716 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
11717 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
11718 (TARGET_FUNCTION_ARG_OFFSET): ...this.
11719 * doc/tm.texi: Regenerate.
11720 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
11721 * config/spu/spu.c (spu_function_arg_offset): New function.
11722 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
11723 * system.h (FUNCTION_ARG_OFFSET): Poison.
11725 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11726 Alan Hayard <alan.hayward@arm.com>
11727 David Sherwood <david.sherwood@arm.com>
11729 * target.def (truly_noop_truncation): New hook.
11730 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
11731 than TRULY_NOOP_TRUNCATION.
11732 * hooks.h (hook_bool_uint_uint_true): Declare.
11733 * hooks.c (hook_bool_uint_uint_true): New function.
11734 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
11735 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
11736 * doc/tm.texi: Regenerate.
11737 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
11738 rather than TRULY_NOOP_TRUNCATION in comments.
11739 (simplify_comparison): Likewise.
11740 (record_truncated_value): Likewise.
11741 * expmed.c (extract_bit_field_1): Likewise.
11742 (extract_split_bit_field): Likewise.
11743 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
11744 instead of TRULY_NOOP_TRUNCATION.
11745 * function.c (assign_parm_setup_block): Likewise.
11746 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
11747 * rtlhooks.c: Include target.h.
11748 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
11749 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
11750 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
11751 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
11752 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
11753 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
11754 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
11755 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
11756 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
11757 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
11758 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
11759 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
11760 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
11761 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
11762 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
11763 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
11764 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
11765 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
11766 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
11767 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
11768 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
11769 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
11770 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
11771 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
11772 * config/mips/mips.c (mips_truly_noop_truncation): New function.
11773 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11774 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11775 rather than TRULY_NOOP_TRUNCATION in comments.
11776 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
11777 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
11778 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
11779 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
11780 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
11781 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
11782 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
11783 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
11784 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
11785 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
11786 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
11787 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11788 rather than TRULY_NOOP_TRUNCATION in comments.
11789 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
11790 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
11791 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
11792 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
11793 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
11794 TRULY_NOOP_TRUNCATION condition.
11795 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
11796 (TRULY_NOOP_TRUNCATION): Delete.
11797 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
11798 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
11799 * config/spu/spu.c (spu_truly_noop_truncation): New function.
11800 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11801 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
11802 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
11803 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
11804 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11805 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11806 rather than TRULY_NOOP_TRUNCATION in comments.
11807 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
11808 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
11809 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
11810 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
11811 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
11812 * system.h (TRULY_NOOP_TRUNCATION): Poison.
11814 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
11817 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
11818 (*cmp_ior): Likewise.
11819 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
11820 (*ior_scc_scc_cmp): Likewise.
11821 (*and_scc_scc): Likewise.
11822 (*and_scc_scc_cmp): Likewise.
11824 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11825 Alan Hayard <alan.hayward@arm.com>
11826 David Sherwood <david.sherwood@arm.com>
11828 * target.def (can_change_mode_class): New hook.
11829 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
11830 (hard_regno_nregs): Likewise.
11831 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
11832 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
11833 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
11834 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
11835 (LOAD_EXTEND_OP): Update accordingly.
11836 * doc/tm.texi: Regenerate.
11837 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
11838 CANNOT_CHANGE_MODE_CLASS.
11839 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
11840 (REG_CAN_CHANGE_MODE_P): ...this new macro.
11841 * combine.c (simplify_set): Update accordingly.
11842 * emit-rtl.c (validate_subreg): Likewise.
11843 * recog.c (general_operand): Likewise.
11844 * regcprop.c (mode_change_ok): Likewise.
11845 * reload1.c (choose_reload_regs): Likewise.
11846 (inherit_piecemeal_p): Likewise.
11847 * rtlanal.c (simplify_subreg_regno): Likewise.
11848 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
11849 instead of CANNOT_CHANGE_MODE_CLASS.
11850 (reload_cse_simplify_operands): Likewise.
11851 * reload.c (push_reload): Use targetm.can_change_mode_class
11852 instead of CANNOT_CHANGE_MODE_CLASS.
11853 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
11854 REG_CANNOT_CHANGE_MODE_P.
11855 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11856 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
11857 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11858 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11859 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11860 (arm_can_change_mode_class): New function.
11861 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11862 than CANNOT_CHANGE_MODE_CLASS in comments.
11863 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11864 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
11865 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
11866 (ix86_can_change_mode_class): ...this new function, inverting the
11867 sense of the return value.
11868 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11869 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11870 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11871 (ia64_can_change_mode_class): New function.
11872 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11873 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
11874 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
11875 (m32c_can_change_mode_class): ...this new function, inverting the
11876 sense of the return value.
11877 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11878 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11879 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
11880 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
11881 (mips_can_change_mode_class): ...this new function, inverting the
11882 sense of the return value.
11883 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11884 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11885 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11886 (msp430_can_change_mode_class): New function.
11887 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11888 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
11889 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11890 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11891 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11892 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
11893 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11894 (pa_cannot_change_mode_class): Replace with...
11895 (pa_can_change_mode_class): ...this new function, inverting the
11896 sense of the return value.
11897 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11898 than CANNOT_CHANGE_MODE_CLASS in comments.
11899 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11900 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
11901 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11902 (pdp11_cannot_change_mode_class): Replace with...
11903 (pdp11_can_change_mode_class): ...this new function, inverting the
11904 sense of the return value.
11905 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11906 * config/powerpcspe/powerpcspe-protos.h
11907 (rs6000_cannot_change_mode_class_ptr): Delete.
11908 * config/powerpcspe/powerpcspe.c
11909 (rs6000_cannot_change_mode_class_ptr): Delete.
11910 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11911 (rs6000_option_override_internal): Assign to
11912 targetm.can_change_mode_class instead of
11913 rs6000_cannot_change_mode_class_ptr.
11914 (rs6000_cannot_change_mode_class): Replace with...
11915 (rs6000_can_change_mode_class): ...this new function, inverting the
11916 sense of the return value.
11917 (rs6000_debug_cannot_change_mode_class): Replace with...
11918 (rs6000_debug_can_change_mode_class): ...this new function.
11919 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11920 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
11921 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11922 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11923 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
11925 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
11926 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11927 (rs6000_option_override_internal): Assign to
11928 targetm.can_change_mode_class instead of
11929 rs6000_cannot_change_mode_class_ptr.
11930 (rs6000_cannot_change_mode_class): Replace with...
11931 (rs6000_can_change_mode_class): ...this new function, inverting the
11932 sense of the return value.
11933 (rs6000_debug_cannot_change_mode_class): Replace with...
11934 (rs6000_debug_can_change_mode_class): ...this new function.
11935 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11936 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
11937 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
11938 (s390_can_change_mode_class): ...this new function, inverting the
11939 sense of the return value.
11940 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11941 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11942 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
11943 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11944 (sh_cannot_change_mode_class): Replace with...
11945 (sh_can_change_mode_class): ...this new function, inverting the
11946 sense of the return value.
11947 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11948 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11949 (sparc_can_change_mode_class): New function.
11950 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11951 * config/spu/spu.c (spu_can_change_mode_class): New function.
11952 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11953 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11954 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11955 (visium_can_change_mode_class): New function.
11956 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
11958 2017-09-15 Richard Biener <rguenther@suse.de>
11960 PR tree-optimization/82217
11961 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
11962 but not undefined case.
11964 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11967 * postreload.c (reload_cse_simplify_operands): Skip
11968 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
11970 2017-09-15 Richard Biener <rguenther@suse.de>
11972 PR tree-optimization/68823
11973 * graphite-scop-detection.c (build_alias_set): If we have a
11974 possible dependence check whether we can handle them by just
11975 looking at the DRs DR_ACCESS_FNs.
11976 (build_scops): If build_alias_set fails, fail the SCOP.
11978 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
11980 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
11981 to support float128 built-in functions that require the ISA 3.0
11983 (BU_FLOAT128_3_HW): Likewise.
11984 (SQRTF128): Add support for the IEEE 128-bit square root and fma
11985 built-in functions.
11986 (FMAF128): Likewise.
11988 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
11989 support for built-in functions that need the ISA 3.0 IEEE 128-bit
11990 floating point instructions.
11991 (rs6000_invalid_builtin): Likewise.
11992 (rs6000_builtin_mask_names): Likewise.
11993 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
11994 (RS6000_BTM_FLOAT128_HW): Likewise.
11995 (RS6000_BTM_COMMON): Likewise.
11996 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
11998 * doc/extend.texi (RS/6000 built-in functions): Document the
11999 IEEE 128-bit floating point square root and fused multiply-add
12000 built-in functions.
12002 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
12004 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
12005 reg (r2) isn't in the set of registers defined in the prologue.
12007 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12008 Alan Hayward <alan.hayward@arm.com>
12009 David Sherwood <david.sherwood@arm.com>
12011 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
12012 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
12013 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
12014 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
12015 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
12017 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
12018 max_vectorization_factor.
12019 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
12021 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12022 Alan Hayward <alan.hayward@arm.com>
12023 David Sherwood <david.sherwood@arm.com>
12025 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
12026 (vect_worthwhile_without_simd_p): Declare.
12027 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
12028 (vectorizable_reduction): Use it.
12029 * tree-vect-stmts.c (vectorizable_shift): Likewise.
12030 (vectorizable_operation): Likewise.
12032 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12033 Alan Hayward <alan.hayward@arm.com>
12034 David Sherwood <david.sherwood@arm.com>
12036 * tree-vectorizer.h (vect_get_num_copies): New function.
12037 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
12038 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12039 (vectorizable_induction): Likewise.
12040 (vectorizable_live_operation): Likewise.
12041 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12042 (vectorizable_bswap): Likewise.
12043 (vectorizable_call): Likewise.
12044 (vectorizable_conversion): Likewise.
12045 (vectorizable_assignment): Likewise.
12046 (vectorizable_shift): Likewise.
12047 (vectorizable_operation): Likewise.
12048 (vectorizable_store): Likewise.
12049 (vectorizable_load): Likewise.
12050 (vectorizable_condition): Likewise.
12051 (vectorizable_comparison): Likewise.
12052 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
12054 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12055 Alan Hayward <alan.hayward@arm.com>
12056 David Sherwood <david.sherwood@arm.com>
12058 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
12059 of vect_init_vector.
12061 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12062 Alan Hayward <alan.hayward@arm.com>
12063 David Sherwood <david.sherwood@arm.com>
12065 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
12066 an inline wrapper that provides a location.
12067 (gimple_build_vector): Likewise.
12068 * gimple-fold.c (gimple_build_vector_from_val): New function.
12069 (gimple_build_vector): Likewise.
12070 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
12071 functions to build the initial value. Always return a gimple value.
12072 (get_initial_defs_for_reduction): Likewise. Only compute
12074 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
12075 vect_init_vector on the results from get_initial_def(s)_for_reduction.
12076 (vectorizable_induction): Use gimple_build_vector rather than
12079 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12080 Alan Hayward <alan.hayward@arm.com>
12081 David Sherwood <david.sherwood@arm.com>
12083 * target.h (vec_perm_indices): New typedef.
12084 (auto_vec_perm_indices): Likewise.
12085 * optabs-query.h: Include target.h
12086 (can_vec_perm_p): Take a vec_perm_indices *.
12087 * optabs-query.c (can_vec_perm_p): Likewise.
12088 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
12089 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12090 * tree-vect-generic.c (lower_vec_perm): Likewise.
12091 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
12092 (vect_grouped_load_supported): Likewise.
12093 (vect_shift_permute_load_chain): Likewise.
12094 (vect_permute_store_chain): Use auto_vec_perm_indices.
12095 (vect_permute_load_chain): Likewise.
12096 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
12097 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
12098 Update uses of can_vec_perm_p.
12099 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
12100 mode with a number of elements. Take a vec_perm_indices *.
12101 (vect_create_epilog_for_reduction): Update accordingly.
12102 Use auto_vec_perm_indices.
12103 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
12104 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
12105 (vect_transform_slp_perm_load): Likewise.
12106 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
12107 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
12108 (vect_gen_perm_mask_checked): Likewise.
12109 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
12110 (vect_gen_perm_mask_checked): Likewise.
12111 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
12112 (vectorizable_store): Likewise.
12113 (vectorizable_load): Likewise.
12114 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
12115 (vectorizable_bswap): Likewise.
12117 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12118 Alan Hayward <alan.hayward@arm.com>
12119 David Sherwood <david.sherwood@arm.com>
12121 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
12122 * tree.c (build_vector): Likewise.
12123 (build_vector_from_ctor): Update accordingly.
12124 (build_vector_from_val): Likewise.
12125 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
12126 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12127 * tree-vect-generic.c (add_rshift): Likewise.
12128 (expand_vector_divmod): Likewise.
12129 (optimize_vector_constructor): Likewise.
12130 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
12131 (vect_transform_slp_perm_load): Likewise.
12132 (vect_schedule_slp_instance): Likewise.
12133 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
12134 (vectorizable_call): Likewise.
12135 (vect_gen_perm_mask_any): Likewise. Add elements in order.
12136 * expmed.c (make_tree): Likewise.
12137 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
12138 a vector passed to build_vector.
12139 (fold_convert_const): Likewise.
12140 (exact_inverse): Likewise.
12141 (fold_ternary_loc): Likewise.
12142 (fold_relational_const): Likewise.
12143 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
12144 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
12145 (const_unop): Likewise. Store the reduction accumulator in a
12146 variable rather than an array.
12147 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
12148 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
12149 the new vector, rather than constructing it after the input arrays.
12150 (native_interpret_vector): Use auto_vec<tree> when building
12151 a vector passed to build_vector. Add elements in order.
12152 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
12153 auto_vec<tree> when building a vector passed to build_vector.
12154 (vect_create_epilog_for_reduction): Likewise.
12155 (vectorizable_induction): Likewise.
12156 (get_initial_def_for_reduction): Likewise. Fix indentation of
12158 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
12160 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
12161 passed to build_vector.
12163 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12164 Alan Hayward <alan.hayward@arm.com>
12165 David Sherwood <david.sherwood@arm.com>
12167 * tree-core.h (tree_base::u): Add an "nelts" field.
12168 (tree_vector): Use VECTOR_CST_NELTS as the length.
12169 * tree.c (tree_size): Likewise.
12170 (make_vector): Initialize VECTOR_CST_NELTS.
12171 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
12172 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
12173 TYPE_VECTOR_SUBPARTS.
12174 * expr.c (const_vector_mask_from_tree): Consistently use "units"
12175 as the number of units, setting it from VECTOR_CST_NELTS.
12176 (const_vector_from_tree): Likewise.
12177 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
12178 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
12179 (fold_negate_expr_1): Likewise.
12180 (fold_convert_const): Likewise.
12181 (const_binop): Likewise. Differentiate the number of output and
12183 (const_unop): Likewise.
12184 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
12185 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
12186 in cases that did the opposite.
12188 2017-09-14 Richard Biener <rguenther@suse.de>
12190 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
12193 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
12195 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
12197 2017-09-14 Jakub Jelinek <jakub@redhat.com>
12200 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
12201 if and where to split a bb, except for splitting before debug insn
12202 sequences followed by non-label real insn. Delete debug insns
12203 in between basic blocks.
12205 * combine.c (make_compound_operation_int): Formatting fixes.
12207 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
12208 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12209 * config/netbsd.h (LINK_EH_SPEC): Likewise.
12210 * config/sol2.h (LINK_EH_SPEC): Likewise.
12211 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12212 * config/s390/linux.h (LINK_SPEC): Likewise.
12213 * config/freebsd.h (LINK_EH_SPEC): Likewise.
12214 * config/openbsd.h (LINK_EH_SPEC): Likewise.
12215 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12216 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
12217 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
12218 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12219 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
12220 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
12222 2017-09-13 Jakub Jelinek <jakub@redhat.com>
12224 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
12226 (ENDFILE_LINUX_SPEC): Likewise.
12227 (LINK_EH_SPEC): Likewise.
12228 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
12229 (LINK_OS_LINUX_SPEC32): Likewise.
12230 (LINK_OS_LINUX_SPEC64): Likewise.
12231 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
12232 (LINK_OS_LINUX_SPEC): Likewise.
12234 2017-09-13 Martin Liska <mliska@suse.cz>
12236 PR middle-end/82154
12237 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
12238 CASE_HIGH is NULL_TREE.
12240 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12241 Alan Hayward <alan.hayward@arm.com>
12242 David Sherwood <david.sherwood@arm.com>
12244 * target.def (secondary_memory_needed): New hook.
12245 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
12246 instead of SECONDARY_MEMORY_NEEDED.
12247 (secondary_memory_needed_mode): Likewise.
12248 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
12249 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
12250 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
12251 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
12252 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
12253 * doc/tm.texi: Regenerate.
12254 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
12255 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
12256 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12257 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
12258 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
12259 * config/i386/i386.c (inline_secondary_memory_needed): Put the
12260 mode argument first and change the reg_class arguments to reg_class_t.
12261 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
12262 Make static. Update the call to inline_secondary_memory_needed.
12263 (ix86_register_move_cost): Update the call to
12264 inline_secondary_memory_needed.
12265 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12266 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
12268 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
12269 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12271 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
12272 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
12273 * config/mips/mips.c (mips_secondary_memory_needed): Make static
12274 and match hook interface. Add comment from mips.h.
12275 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12276 * config/mmix/mmix.md (truncdfsf2): Refer to
12277 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12279 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
12280 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
12281 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12282 (pa_secondary_memory_needed): New function.
12283 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
12284 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
12285 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12286 (pdp11_secondary_memory_needed): Make static and match hook interface.
12287 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
12288 * config/powerpcspe/powerpcspe-protos.h
12289 (rs6000_secondary_memory_needed_ptr): Delete.
12290 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
12292 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12293 (rs6000_option_override_internal): Assign to
12294 targetm.secondary_memory_needed rather than
12295 rs6000_secondary_memory_needed_ptr.
12296 (rs6000_secondary_memory_needed): Match hook interface.
12297 (rs6000_debug_secondary_memory_needed): Likewise.
12298 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
12299 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
12300 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
12301 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12302 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
12303 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
12305 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
12306 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12307 (rs6000_option_override_internal): Assign to
12308 targetm.secondary_memory_needed rather than
12309 rs6000_secondary_memory_needed_ptr.
12310 (rs6000_secondary_memory_needed): Match hook interface.
12311 (rs6000_debug_secondary_memory_needed): Likewise.
12312 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
12313 * config/s390/s390.c (s390_secondary_memory_needed): New function.
12314 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12315 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
12316 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12317 (sparc_secondary_memory_needed): New function.
12318 * lra-constraints.c (check_and_process_move): Refer to
12319 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12321 (curr_insn_transform): Likewise.
12322 (process_alt_operands): Use targetm.secondary_memory_needed
12323 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12324 (check_secondary_memory_needed_p): Likewise.
12325 (choose_split_class): Likewise.
12326 * reload.c: Unconditionally include code that was previously
12327 conditional on SECONDARY_MEMORY_NEEDED.
12328 (push_secondary_reload): Use targetm.secondary_memory_needed
12329 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12330 (push_reload): Likewise.
12331 * reload1.c: Unconditionally include code that was previously
12332 conditional on SECONDARY_MEMORY_NEEDED.
12333 (choose_reload_regs): Use targetm.secondary_memory_needed
12334 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12335 (gen_reload): Likewise.
12336 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
12338 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12339 Alan Hayward <alan.hayward@arm.com>
12340 David Sherwood <david.sherwood@arm.com>
12342 * target.def (secondary_memory_needed_mode): New hook:
12343 * targhooks.c (default_secondary_memory_needed_mode): Declare.
12344 * targhooks.h (default_secondary_memory_needed_mode): New function.
12345 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
12346 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
12347 * doc/tm.texi: Regenerate.
12348 * lra-constraints.c (check_and_process_move): Use
12349 targetm.secondary_memory_needed_mode instead of
12350 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
12351 (curr_insn_transform): Likewise.
12352 * reload.c (get_secondary_mem): Likewise.
12353 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12354 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
12356 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12357 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12358 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
12359 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12360 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
12362 * config/powerpcspe/powerpcspe-protos.h
12363 (rs6000_secondary_memory_needed_mode): Delete.
12364 * config/powerpcspe/powerpcspe.c
12365 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12366 (rs6000_secondary_memory_needed_mode): Make static.
12367 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12368 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
12370 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12372 (rs6000_secondary_memory_needed_mode): Make static.
12373 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12374 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
12375 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12376 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12377 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12379 (sparc_secondary_memory_needed_mode): New function.
12380 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
12382 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
12384 * config/aarch64/constraints.md (Umq): New constraint.
12385 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
12387 (mov<mode>): Update condition.
12389 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12391 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
12392 when bitposition is the same.
12394 2017-09-13 Richard Biener <rguenther@suse.de>
12396 * dwarf2out.c (output_die_symbol): Remove.
12397 (output_die): Do not output a DIEs symbol.
12399 2017-09-13 Richard Biener <rguenther@suse.de>
12401 PR middle-end/82128
12402 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
12403 default-def to avoid breaking iterator update with the weird
12404 interaction with cgraph_update_edges_for_call_stmt_node.
12406 2017-09-13 Richard Biener <rguenther@suse.de>
12408 * tree-cfg.c (verify_gimple_assign_binary): Add verification
12409 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
12410 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
12411 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
12413 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
12415 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
12416 Disable pc relative literal load irrespective of
12417 TARGET_FIX_ERR_A53_84341 for default.
12419 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
12421 * config/sparc/sparc.c (output_return): Output the source location of
12422 the insn in the delay slot, if any.
12423 (output_sibcall): Likewise.
12425 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
12428 * common.opt (-static-pie): New alias.
12429 (shared): Negate static-pie.
12430 (-no-pie): Update help text.
12432 (static-pie): New option.
12433 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
12434 -static-pie support.
12435 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
12436 (LINK_EH_SPEC): Likewise.
12437 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12438 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
12439 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
12440 * gcc.c (LINK_COMMAND_SPEC): Likewise.
12441 (init_gcc_specs): Likewise.
12442 (init_spec): Likewise.
12443 (display_help): Update help message for -pie.
12444 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
12447 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
12449 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
12450 (movdi_aarch64): Likewise.
12451 (movti_aarch64): Likewise.
12453 2017-09-12 Simon Wright <simon@pushface.org>
12456 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
12457 calculation of the minor version, always output as 0.
12459 2017-09-12 Jakub Jelinek <jakub@redhat.com>
12462 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
12463 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
12464 on it early, rather than manual conversion late. For
12465 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
12466 instead of performing manual conversion.
12468 2017-09-12 Carl Love <cel@us.ibm.com>
12470 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
12471 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
12473 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12474 VMULOSW): Add definitions.
12475 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12476 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12477 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12478 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12479 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12481 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
12483 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
12485 (movti_aarch64): Likewise.
12486 (movdf_aarch64): Likewise.
12487 (movtf_aarch64): Likewise.
12488 (load_pairdi): Likewise.
12489 (store_pairdi): Likewise.
12490 (load_pairdf): Likewise.
12491 (store_pairdf): Likewise.
12492 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
12493 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
12494 (ldr_got_small_<mode>): Likewise.
12495 (ldr_got_small_28k_<mode>): Likewise.
12496 (ldr_got_tiny): Likewise.
12497 * config/aarch64/iterators.md (ldst_sz): New.
12498 (ldpstp_sz): Likewise.
12499 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
12501 (thunderx_load): Split load_8 to load_16.
12502 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
12504 (thunderx2t99_storepair_basic): Split store_8 to store_16.
12505 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
12506 (xgene1_store_pair): Split store_8 to store_16.
12507 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
12508 (falkor_st_0_st_sd): Split store_8 to store_16.
12510 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
12512 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
12513 and store1/2/3/4 to store_4/8/12/16.
12514 * config/aarch64/aarch64.md: Update for rename.
12515 * config/arm/arm.md: Likewise.: Likewise.
12516 * config/arm/arm.c: Likewise.
12517 * config/arm/thumb1.md: Likewise.
12518 * config/arm/thumb2.md: Likewise.
12519 * config/arm/vfp.md: Likewise.
12520 * config/arm/arm-generic.md: Likewise.
12521 * config/arm/arm1020e.md: Likewise.
12522 * config/arm/arm1026ejs.md: Likewise.
12523 * config/arm/arm1136jfs.md: Likewise.
12524 * config/arm/arm926ejs.md: Likewise.
12525 * config/arm/cortex-a15.md: Likewise.
12526 * config/arm/cortex-a17.md: Likewise.
12527 * config/arm/cortex-a5.md: Likewise.
12528 * config/arm/cortex-a53.md: Likewise.
12529 * config/arm/cortex-a57.md: Likewise.
12530 * config/arm/cortex-a7.md: Likewise.
12531 * config/arm/cortex-a8.md: Likewise.
12532 * config/arm/cortex-a9.md: Likewise.
12533 * config/arm/cortex-m4.md: Likewise.
12534 * config/arm/cortex-m7.md: Likewise.
12535 * config/arm/cortex-r4.md: Likewise.
12536 * config/arm/exynos-m1.md: Likewise.
12537 * config/arm/fa526.md: Likewise.
12538 * config/arm/fa606te.md: Likewise.
12539 * config/arm/fa626te.md: Likewise.
12540 * config/arm/fa726te.md: Likewise.
12541 * config/arm/fmp626.md: Likewise.
12542 * config/arm/iwmmxt.md: Likewise.
12543 * config/arm/ldmstm.md: Likewise.
12544 * config/arm/marvell-pj4.md: Likewise.
12545 * config/arm/xgene1.md: Likewise.
12546 * config/aarch64/thunderx.md: Likewise.
12547 * config/aarch64/thunderx2t99.md: Likewise.
12548 * config/aarch64/falkor.md: Likewise.
12550 2017-09-12 Martin Liska <mliska@suse.cz>
12552 * attribs.c (private_lookup_attribute): New function.
12553 * attribs.h (private_lookup_attribute): Declared here.
12554 (lookup_attribute): Called from this place.
12556 2017-09-12 Richard Biener <rguenther@suse.de>
12558 PR tree-optimization/82157
12559 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
12560 stmts with side-effects.
12562 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12563 Alan Hayward <alan.hayward@arm.com>
12564 David Sherwood <david.sherwood@arm.com>
12566 * target.def (hard_regno_nregs): New hook.
12567 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
12568 * targhooks.h (default_hard_regno_nregs): Declare.
12569 * targhooks.c (default_hard_regno_nregs): New function.
12570 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
12571 (TARGET_HARD_REGNO_NREGS): ...this hook.
12572 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
12573 (CLASS_MAX_NREGS): Likewise.
12574 * doc/tm.texi: Regenerate.
12575 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
12576 instead of HARD_REGNO_NREGS.
12577 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
12578 HARD_REGNO_NREGS in the comment.
12579 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
12580 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
12581 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
12582 Return an unsigned int.
12583 (TARGET_HARD_REGNO_NREGS): Redefine.
12584 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
12585 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
12586 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
12587 (arc_hard_regno_nregs): New function.
12588 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
12589 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
12590 (arm_hard_regno_nregs): New function.
12591 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
12592 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
12593 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
12594 (TARGET_HARD_REGNO_NREGS): Redefine.
12595 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
12596 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
12597 (HARD_REGNO_NREGS): Delete.
12598 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
12599 (cr16_hard_regno_nregs): New function.
12600 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
12601 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
12602 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
12603 (cris_hard_regno_nregs): New function.
12604 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
12605 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
12606 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
12607 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
12608 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
12609 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
12610 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
12611 (frv_hard_regno_nregs): Make static. Take and return an
12613 (frv_class_max_nregs): Remove outdated copy of documentation.
12614 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
12615 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
12616 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
12617 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
12618 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
12619 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
12620 (TARGET_HARD_REGNO_NREGS): Redefine.
12621 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
12622 (CLASS_MAX_NREGS): Update comment.
12623 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
12624 (ia64_hard_regno_nregs): New function.
12625 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
12626 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
12627 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
12628 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
12629 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
12631 (m32c_hard_regno_nregs): Likewise. Make static.
12632 (TARGET_HARD_REGNO_NREGS): Redefine.
12633 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
12634 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
12635 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
12636 (m68k_hard_regno_nregs): New function.
12637 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
12638 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
12639 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
12640 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
12641 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
12642 Take and return an unsigned int.
12643 (TARGET_HARD_REGNO_NREGS): Redefine.
12644 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
12645 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
12646 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
12647 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
12648 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
12649 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
12650 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
12651 (msp430_hard_regno_nregs): Make static. Take and return an
12653 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
12654 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
12655 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
12656 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
12657 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
12658 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
12659 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
12660 (TARGET_HARD_REGNO_NREGS): Redefine.
12661 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
12662 (PA_HARD_REGNO_NREGS): ...this.
12663 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
12664 (PA_HARD_REGNO_NREGS): ...this.
12665 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
12666 (pa_hard_regno_nregs): New function.
12667 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
12668 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
12669 (pdp11_hard_regno_nregs): New function.
12670 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
12671 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
12672 (rs6000_hard_regno_nregs_hook): New function.
12673 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
12674 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
12675 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
12676 Take and return an unsigned int. Move earlier in file.
12677 (TARGET_HARD_REGNO_NREGS): Redefine.
12678 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
12679 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
12680 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
12681 (rl78_hard_regno_nregs): Make static. Take and return an
12683 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
12684 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
12685 (rs6000_hard_regno_nregs_hook): New function.
12686 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
12687 * config/rx/rx.c (rx_hard_regno_nregs): New function.
12688 (TARGET_HARD_REGNO_NREGS): Redefine.
12689 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
12690 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
12691 instead of HARD_REGNO_NREGS.
12692 (s390_hard_regno_nregs): New function.
12693 (s390_hard_regno_mode_ok): Add comment from s390.h.
12694 (TARGET_HARD_REGNO_NREGS): Redefine.
12695 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
12696 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
12697 (sh_hard_regno_nregs): New function.
12698 (sh_pass_in_reg_p): Use it.
12699 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
12700 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
12701 (sparc_hard_regno_nregs): New function.
12702 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
12703 * config/spu/spu.c (spu_hard_regno_nregs): New function.
12704 (spu_function_arg_advance): Use it, supplying a valid register number.
12705 (TARGET_HARD_REGNO_NREGS): Redefine.
12706 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
12707 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
12708 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
12709 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
12710 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
12711 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
12712 (CLASS_MAX_NREGS): Remove copy of old documentation.
12713 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
12714 (visium_hard_regno_nregs): New function.
12715 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
12716 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
12717 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
12718 xtensa_hard_regno_nregs): New function.
12719 * system.h (HARD_REGNO_NREGS): Poison.
12721 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12723 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
12724 hard_regno_nregs instead of HARD_REGNO_NREGS.
12725 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
12726 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
12727 (c6x_expand_epilogue): Likewise.
12728 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
12729 (frv_read_iacc_argument): Likewise.
12730 * config/sh/sh.c: Include regs.h.
12731 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12732 (regs_used): Likewise.
12733 (output_stack_adjust): Likewise.
12734 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
12735 * expmed.c: Include regs.h.
12736 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12737 * ree.c: Include regs.h.
12738 (combine_reaching_defs): Use hard_regno_nregs instead of
12740 (add_removable_extension): Likewise.
12742 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12744 * regs.h (hard_regno_nregs): Turn into a function.
12745 (end_hard_regno): Update accordingly.
12746 * caller-save.c (setup_save_areas): Likewise.
12747 (save_call_clobbered_regs): Likewise.
12748 (replace_reg_with_saved_mem): Likewise.
12749 (insert_restore): Likewise.
12750 (insert_save): Likewise.
12751 * combine.c (can_change_dest_mode): Likewise.
12752 (move_deaths): Likewise.
12753 (distribute_notes): Likewise.
12754 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
12755 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
12756 (rs6000_split_multireg_move): Likewise.
12757 (rs6000_register_move_cost): Likewise.
12758 (rs6000_memory_move_cost): Likewise.
12759 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
12760 (rs6000_split_multireg_move): Likewise.
12761 (rs6000_register_move_cost): Likewise.
12762 (rs6000_memory_move_cost): Likewise.
12763 * cselib.c (cselib_reset_table): Likewise.
12764 (cselib_lookup_1): Likewise.
12765 * emit-rtl.c (set_mode_and_regno): Likewise.
12766 * function.c (aggregate_value_p): Likewise.
12767 * ira-color.c (setup_profitable_hard_regs): Likewise.
12768 (check_hard_reg_p): Likewise.
12769 (calculate_saved_nregs): Likewise.
12770 (assign_hard_reg): Likewise.
12771 (improve_allocation): Likewise.
12772 (calculate_spill_cost): Likewise.
12773 * ira-emit.c (modify_move_list): Likewise.
12774 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
12775 (ira_hard_reg_in_set_p): Likewise.
12776 * ira.c (setup_reg_mode_hard_regset): Likewise.
12777 (clarify_prohibited_class_mode_regs): Likewise.
12778 (check_allocation): Likewise.
12779 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12780 (lra_setup_reg_renumber): Likewise.
12781 (setup_try_hard_regno_pseudos): Likewise.
12782 (spill_for): Likewise.
12783 (assign_hard_regno): Likewise.
12784 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
12785 * lra-constraints.c (in_class_p): Likewise.
12786 (lra_constraint_offset): Likewise.
12787 (simplify_operand_subreg): Likewise.
12788 (lra_constraints): Likewise.
12789 (split_reg): Likewise.
12790 (split_if_necessary): Likewise.
12791 (invariant_p): Likewise.
12792 (inherit_in_ebb): Likewise.
12793 * lra-lives.c (process_bb_lives): Likewise.
12794 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
12795 (get_hard_regs): Likewise.
12796 (do_remat): Likewise.
12797 * lra-spills.c (assign_spill_hard_regs): Likewise.
12798 * mode-switching.c (create_pre_exit): Likewise.
12799 * postreload.c (reload_combine_recognize_pattern): Likewise.
12800 * recog.c (peep2_find_free_register): Likewise.
12801 * regcprop.c (kill_value_regno): Likewise.
12802 (set_value_regno): Likewise.
12803 (copy_value): Likewise.
12804 (maybe_mode_change): Likewise.
12805 (find_oldest_value_reg): Likewise.
12806 (copyprop_hardreg_forward_1): Likewise.
12807 * regrename.c (check_new_reg_p): Likewise.
12808 (regrename_do_replace): Likewise.
12809 * reload.c (push_reload): Likewise.
12810 (combine_reloads): Likewise.
12811 (find_dummy_reload): Likewise.
12812 (operands_match_p): Likewise.
12813 (find_reloads): Likewise.
12814 (find_equiv_reg): Likewise.
12815 (reload_adjust_reg_for_mode): Likewise.
12816 * reload1.c (count_pseudo): Likewise.
12817 (count_spilled_pseudo): Likewise.
12818 (find_reg): Likewise.
12819 (clear_reload_reg_in_use): Likewise.
12820 (free_for_value_p): Likewise.
12821 (allocate_reload_reg): Likewise.
12822 (choose_reload_regs): Likewise.
12823 (reload_adjust_reg_for_temp): Likewise.
12824 (emit_reload_insns): Likewise.
12825 (delete_output_reload): Likewise.
12826 * rtlanal.c (subreg_get_info): Likewise.
12827 * sched-deps.c (sched_analyze_reg): Likewise.
12828 * sel-sched.c (init_regs_for_mode): Likewise.
12829 (mark_unavailable_hard_regs): Likewise.
12830 (choose_best_reg_1): Likewise.
12831 (verify_target_availability): Likewise.
12832 * valtrack.c (dead_debug_insert_temp): Likewise.
12833 * var-tracking.c (track_loc_p): Likewise.
12834 (emit_note_insn_var_location): Likewise.
12835 * varasm.c (make_decl_rtl): Likewise.
12836 * reginfo.c (choose_hard_reg_mode): Likewise.
12837 (init_reg_modes_target): Refer directly to
12838 this_target_regs->x_hard_regno_nregs.
12840 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12842 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
12843 instead of hard_regno_nregs.
12845 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12847 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
12848 end_hard_regno instead of hard_regno_nregs.
12849 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
12850 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
12851 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
12852 * ira-color.c (improve_allocation): Likewise.
12853 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12854 * lra-lives.c (mark_regno_live): Likewise.
12855 (mark_regno_dead): Likewise.
12856 * lra-remat.c (operand_to_remat): Likewise.
12857 * lra.c (collect_non_operand_hard_regs): Likewise.
12858 * postreload.c (reload_combine_note_store): Likewise.
12859 (move2add_valid_value_p): Likewise.
12860 * reload.c (regno_clobbered_p): Likewise.
12862 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12864 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
12866 * config/v850/v850.c (v850_reorg): Likewise.
12867 * reload.c (refers_to_regno_for_reload_p): Likewise.
12868 (find_equiv_reg): Likewise.
12869 * reload1.c (reload_reg_reaches_end_p): Likewise.
12871 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12873 * caller-save.c (add_used_regs): Use REG_NREGS instead of
12875 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
12876 * config/arm/arm.c (output_move_neon): Likewise.
12877 (arm_attr_length_move_neon): Likewise.
12878 (neon_split_vcombine): Likewise.
12879 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
12880 (c6x_mark_reg_written): Likewise.
12881 (c6x_dwarf_register_span): Likewise.
12882 * config/i386/i386.c (ix86_save_reg): Likewise.
12883 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
12884 (rws_access_reg): Likewise.
12885 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
12886 * mode-switching.c (create_pre_exit): Likewise.
12887 * ree.c (combine_reaching_defs): Likewise.
12888 (add_removable_extension): Likewise.
12889 * regcprop.c (find_oldest_value_reg): Likewise.
12890 (copyprop_hardreg_forward_1): Likewise.
12891 * reload.c (reload_inner_reg_of_subreg): Likewise.
12892 (push_reload): Likewise.
12893 (combine_reloads): Likewise.
12894 (find_dummy_reload): Likewise.
12895 (reload_adjust_reg_for_mode): Likewise.
12896 * reload1.c (find_reload_regs): Likewise.
12897 (forget_old_reloads_1): Likewise.
12898 (reload_reg_free_for_value_p): Likewise.
12899 (reload_adjust_reg_for_temp): Likewise.
12900 (emit_reload_insns): Likewise.
12901 (delete_output_reload): Likewise.
12902 * sel-sched.c (choose_best_reg_1): Likewise.
12903 (choose_best_pseudo_reg): Likewise.
12905 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12906 Alan Hayward <alan.hayward@arm.com>
12907 David Sherwood <david.sherwood@arm.com>
12909 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
12910 * target.def (slow_unaligned_access): New hook.
12911 * targhooks.h (default_slow_unaligned_access): Declare.
12912 * targhooks.c (default_slow_unaligned_access): New function.
12913 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
12914 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
12915 * doc/tm.texi: Regenerate.
12916 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
12917 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
12918 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
12920 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
12921 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
12923 (rs6000_slow_unaligned_access): New function.
12924 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
12925 (expand_block_compare): Likewise.
12926 (expand_strn_compare): Likewise.
12927 (rs6000_rtx_costs): Likewise.
12928 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
12929 (riscv_slow_unaligned_access): Likewise.
12930 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
12931 (riscv_slow_unaligned_access_p): ...this and make static.
12932 (riscv_option_override): Update accordingly.
12933 (riscv_slow_unaligned_access): New function.
12934 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
12935 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
12936 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
12937 (rs6000_slow_unaligned_access): New function.
12938 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
12939 (rs6000_rtx_costs): Likewise.
12940 * config/rs6000/rs6000-string.c (expand_block_compare)
12941 (expand_strn_compare): Use targetm.slow_unaligned_access instead
12942 of SLOW_UNALIGNED_ACCESS.
12943 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
12944 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
12945 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
12946 of SLOW_UNALIGNED_ACCESS.
12947 * expmed.c (simple_mem_bitfield_p): Likewise.
12948 * expr.c (alignment_for_piecewise_move): Likewise.
12949 (emit_group_load_1): Likewise.
12950 (emit_group_store): Likewise.
12951 (copy_blkmode_from_reg): Likewise.
12952 (emit_push_insn): Likewise.
12953 (expand_assignment): Likewise.
12954 (store_field): Likewise.
12955 (expand_expr_real_1): Likewise.
12956 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
12957 * lra-constraints.c (simplify_operand_subreg): Likewise.
12958 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12959 * gimple-ssa-store-merging.c: Likewise in block comment at start
12961 * tree-ssa-strlen.c: Include target.h.
12962 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
12963 of SLOW_UNALIGNED_ACCESS.
12964 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
12966 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12968 PR rtl-optimization/82185
12969 * expmed.c (emit_store_flag_int): Only test tem if it has been
12972 2017-09-12 Richard Biener <rguenther@suse.de>
12974 PR middle-end/82149
12975 * match.pd ((FTYPE) N CMP CST): Fix typo.
12977 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
12979 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
12981 (mips_near_type_p): Add 'short_call' attribute as a synonym
12983 * doc/extend.texi (short_call): Document new function attribute.
12985 2017-09-12 Jakub Jelinek <jakub@redhat.com>
12988 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
12989 assertion check that in the condition.
12990 (get_atomic_generic_size): Likewise. Before testing if parameter
12991 has pointer type, if it has array type, call for C++
12992 default_conversion to perform array-to-pointer conversion.
12994 2017-09-12 Richard Biener <rguenther@suse.de>
12996 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
12997 for operations we cannot scalarize.
12999 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
13001 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
13002 vectors heap vectors. Clean up comments.
13003 Make visited_bbs a reference.
13004 (profitable_jump_thread_path): Make GC
13005 vectors heap vectors. Clean up comments.
13007 (convert_and_register_jump_thread_path): Make GC vectors heap
13009 (check_subpath_and_update_thread_path): Same. Clean up comments.
13010 Make visited_bbs a reference.
13011 (handle_phi): Abstract common code to to
13012 register_jump_thread_path_if_profitable.
13013 Rename VAR_BB to DEF_BB.
13015 Make GC vectors heap vectors.
13016 Make visited_bbs a reference.
13017 (handle_assignment): Same.
13018 (register_jump_thread_path_if_profitable): New.
13019 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
13021 Make GC vectors heap vectors. Clean up comments.
13022 Make visited_bbs a reference.
13023 (find_jump_threads_backwards): Make visited_bbs live in the stack.
13024 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
13027 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
13030 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
13031 words of E_DImode object are reachable by xtensa_uimm8x4 access.
13033 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
13035 Revert r251800 and r251799.
13037 2017-09-11 Martin Jambor <mjambor@suse.cz>
13040 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
13041 arguments in advance.
13042 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
13043 use it to find predecessor edges.
13044 (naive_outof_ssa): Collect vector of predecessors.
13046 2017-09-08 Jason Merrill <jason@redhat.com>
13048 PR c++/70029 - ICE with ref-qualifier and -flto
13049 * langhooks.h (struct lang_hooks_for_types): Add
13050 copy_lang_qualifiers.
13051 * attribs.c (build_type_attribute_qual_variant): Use it.
13052 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
13054 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
13055 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
13057 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
13060 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
13061 (*mulsi3_sp64): New instruction.
13062 (mulsi3): New expander.
13064 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
13066 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
13068 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13070 * sancov.c: Include memmodel.h.
13072 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
13075 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
13078 2017-09-07 Carl Love <cel@us.ibm.com>
13080 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
13081 the sldi instruction.
13083 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
13085 * sancov.c: Include tm_p.h.
13087 2017-09-07 Jakub Jelinek <jakub@redhat.com>
13090 * output.h (switch_to_other_text_partition): New declaration.
13091 * varasm.c (switch_to_other_text_partition): New function.
13092 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
13093 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
13094 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
13095 to the other text partition before emitting LCL label and switch back
13096 after emitting the word after it.
13098 2017-09-07 Richard Biener <rguenther@suse.de>
13100 * passes.def (pass_split_crit_edges): Remove instance before PRE.
13101 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
13102 critical edges here, after loop init.
13103 (pass_data_pre): Remove PROP_no_crit_edges flags.
13104 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
13105 for valueization of call args to avoid leaking VN_TOP.
13106 (visit_use): Assert we do not visit default defs.
13107 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
13108 Use error_mark_node to more easily detect leaking VN_TOP.
13109 All default-defs are varying, not VN_TOP. Mark them visited.
13110 (run_scc_vn): Make code match comment.
13112 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13114 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
13115 OPTION_MASK_FLOAT128_KEYWORD.
13116 (POWERPC_MASKS): Likewise.
13117 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
13118 support for the -mfloat128-type option, and make -mfloat128
13119 default on PowerPC Linux systems. Define or undefine
13120 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
13121 Define __float128 to be __ieee128 if IEEE 128-bit support is
13122 enabled, or undefine it.
13123 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
13124 Delete defining __FLOAT128_TYPE__.
13125 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
13126 -mfloat128-type option and make -mfloat128 default on PowerPC
13128 (TARGET_FLOAT128_TYPE): Likewise.
13129 (-mfloat128-type): Likewise.
13130 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13131 Delete the -mfloat128-type option and make -mfloat128 default on
13132 PowerPC Linux systems. Always use __ieee128 to be the keyword for
13133 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
13134 128-bit floating point is enabled. Change tests from using
13135 -mfloat128-type to -mfloat128.
13136 (rs6000_mangle_type): Use the correct mangling for the __float128
13137 type even if normal long double is restricted to 64-bits.
13138 (floatn_mode): Enable the _Float128 type by default on VSX Linux
13140 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
13141 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
13142 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
13144 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
13145 documentation for -mfloat128.
13147 2017-09-06 Olivier Hainque <hainque@adacore.com>
13149 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
13151 2017-09-06 Wish Wu <wishwu007@gmail.com>
13152 Jakub Jelinek <jakub@redhat.com>
13154 * asan.c (initialize_sanitizer_builtins): Add
13155 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
13156 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
13157 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
13158 BT_FN_VOID_UINT64_PTR variables.
13159 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
13160 (BT_FN_VOID_UINT16_UINT16): Likewise.
13161 (BT_FN_VOID_UINT32_UINT32): Likewise.
13162 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
13163 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
13164 (BT_FN_VOID_UINT64_PTR): Likewise.
13165 * common.opt (flag_sanitize_coverage): New variable.
13166 (fsanitize-coverage=trace-pc): Remove.
13167 (fsanitize-coverage=): Add.
13168 * flag-types.h (enum sanitize_coverage_code): New enum.
13169 * fold-const.c (fold_range_test): Disable non-short-circuit
13170 optimization if flag_sanitize_coverage.
13171 (fold_truth_andor): Likewise.
13172 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
13173 * opts.c (COVERAGE_SANITIZER_OPT): Define.
13174 (coverage_sanitizer_opts): New array.
13175 (get_closest_sanitizer_option): Add OPTS argument, handle also
13176 OPT_fsanitize_coverage_.
13177 (parse_sanitizer_options): Adjusted to also handle
13178 OPT_fsanitize_coverage_.
13179 (common_handle_option): Add OPT_fsanitize_coverage_.
13180 * sancov.c (instrument_comparison, instrument_switch): New function.
13181 (sancov_pass): Add trace-cmp support.
13182 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
13183 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
13184 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
13185 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
13186 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
13187 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
13188 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
13189 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
13190 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
13191 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
13193 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13195 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13196 error. Only quit immediately if parsing is complete.
13197 (BEGIN): Initialize fatal_err and parse_done.
13198 (begin fpu, end fpu): Check number of arguments.
13199 (begin arch, end arch): Likewise.
13200 (begin cpu, end cpu): Likewise.
13201 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13202 (optalias): Likewise.
13204 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13206 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13207 * config/arm/arm-isa.h: Delete. Move definitions to ...
13208 * arm-cpus.in: ... here. Use new feature and fgroup values.
13209 * config/arm/arm.c (arm_option_override): Use lower case for feature
13211 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13212 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13213 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13215 (print_isa_bits_for): New function.
13216 (gen_isa): New function.
13217 (gen_comm_data): Use print_isa_bits_for.
13218 (define feature): New keyword.
13219 (define fgroup): New keyword.
13220 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
13221 (arm-isa.h): Add rule to generate file.
13222 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13223 case for feature bit names.
13225 2017-09-06 Richard Biener <rguenther@suse.de>
13227 * tree-ssa-pre.c (NECESSARY): Remove.
13228 (create_expression_by_pieces): Do not touch pass-local flags.
13229 (insert_into_preds_of_block): Likewise.
13230 (do_pre_regular_insertion): Likewise.
13231 (eliminate_insert): Likewise.
13232 (eliminate_dom_walker::before_dom_children): Likewise.
13233 (fini_eliminate): Do not look at inserted_exprs.
13234 (mark_operand_necessary): Remove.
13235 (remove_dead_inserted_code): Replace with simple work-list
13236 algorithm based on inserted_exprs and SSA uses.
13237 (pass_pre::execute): Re-order fini_eliminate and
13238 remove_dead_inserted_code.
13240 2017-09-06 Olivier Hainque <hainque@adacore.com>
13242 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
13243 for VxWorks 7. Adjust surrounding comments.
13245 2017-09-06 Richard Biener <rguenther@suse.de>
13247 * gimple-ssa-strength-reduction.c
13248 (find_candidates_dom_walker::before_dom_children): Also allow
13251 2017-09-06 Richard Biener <rguenther@suse.de>
13253 PR tree-optimization/82108
13254 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
13255 for gap in the non-permutation SLP case.
13257 2017-09-06 Martin Jambor <mjambor@suse.cz>
13259 PR tree-optimization/82078
13260 * tree-sra.c (sort_and_splice_var_accesses): Move call to
13261 add_access_to_work_queue...
13262 (build_accesses_from_assign): ...here.
13263 (propagate_all_subaccesses): Make sure racc is the group
13264 representative, if there is one.
13266 2017-09-06 Jakub Jelinek <jakub@redhat.com>
13268 PR middle-end/82095
13269 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
13272 2017-09-06 Richard Biener <rguenther@suse.de>
13274 * gimple-ssa-strength-reduction.c
13275 (find_candidates_dom_walker::before_doom_children): Use a
13276 type and not a mode check.
13278 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
13281 * config/arm/predicates.md (arm_general_adddi_operand): Create new
13283 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
13285 2017-09-05 Jeff Law <law@redhat.com>
13287 PR tree-optimization/64910
13288 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
13289 cases where we have 3 or more operands.
13291 2017-09-05 Jakub Jelinek <jakub@redhat.com>
13293 PR middle-end/81768
13294 * omp-low.c (lower_omp_for): Recompute tree invariant if
13295 gimple_omp_for_initial/final is ADDR_EXPR.
13297 PR middle-end/81768
13298 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
13299 into gimple val before gimplification fo the COND_EXPR.
13301 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
13303 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
13304 REGION_COPY argument.
13305 (thread_through_all_blocks): Remove unused argument to
13306 duplicate_thread_path.
13308 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13309 Alan Hayward <alan.hayward@arm.com>
13310 David Sherwood <david.sherwood@arm.com>
13312 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
13313 Take a scalar_mode rather than a machine_mode.
13314 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13315 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
13316 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13317 (aarch64_gen_adjusted_ldpstp): Likewise.
13318 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
13320 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13321 Alan Hayward <alan.hayward@arm.com>
13322 David Sherwood <david.sherwood@arm.com>
13324 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
13325 Take a scalar_int_mode instead of a machine_mode.
13326 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13327 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13328 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13329 (aarch64_simd_attr_length_rglist): Delete.
13330 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
13331 a scalar_int_mode instead of a machine_mode.
13332 (aarch64_add_offset): Likewise.
13333 (aarch64_internal_mov_immediate): Likewise
13334 (aarch64_add_constant_internal): Likewise.
13335 (aarch64_add_constant): Likewise.
13336 (aarch64_movw_imm): Likewise.
13337 (aarch64_rtx_arith_op_extract_p): Likewise.
13338 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13339 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13340 Remove assert that the mode isn't a vector.
13341 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13342 (aarch64_expand_mov_immediate): Update calls after above changes.
13343 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
13344 (aarch64_and_bitmask_imm): Check for scalar integer modes.
13345 (aarch64_move_imm): Likewise.
13346 (aarch64_can_const_movi_rtx_p): Likewise.
13347 (aarch64_strip_extend): Likewise.
13348 (aarch64_extr_rtx_p): Likewise.
13349 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
13350 a CONST_INT when the mode parameter is VOIDmode.
13351 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
13353 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13355 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
13356 * stor-layout.c (bitwise_mode_for_mode): Likewise.
13357 (bitwise_type_for_mode): Update accordingly.
13359 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13361 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
13362 * stor-layout.c (mode_for_size_tree): Likewise.
13363 (mode_for_array): Update accordingly.
13364 (layout_decl): Likewise.
13365 (compute_record_mode): Likewise. Only set the mode once.
13367 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13369 * target.def (get_mask_mode): Change return type to opt_mode.
13371 * doc/tm.texi: Regenerate.
13372 * targhooks.h (default_get_mask_mode): Return an opt_mode.
13373 * targhooks.c (default_get_mask_mode): Likewise.
13374 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
13375 * optabs-query.c (can_vec_mask_load_store_p): Update use of
13376 targetm.get_mask_mode.
13377 * tree.c (build_truth_vector_type): Likewise.
13379 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13381 * machmode.h (mode_for_vector): Return an opt_mode.
13382 * stor-layout.c (mode_for_vector): Likewise.
13383 (mode_for_int_vector): Update accordingly.
13384 (layout_type): Likewise.
13385 * config/i386/i386.c (emit_memmov): Likewise.
13386 (ix86_expand_set_or_movmem): Likewise.
13387 (ix86_expand_vector_init): Likewise.
13388 (ix86_get_mask_mode): Likewise.
13389 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
13391 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
13392 * expmed.c (extract_bit_field_1): Likewise.
13393 * expr.c (expand_expr_real_2): Likewise.
13394 * optabs-query.c (can_vec_perm_p): Likewise.
13395 (can_vec_mask_load_store_p): Likewise.
13396 * optabs.c (expand_vec_perm): Likewise.
13397 * targhooks.c (default_get_mask_mode): Likewise.
13398 * tree-vect-stmts.c (vectorizable_store): Likewise.
13399 (vectorizable_load): Likewise.
13400 (get_vectype_for_scalar_type_and_size): Likewise.
13402 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13404 * machmode.h (mode_for_int_vector): New function.
13405 * stor-layout.c (mode_for_int_vector): Likewise.
13406 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
13407 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
13408 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
13409 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
13410 (s390_expand_vcond): Likewise.
13412 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13414 * machmode.h (opt_machine_mode): New type.
13415 (opt_mode<T>): Allow construction from anything that can be
13417 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
13418 (mode_for_size): Return an opt_machine_mode.
13419 * stor-layout.c (mode_for_size): Likewise.
13420 (mode_for_size_tree): Update call accordingly.
13421 (bitwise_mode_for_mode): Likewise.
13422 (make_fract_type): Likewise.
13423 (make_accum_type): Likewise.
13424 * caller-save.c (replace_reg_with_saved_mem): Update call
13426 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13427 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13428 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13429 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13430 * expmed.c (extract_bit_field_1): Likewise.
13431 * reload.c (get_secondary_mem): Likewise.
13432 * varasm.c (assemble_integer): Likewise.
13433 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
13436 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13438 * machmode.h (decimal_float_mode_for_size): New function.
13439 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
13440 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
13441 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
13442 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
13443 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
13444 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
13446 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13448 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
13449 (get_builtin_sync_mode): Likewise.
13450 (expand_ifn_atomic_compare_exchange): Likewise.
13451 (expand_builtin_atomic_clear): Likewise.
13452 (expand_builtin_atomic_test_and_set): Likewise.
13453 (fold_builtin_atomic_always_lock_free): Likewise.
13454 * calls.c (compute_argument_addresses): Likewise.
13455 (emit_library_call_value_1): Likewise.
13456 (store_one_arg): Likewise.
13457 * combine.c (combine_instructions): Likewise.
13458 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
13459 * config/arm/arm.c (arm_function_value): Likewise.
13460 (aapcs_allocate_return_reg): Likewise.
13461 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
13462 * config/i386/i386.c (construct_container): Likewise.
13463 (ix86_gimplify_va_arg): Likewise.
13464 (ix86_expand_sse_cmp): Likewise.
13465 (emit_memmov): Likewise.
13466 (emit_memset): Likewise.
13467 (expand_small_movmem_or_setmem): Likewise.
13468 (ix86_expand_pextr): Likewise.
13469 (ix86_expand_pinsr): Likewise.
13470 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
13471 * config/microblaze/microblaze.c (microblaze_block_move_straight):
13473 * config/mips/mips.c (mips_function_value_1) Likewise.
13474 (mips_block_move_straight): Likewise.
13475 (mips_expand_ins_as_unaligned_store): Likewise.
13476 * config/powerpcspe/powerpcspe.c
13477 (rs6000_darwin64_record_arg_advance_flush): Likewise.
13478 (rs6000_darwin64_record_arg_flush): Likewise.
13479 * config/rs6000/rs6000.c
13480 (rs6000_darwin64_record_arg_advance_flush): Likewise.
13481 (rs6000_darwin64_record_arg_flush): Likewise.
13482 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
13483 (sparc_function_value_1): Likewise.
13484 * config/spu/spu.c (adjust_operand): Likewise.
13485 (spu_emit_branch_or_set): Likewise.
13486 (arith_immediate_p): Likewise.
13487 * emit-rtl.c (gen_lowpart_common): Likewise.
13488 * expr.c (expand_expr_real_1): Likewise.
13489 * function.c (assign_parm_setup_block): Likewise.
13490 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
13491 * reload1.c (alter_reg): Likewise.
13492 * stor-layout.c (mode_for_vector): Likewise.
13493 (layout_type): Likewise.
13495 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13497 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
13498 (spu_convert_move): Likewise.
13499 * lower-subreg.c (resolve_simple_move): Likewise.
13501 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13504 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
13505 define_insn to a define_expand.
13506 (altivec_vsum2sws_direct): New define_insn.
13507 (altivec_vsumsws): Convert from a define_insn to a define_expand.
13509 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
13511 * config/arm/arm.c (arm_option_params_internal): Improve setting of
13514 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
13519 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
13520 realign stack if stack alignment needed is less than incoming
13523 2017-09-05 Marek Polacek <polacek@redhat.com>
13526 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
13529 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
13531 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
13533 2017-09-05 Richard Biener <rguenther@suse.de>
13535 PR tree-optimization/82084
13536 * fold-const.c (can_native_encode_string_p): Handle wide characters.
13538 2017-09-05 Richard Biener <rguenther@suse.de>
13540 PR tree-optimization/82102
13541 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
13543 2017-09-05 Martin Liska <mliska@suse.cz>
13545 PR tree-optimization/82032
13546 * tree-cfg.c (generate_range_test): New function.
13547 * tree-cfg.h (generate_range_test): Declared here.
13548 * tree-cfgcleanup.c (convert_single_case_switch): New function.
13549 (cleanup_control_expr_graph): Use it.
13550 * tree-switch-conversion.c (try_switch_expansion): Remove
13552 (emit_case_nodes): Use generate_range_test.
13554 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
13557 * config/i386/i386.md (*<btsc><mode>_mask): Add
13558 TARGET_USE_BT to insn constraint.
13559 (*btr<mode>_mask): Ditto.
13561 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
13563 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
13564 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
13566 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
13569 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
13570 TARGET_NEON and TARGET_IWMMXT.
13571 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
13572 TARGET_NEON and TARGET_IWMMXT.
13573 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
13575 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
13577 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
13578 (ix86_rewrite_tls_address): Ditto.
13579 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
13580 (ix86_rewrite_tls_address_1): Ditto.
13581 (ix86_rewrite_tls_address): Ditto.
13582 * config/i386/predicates.md (tls_address_pattern): New predicate.
13583 * config/i386/i386.md (TLS address splitter): New splitter.
13585 2017-09-04 Richard Biener <rguenther@suse.de>
13587 PR tree-optimization/82084
13588 * fold-const.h (can_native_encode_string_p): Declare.
13589 * fold-const.c (can_native_encode_string_p): Factor out from ...
13590 (native_encode_string): ... here.
13591 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
13592 vectorizing stores from constants we later cannot handle.
13594 2017-09-04 Marek Polacek <polacek@redhat.com>
13597 * doc/invoke.texi: Update -Wtautological-compare documentation.
13599 2017-09-04 Jeff Law <law@redhat.com>
13601 PR tree-optimization/64910
13602 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
13603 swap the first and last operand if the last is a constant.
13605 2017-09-04 Marek Polacek <polacek@redhat.com>
13608 * convert.c (do_narrow): When sanitizing signed integer overflows,
13609 bail out for signed types.
13610 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
13612 2017-09-04 Richard Biener <rguenther@suse.de>
13614 PR tree-optimization/82060
13615 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13616 Move devirtualization after stmt folding and before EH/AB/noreturn
13617 cleanup to get the stmt refs canonicalized. Use a bool instead
13618 of gimple_modified_p since that doesn't work for NOPs. Schedule
13619 NOPs generated by folding for removal.
13621 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13622 Alan Hayward <alan.hayward@arm.com>
13623 David Sherwood <david.sherwood@arm.com>
13625 * coretypes.h (pad_direction): New enum.
13626 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
13627 (FUNCTION_ARG_PADDING): Likewise.
13628 * target.def (function_arg_padding): New hook.
13629 * targhooks.h (default_function_arg_padding): Declare.
13630 * targhooks.c (default_function_arg_padding): New function.
13631 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
13632 (TARGET_FUNCTION_ARG_PADDING): ...this.
13633 * doc/tm.texi: Regenerate.
13634 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
13635 instead of direction.
13636 (compute_argument_addresses): Likewise.
13637 (load_register_parameters): Likewise.
13638 (emit_library_call_value_1): Likewise.
13639 (store_one_arg): Use targetm.calls.function_arg_padding instead
13640 of FUNCTION_ARG_PADDING.
13641 (must_pass_in_stack_var_size_or_pad): Likewise.
13642 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
13643 (emit_group_store): Likewise.
13644 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
13645 instead of FUNCTION_ARG_PADDING.
13646 (emit_push_insn): Likewise, and propagate enum change throughout
13648 * function.h (direction): Delete.
13649 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
13651 * function.c (assign_parm_find_stack_rtl): Likewise.
13652 (assign_parm_setup_block_p): Likewise.
13653 (assign_parm_setup_block): Likewise.
13654 (gimplify_parameters): Likewise.
13655 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
13656 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
13658 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
13659 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13660 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
13661 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
13662 (aarch64_function_arg_padding): ...this new function.
13663 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
13664 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13665 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
13666 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13667 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
13668 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13669 (arm_pad_arg_upward): Replace with...
13670 (arm_function_arg_padding): ...this new function.
13671 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
13673 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
13674 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
13675 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13676 (ia64_hpux_function_arg_padding): Replace with...
13677 (ia64_function_arg_padding): ...this new function. Use pad_direction
13678 instead of direction. Check for TARGET_HPUX.
13679 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
13680 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13681 (iq2000_function_arg_padding): New function.
13682 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
13683 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
13684 (mips_function_arg_padding): ...this new function.
13685 (mips_pad_reg_upward): Update accordingly.
13686 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13687 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
13688 targetm.calls.function_arg_padding.
13689 (FUNCTION_ARG_PADDING): Delete.
13690 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13691 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
13692 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13693 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
13694 (nios2_block_reg_padding): Return pad_direction instead of direction.
13695 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
13696 instead of direction.
13697 (nios2_function_arg_padding): Likewise. Make static.
13698 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13699 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
13700 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
13701 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
13702 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13703 (pa_function_arg_padding): Make static. Return pad_direction instead
13705 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
13706 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13707 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
13708 instead of direction. Use targetm.calls.function_arg_padding.
13709 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
13710 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13711 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
13712 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
13713 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
13715 (function_arg_padding): Rename to...
13716 (rs6000_function_arg_padding): ...this. Make static. Return
13717 pad_direction instead of direction.
13718 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13719 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
13720 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13721 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
13722 instead of direction. Use targetm.calls.function_arg_padding.
13723 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
13724 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13725 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
13726 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
13727 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13728 (function_arg_padding): Rename to...
13729 (rs6000_function_arg_padding): ...this. Make static. Return
13730 pad_direction instead of direction.
13731 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13732 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
13733 * config/s390/s390.c (s390_function_arg_padding): New function.
13734 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13735 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
13736 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
13737 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13738 (function_arg_padding): Rename to...
13739 (sparc_function_arg_padding): ...this. Make static. Return
13740 pad_direction instead of direction.
13741 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
13742 * config/spu/spu.c (spu_function_arg_padding): New function.
13743 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13744 * system.h (FUNCTION_ARG_PADDING): Poison.
13746 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13747 Alan Hayward <alan.hayward@arm.com>
13748 David Sherwood <david.sherwood@arm.com>
13750 * target.def (modes_tieable_p): New hook.
13751 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
13752 (TARGET_MODES_TIEABLE_P): ...this.
13753 * doc/tm.texi.in: Regenerate.
13754 * hooks.h (hook_bool_mode_mode_true): Declare.
13755 * hooks.c (hook_bool_mode_mode_true): New function.
13756 * combine.c (subst): Use targetm.modes_tieable_p instead of
13758 * dse.c (find_shift_sequence): Likewise.
13759 * expmed.c (extract_low_bits): Likewise.
13760 * lower-subreg.c: Include target.h.
13761 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
13763 * rtlanal.c (rtx_cost): Likewise.
13764 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
13765 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
13766 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
13767 (TARGET_MODES_TIEABLE_P): Redefine.
13768 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
13769 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
13770 (TARGET_MODES_TIEABLE_P): Redefine.
13771 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
13772 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
13773 (arc_modes_tieable_p): New function.
13774 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
13775 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
13776 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
13777 (arm_modes_tieable_p): Make static.
13778 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
13779 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
13780 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
13781 (TARGET_MODES_TIEABLE_P): Redefine.
13782 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
13783 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
13784 (TARGET_MODES_TIEABLE_P): Redefine.
13785 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
13786 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
13787 (cr16_modes_tieable_p): New function.
13788 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
13789 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
13790 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
13791 (TRULY_NOOP_TRUNCATION): Update comment.
13792 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
13793 (TRULY_NOOP_TRUNCATION): Update comment.
13794 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
13795 (frv_modes_tieable_p): New function.
13796 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
13797 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
13798 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
13799 (TARGET_MODES_TIEABLE_P): Redefine.
13800 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
13801 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
13802 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
13803 (TARGET_MODES_TIEABLE_P): Redefine.
13804 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
13805 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
13806 (ia64_modes_tieable_p): New function.
13807 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
13808 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
13809 (iq2000_modes_tieable_p): New function.
13810 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
13811 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
13812 (lm32_modes_tieable_p): New function.
13813 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
13814 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
13815 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
13816 (TARGET_MODES_TIEABLE_P): Redefine.
13817 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
13818 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
13819 (m32r_modes_tieable_p): New function.
13820 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
13821 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
13822 (m68k_modes_tieable_p): New function.
13823 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
13824 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
13825 (mcore_modes_tieable_p): New function.
13826 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
13827 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
13829 (TARGET_MODES_TIEABLE_P): Redefine.
13830 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
13831 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
13832 * config/mips/mips.c (mips_modes_tieable_p): Make static.
13833 (TARGET_MODES_TIEABLE_P): Redefine.
13834 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
13835 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
13836 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
13837 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
13838 (mn10300_modes_tieable_p): ...this and make static.
13839 (TARGET_MODES_TIEABLE_P): Redefine.
13840 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
13841 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
13842 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
13843 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
13844 (msp430_modes_tieable_p): Make static.
13845 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
13846 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
13847 (TARGET_MODES_TIEABLE_P): Redefine.
13848 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
13849 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
13850 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
13851 (TARGET_MODES_TIEABLE_P): Redefine.
13852 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
13853 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
13854 * config/pa/pa.c (pa_modes_tieable_p): Make static.
13855 (TARGET_MODES_TIEABLE_P): Redefine.
13856 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
13857 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
13858 (pdp11_modes_tieable_p): New function.
13859 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
13860 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
13861 (rs6000_modes_tieable_p): New function.
13862 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
13863 * config/powerpcspe/powerpcspe.md: Update comment.
13864 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
13865 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
13866 (TARGET_MODES_TIEABLE_P): Redefine.
13867 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
13868 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
13869 (rl78_modes_tieable_p): New function.
13870 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
13871 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
13872 (rs6000_modes_tieable_p): New function.
13873 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
13874 * config/rs6000/rs6000.md: Update comment.
13875 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
13876 * config/rx/rx.c (rx_modes_tieable_p): New function.
13877 (TARGET_MODES_TIEABLE_P): Redefine.
13878 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
13879 * config/s390/s390.c (s390_modes_tieable_p): New function.
13880 (TARGET_MODES_TIEABLE_P): Redefine.
13881 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
13882 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
13883 (sh_modes_tieable_p): New function.
13884 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
13885 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
13886 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
13887 (sparc_modes_tieable_p): Make static.
13888 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
13889 * config/spu/spu.c (spu_modes_tieable_p): New function.
13890 (TARGET_MODES_TIEABLE_P): Redefine.
13891 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
13892 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
13893 (TARGET_MODES_TIEABLE_P): Redefine.
13894 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
13895 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
13896 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
13897 * config/v850/v850.c (v850_modes_tieable_p): New function.
13898 (TARGET_MODES_TIEABLE_P): Redefine.
13899 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
13900 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
13901 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
13902 (visium_modes_tieable_p): New function.
13903 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
13904 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
13905 (xtensa_modes_tieable_p): New function.
13906 * system.h (MODES_TIEABLE_P): Poison.
13908 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13909 Alan Hayward <alan.hayward@arm.com>
13910 David Sherwood <david.sherwood@arm.com>
13912 * target.def (hard_regno_mode_ok): New hook.
13913 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
13914 (TARGET_HARD_REGNO_MODE_OK): ...this.
13915 * doc/tm.texi.in: Regenerate.
13916 * hooks.h (hook_bool_uint_mode_true): Declare.
13917 * hooks.c (hook_bool_uint_mode_true): New function.
13918 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
13919 HARD_REGNO_MODE_OK.
13920 * genpreds.c (write_insn_preds_c): Add an include of target.h.
13921 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
13922 instead of HARD_REGNO_MODE_OK.
13923 * caller-save.c: Include target.h.
13924 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
13925 HARD_REGNO_MODE_OK.
13926 * combine.c (can_combine_p): Likewise.
13927 (combinable_i3pat): Likewise.
13928 (can_change_dest_mode): Likewise.
13929 * expr.c (init_expr_target): Likewise.
13930 (convert_move): Likewise.
13931 (convert_modes): Likewise.
13932 * ira.c (setup_prohibited_class_mode_regs): Likewise.
13933 (setup_prohibited_mode_move_regs): Likewise.
13934 * ira.h (target_ira): Likewise.
13935 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13936 * lra-constraints.c (process_alt_operands): Likewise.
13937 (split_reg): Likewise.
13938 * recog.c (peep2_find_free_register): Likewise.
13939 * ree.c (combine_reaching_defs): Likewise.
13940 * regcprop.c (maybe_mode_change): Likewise.
13941 * reginfo.c (init_reg_sets_1): Likewise.
13942 (choose_hard_reg_mode): Likewise.
13943 (simplifiable_subregs): Likewise.
13944 * regrename.c (check_new_reg_p): Likewise.
13945 * reload.c (find_valid_class): Likewise.
13946 (find_valid_class_1): Likewise.
13947 (reload_inner_reg_of_subreg): Likewise.
13948 (push_reload): Likewise.
13949 (combine_reloads): Likewise.
13950 (find_dummy_reload): Likewise.
13951 (find_reloads): Likewise.
13952 * reload1.c (find_reg): Likewise.
13953 (set_reload_reg): Likewise.
13954 (allocate_reload_reg): Likewise.
13955 (choose_reload_regs): Likewise.
13956 (reload_adjust_reg_for_temp): Likewise.
13957 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
13958 (simplify_subreg_regno): Likewise.
13959 * sel-sched.c (init_regs_for_mode): Likewise.
13960 * varasm.c (make_decl_rtl): Likewise.
13961 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
13962 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
13963 HARD_REGNO_MODE_OK.
13964 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
13965 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
13966 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13967 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
13968 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
13969 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13970 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
13971 (arc_mode_class): Delete.
13972 (HARD_REGNO_MODE_OK): Delete.
13973 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13974 (arc_hard_regno_mode_ok): Rename old array to...
13975 (arc_hard_regno_mode_ok_modes): ...this.
13976 (arc_conditional_register_usage): Update accordingly.
13977 (arc_mode_class): Make static.
13978 (arc_hard_regno_mode_ok): New function.
13979 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
13980 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
13981 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13982 (arm_hard_regno_mode_ok): Make static.
13983 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
13984 HARD_REGNO_MODE_OK.
13985 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
13986 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
13987 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
13989 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13990 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
13991 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
13992 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
13993 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
13994 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13995 * config/bfin/predicates.md (valid_reg_operand): Use
13996 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
13997 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
13998 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
13999 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14000 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
14001 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
14002 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14003 (cr16_hard_regno_mode_ok): Make static and return a bool.
14004 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
14005 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14006 (cris_hard_regno_mode_ok): New function.
14007 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
14008 (epiphany_mode_class): Delete.
14009 (HARD_REGNO_MODE_OK): Delete.
14010 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
14011 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14012 (hard_regno_mode_ok): Rename to...
14013 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
14014 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
14015 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
14016 HARD_REGNO_MODE_OK.
14017 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
14018 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
14019 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14020 (frv_hard_regno_mode_ok): Make static and return a bool.
14021 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
14022 HARD_REGNO_MODE_OK.
14023 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
14024 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
14025 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
14026 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
14028 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14029 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
14030 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
14031 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
14033 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14034 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
14035 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14036 (ia64_hard_regno_mode_ok): New function.
14037 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
14038 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14039 (iq2000_hard_regno_mode_ok): New function.
14040 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
14041 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14042 (lm32_hard_regno_mode_ok): New function.
14043 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
14044 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
14045 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
14046 instead of HARD_REGNO_MODE_OK.
14047 (m32c_hard_regno_ok): Rename to...
14048 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
14049 (m32c_cannot_change_mode_class): Update accordingly.
14050 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14051 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
14052 (m32r_mode_class): Delete.
14053 (HARD_REGNO_MODE_OK): Delete.
14054 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14055 (m32r_hard_regno_mode_ok): Rename to...
14056 (m32r_hard_regno_modes): ...this.
14057 (m32r_mode_class): Make static.
14058 (m32r_hard_regno_mode_ok): New function.
14059 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
14060 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
14061 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14062 (m68k_hard_regno_mode_ok): Make static.
14063 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
14064 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14065 (mcore_hard_regno_mode_ok): New function.
14066 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
14067 (HARD_REGNO_MODE_OK): Delete.
14068 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
14070 (microblaze_hard_regno_mode_ok_p): ...this and make static.
14071 (microblaze_hard_regno_mode_ok): New function.
14072 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14073 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
14074 (mips_hard_regno_mode_ok): Delete.
14075 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
14076 (mips_hard_regno_mode_ok_p): ...this and make static.
14077 (mips_hard_regno_mode_ok_p): Rename to...
14078 (mips_hard_regno_mode_ok_uncached): ...this.
14079 (mips_hard_regno_mode_ok): New function.
14080 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
14081 of HARD_REGNO_MODE_OK.
14082 (mips_option_override): Update after above name changes.
14083 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14084 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
14085 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
14086 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
14087 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
14088 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14089 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
14090 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
14091 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
14092 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14093 (msp430_hard_regno_mode_ok): Make static and return a bool.
14094 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
14095 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
14096 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
14098 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14099 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
14100 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
14101 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
14102 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
14103 (PA_HARD_REGNO_MODE_OK): ...this
14104 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
14105 (PA_HARD_REGNO_MODE_OK): ...this.
14106 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14107 (pa_hard_regno_mode_ok): New function.
14108 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
14109 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14110 (pdp11_hard_regno_mode_ok): New function.
14111 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
14112 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
14114 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
14116 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14117 (rs6000_hard_regno_mode_ok): Rename to...
14118 (rs6000_hard_regno_mode_ok_uncached): ...this.
14119 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14120 (rs6000_hard_regno_mode_ok): New function.
14121 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
14122 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
14123 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
14124 (riscv_hard_regno_mode_ok): ...this and make static.
14125 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14126 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
14127 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
14128 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14129 (rl78_hard_regno_mode_ok): Make static and return bool.
14130 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
14131 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
14133 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
14134 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14135 (rs6000_hard_regno_mode_ok): Rename to...
14136 (rs6000_hard_regno_mode_ok_uncached): ...this.
14137 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14138 (rs6000_hard_regno_mode_ok): New function.
14139 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
14140 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
14141 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14142 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
14143 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
14144 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
14145 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14146 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
14147 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
14148 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14149 (sh_hard_regno_mode_ok): Make static.
14150 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
14151 instead of HARD_REGNO_MODE_OK.
14152 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
14153 (sparc_mode_class): Delete.
14154 (HARD_REGNO_MODE_OK): Delete.
14155 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14156 (hard_regno_mode_classes): Make static.
14157 (sparc_mode_class): Likewise.
14158 (sparc_hard_regno_mode_ok): New function.
14159 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
14160 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
14161 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
14163 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14164 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
14165 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
14166 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
14167 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
14168 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14169 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
14170 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
14171 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14172 (visium_hard_regno_mode_ok): New function.
14173 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
14174 instead of HARD_REGNO_MODE_OK.
14175 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
14176 (HARD_REGNO_MODE_OK): Delete.
14177 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
14178 (xtensa_hard_regno_mode_ok_p): ...this and make static.
14179 (xtensa_option_override): Update accordingly.
14180 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14181 (xtensa_hard_regno_mode_ok): New function.
14182 * system.h (HARD_REGNO_MODE_OK): Poison.
14184 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14185 Alan Hayward <alan.hayward@arm.com>
14186 David Sherwood <david.sherwood@arm.com>
14188 * target.def (hard_regno_call_part_clobbered): New hook.
14189 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
14190 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
14191 * doc/tm.texi: Regenerate.
14192 * hooks.h (hook_bool_uint_mode_false): Declare.
14193 * hooks.c (hook_bool_uint_mode_false): New function.
14194 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14195 * cselib.c (cselib_process_insn): Use
14196 targetm.hard_regno_call_part_clobbered instead of
14197 HARD_REGNO_CALL_PART_CLOBBERED.
14198 * ira-conflicts.c (ira_build_conflicts): Likewise.
14199 * ira-costs.c (ira_tune_allocno_costs): Likewise.
14200 * lra-constraints.c (need_for_call_save_p): Likewise.
14201 * lra-lives.c: Include target.h.
14202 (check_pseudos_live_through_calls): Use
14203 targetm.hard_regno_call_part_clobbered instead of
14204 HARD_REGNO_CALL_PART_CLOBBERED.
14205 * regcprop.c: Include target.h.
14206 (copyprop_hardreg_forward_1): Use
14207 targetm.hard_regno_call_part_clobbered instead of
14208 HARD_REGNO_CALL_PART_CLOBBERED.
14209 * reginfo.c (choose_hard_reg_mode): Likewise.
14210 * regrename.c (check_new_reg_p): Likewise.
14211 * reload.c (find_equiv_reg): Likewise.
14212 * reload1.c (emit_reload_insns): Likewise.
14213 * sched-deps.c (deps_analyze_insn): Likewise.
14214 * sel-sched.c (init_regs_for_mode): Likewise.
14215 (mark_unavailable_hard_regs): Likewise.
14216 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
14217 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14218 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
14220 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14221 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14222 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
14224 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
14226 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14227 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14228 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
14230 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14231 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14232 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
14234 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14235 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
14237 * config/powerpcspe/powerpcspe.c
14238 (rs6000_hard_regno_call_part_clobbered): New function.
14239 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14240 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14241 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
14243 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14244 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14245 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
14247 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14248 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14249 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
14251 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14252 Alan Hayward <alan.hayward@arm.com>
14253 David Sherwood <david.sherwood@arm.com>
14255 * rtl.h (subreg_memory_offset): Declare.
14256 * emit-rtl.c (subreg_memory_offset): New function.
14257 * expmed.c (store_bit_field_1): Use it.
14258 * expr.c (undefined_operand_subword_p): Likewise.
14259 * simplify-rtx.c (simplify_subreg): Likewise.
14261 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
14263 PR rtl-optimization/57448
14266 * optabs.c (expand_atomic_load): Place compiler memory barriers if
14267 using atomic_load pattern.
14268 (expand_atomic_store): Likewise.
14270 2017-09-04 Jakub Jelinek <jakub@redhat.com>
14273 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
14274 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
14275 handling. Use replace_call_with_value with NULL instead of
14276 gsi_replace, unlink_stmt_vdef and release_defs.
14278 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
14281 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
14283 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14286 * rtl.h (emit_library_call_value_1): Declare.
14287 (emit_library_call): Replace declaration with a series of overloads.
14288 Remove the parameter count argument.
14289 (emit_library_call_value): Likewise.
14290 * calls.c (emit_library_call_value_1): Make global. Replace varargs
14291 with an "rtx_mode_t *".
14292 (emit_library_call_value): Delete.
14293 (emit_library_call): Likewise.
14294 * asan.c (asan_emit_stack_protection): Update calls accordingly.
14295 (asan_emit_allocas_unpoison): Likewise.
14296 * builtins.c (expand_builtin_powi): Likewise.
14297 (expand_asan_emit_allocas_unpoison): Likewise.
14298 * cfgexpand.c (expand_main_function): Likewise.
14299 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
14300 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
14301 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
14302 * config/arm/arm.c (arm_trampoline_init): Likewise.
14303 (arm_call_tls_get_addr): Likewise.
14304 (arm_expand_divmod_libfunc): Likewise.
14305 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
14306 (smulsi3_highpart): Likewise.
14307 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
14308 (c6x_expand_compare): Likewise.
14309 (c6x_expand_movmem): Likewise.
14310 * config/frv/frv.c (frv_trampoline_init): Likewise.
14311 * config/i386/i386.c (ix86_trampoline_init): Likewise.
14312 (ix86_expand_divmod_libfunc): Likewise.
14313 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
14314 (ia64_expand_compare): Likewise.
14315 (ia64_profile_hook): Likewise.
14316 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
14317 (nonlocal_goto): Likewise.
14318 (restore_stack_nonlocal): Likewise.
14319 * config/m32r/m32r.c (block_move_call): Likewise.
14320 (m32r_trampoline_init): Likewise.
14321 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
14322 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
14323 (m68k_call_m68k_read_tp): Likewise.
14324 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
14325 (microblaze_expand_divide): Likewise.
14326 * config/mips/mips.h (mips_args): Likewise.
14327 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
14328 (MIPS_ICACHE_SYNC): Likewise.
14329 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
14330 (nios2_trampoline_init): Likewise.
14331 * config/pa/pa.c (hppa_tls_call): Likewise.
14332 (pa_trampoline_init): Likewise.
14333 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
14334 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
14335 (expand_strn_compare): Likewise.
14336 (rs6000_generate_compare): Likewise.
14337 (rs6000_expand_float128_convert): Likewise.
14338 (output_profile_hook): Likewise.
14339 (rs6000_trampoline_init): Likewise.
14340 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
14341 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
14342 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
14343 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
14344 (rs6000_generate_compare): Likewise.
14345 (rs6000_expand_float128_convert): Likewise.
14346 (output_profile_hook): Likewise.
14347 (rs6000_trampoline_init): Likewise.
14348 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
14349 * config/sh/sh.c (sh_trampoline_init): Likewise.
14350 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
14351 (sparc_emit_float_lib_cmp): Likewise.
14352 (sparc32_initialize_trampoline): Likewise.
14353 (sparc64_initialize_trampoline): Likewise.
14354 (sparc_profile_hook): Likewise.
14355 * config/spu/spu.c (ea_load_store): Likewise.
14356 * config/spu/spu.md (floatunssidf2): Likewise.
14357 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
14358 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
14359 * config/visium/visium.c (expand_block_move_4): Likewise.
14360 (expand_block_move_2): Likewise.
14361 (expand_block_move_1): Likewise.
14362 (expand_block_set_4): Likewise.
14363 (expand_block_set_2): Likewise.
14364 (expand_block_set_1): Likewise.
14365 (visium_trampoline_init): Likewise.
14366 (visium_profile_hook): Likewise.
14367 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
14368 (xtensa_setup_frame_addresses): Likewise.
14369 (xtensa_trampoline_init): Likewise.
14370 * except.c (sjlj_emit_function_enter): Likewise.
14371 (sjlj_emit_function_exit): Likewise.
14372 * explow.c (allocate_dynamic_stack_space): Likewise.
14373 (probe_stack_range): Likewise.
14374 * expr.c (convert_mode_scalar): Likewise.
14375 * optabs.c (expand_binop): Likewise.
14376 (expand_twoval_binop_libfunc): Likewise.
14377 (expand_unop): Likewise.
14378 (prepare_cmp_insn): Likewise.
14379 (prepare_float_lib_cmp): Likewise.
14380 (expand_float): Likewise.
14381 (expand_fix): Likewise.
14382 (expand_fixed_convert): Likewise.
14383 (maybe_emit_sync_lock_test_and_set): Likewise.
14384 (expand_atomic_compare_and_swap): Likewise.
14385 (expand_mem_thread_fence): Likewise.
14386 (expand_atomic_fetch_op): Likewise.
14388 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
14390 * doc/generic.texi (OpenACC): Adjust URL.
14391 * doc/invoke.texi (C Dialect Options): Ditto.
14393 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
14395 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
14396 predicate for operand 1. Add (m,<S>) constraint.
14397 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
14398 Prevent memory operand 1 with register operand 2.
14400 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
14402 PR rtl-optimization/82024
14403 * combine.c (try_combine): If the combination result is a PARALLEL,
14404 and we only need to retain the SET in there that would be placed
14405 at I2, check that we can place that at I3 instead, before doing so.
14407 2017-09-01 Jakub Jelinek <jakub@redhat.com>
14410 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
14412 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
14413 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
14416 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
14417 Jeff Law <law@redhat.com>
14419 * varasm.c (bss_initializer_p): Do not put constants into .bss
14420 (categorize_decl_for_section): Handle bss_initializer_p returning
14421 false when DECL_INITIAL is NULL.
14423 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14426 * config/s390/s390.c (s390_can_inline_p): New function.
14428 2017-09-01 Jeff Law <law@redhat.com>
14430 PR tree-optimization/82052
14431 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
14432 Always initialize the returned slot after a hash table miss
14433 when INSERT is true.
14435 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
14437 * config/s390/s390.md (mem_signal_fence): Remove.
14438 * doc/md.texi (mem_signal_fence): Remove.
14439 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
14441 * target-insns.def (mem_signal_fence): Remove.
14443 2017-09-01 Jakub Jelinek <jakub@redhat.com>
14446 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
14449 * asan.c (create_odr_indicator): Strip name encoding from assembler
14450 name before appending it after __odr_asan_.
14452 2017-09-01 Martin Liska <mliska@suse.cz>
14454 PR tree-optimization/82059
14455 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
14456 frequency only when an edge is redirected.
14458 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14460 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
14461 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
14462 (arc_conditional_register_usage): Remove ARC600 lp_count
14464 (arc_file_start): Emit Tag_ARC_CPU_variation.
14465 (arc_can_use_doloop_p): New conditions to use ZOLs.
14466 (hwloop_fail): New function.
14467 (hwloop_optimize): Likewise.
14468 (hwloop_pattern_reg): Likewise.
14469 (arc_doloop_hooks): New struct, to be used with reorg_loops.
14470 (arc_reorg_loops): New function, calls reorg_loops.
14471 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
14472 (arc600_corereg_hazard): Remove ZOL checking, case handled by
14474 (arc_loop_hazard): Remove function, functionality moved into
14476 (arc_hazard): Remove arc_loop_hazard call.
14477 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
14478 into hwloop_optimize.
14479 (arc_label_align): Remove ZOL handling.
14480 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
14481 * config/arc/arc.md (doloop_begin): Remove pattern.
14482 (doloop_begin_i): Likewise.
14483 (doloop_end_i): Likewise.
14484 (doloop_fallback): Likewise.
14485 (doloop_fallback_m): Likewise.
14486 (doloop_end): Reimplement expand.
14487 (arc_lp): New pattern for LP instruction.
14488 (loop_end): New pattern.
14489 (loop_fail): Likewise.
14490 (decrement_and_branch_until_zero): Likewise.
14491 * config/arc/arc.opt (mlpc-width): New option.
14492 * doc/invoke.texi (mlpc-width): Document option.
14494 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14496 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
14497 (arc_ccfsm_advance): Fix checking for delay slots.
14498 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
14500 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14502 * config/arc/arc.md (movqi_insn): Add stores to save constant long
14504 (movhi_insn): Update store instruction constraint which are saving
14505 6-bit short immediates.
14506 (movsi_insn): Consider also short scaled load operations.
14507 (zero_extendhisi2_i): Use Usd constraint instead of T.
14508 (extendhisi2_i): Add q constraint.
14509 (arc_clzsi2): Add type and length attributes.
14510 (arc_ctzsi2): Likewise.
14511 * config/arc/constraints.md (Usc): Update constraint, the
14512 assembler can parse two relocations for a single instruction.
14514 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14516 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
14517 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
14519 2017-08-31 Olivier Hainque <hainque@adacore.com>
14521 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
14522 match as powerpc-wrs-vxworks*.
14524 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
14526 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
14527 register constraint for by-element operand.
14528 (aarch64_mls_elt_merge<mode>): Likewise.
14530 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14532 * config/arc/arc.c (arc_can_follow_jump): Check for short
14535 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14537 * config.gcc: Use g.opt for arc.
14538 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
14539 functionality moved to ...
14540 (legitimate_scaled_address_p): New function, ...here.
14541 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
14542 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
14543 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
14545 (arc_override_options): Handle G option.
14546 (arc_output_pic_addr_const): Correct function definition.
14547 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
14548 (arc_decl_anon_ns_mem_p): Delete.
14549 (arc_in_small_data_p): Overhaul this function to take into
14550 consideration the value given via G option.
14551 (arc_rewrite_small_data_1): Renamed and corrected old
14552 arc_rewrite_small_data function.
14553 (arc_rewrite_small_data): New function.
14554 (small_data_pattern): Don't use pic_offset_table_rtx.
14555 * config/arc/arc.h (CC1_SPEC): Recognize G option.
14556 * config/arc/simdext.md (movmisalignv2hi): Use
14557 prepare_move_operands function.
14559 (movmisalign*): Likewise.
14560 * doc/invoke.texi (ARC options): Document -G option.
14562 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14564 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
14566 * config/arc/arc.c (arc_print_operand): Output scalled address for
14567 sdata whenever is possible.
14568 (arc_in_small_data_p): Allow sdata for 64bit datum when double
14569 load/stores are available.
14570 (compact_sda_memory_operand): Check for the alignment required by
14571 code density instructions.
14572 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
14574 * config/arc/constraints.md (Usd): Update constraint.
14575 (Us0): New constraint.
14576 (Usc): Update constraint.
14578 2017-08-31 Richard Biener <rguenther@suse.de>
14580 PR middle-end/82054
14581 * dwarf2out.c (dwarf2out_early_global_decl): Process each
14582 function only once.
14584 2017-08-31 Tamar Christina <tamar.christina@arm.com>
14586 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
14587 Resize type_signature.
14589 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14590 Alan Hayward <alan.hayward@arm.com>
14591 David Sherwood <david.sherwood@arm.com>
14593 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
14594 subregs whose inner modes can be stored in GPRs.
14595 (aarch64_classify_index): Likewise.
14597 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14598 Alan Hayward <alan.hayward@arm.com>
14599 David Sherwood <david.sherwood@arm.com>
14601 * config/aarch64/iterators.md (V_cmp_result): Rename to...
14602 (V_INT_EQUIV): ...this.
14603 (v_cmp_result): Rename to...
14604 (v_int_equiv): ...this.
14605 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
14606 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
14607 (copysign<mode>3): Likewise.
14608 (aarch64_simd_bsl<mode>_internal): Likewise.
14609 (aarch64_simd_bsl<mode>): Likewise.
14610 (vec_cmp<mode><mode>): Likewise.
14611 (vcond<mode><mode>): Likewise.
14612 (vcond<v_cmp_mixed><mode>): Likewise.
14613 (vcondu<mode><v_cmp_mixed>): Likewise.
14614 (aarch64_cm<optab><mode>): Likewise.
14615 (aarch64_cmtst<mode>): Likewise.
14616 (aarch64_fac<optab><mode>): Likewise.
14617 (vec_perm_const<mode>): Likewise.
14618 (vcond_mask_<mode><v_cmp_result>): Rename to...
14619 (vcond_mask_<mode><v_int_equiv>): ...this.
14620 (vec_cmp<mode><v_cmp_result>): Rename to...
14621 (vec_cmp<mode><v_int_equiv>): ...this.
14623 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14624 Alan Hayward <alan.hayward@arm.com>
14625 David Sherwood <david.sherwood@arm.com>
14627 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
14629 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
14630 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
14631 (UNSPEC_LD4_DREG): New unspecs.
14632 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
14633 (aarch64_ld2<mode>_dreg_be): Replace with...
14634 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
14636 (aarch64_ld3<mode>_dreg_le)
14637 (aarch64_ld3<mode>_dreg_be): Replace with...
14638 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
14640 (aarch64_ld4<mode>_dreg_le)
14641 (aarch64_ld4<mode>_dreg_be): Replace with...
14642 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
14645 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14647 PR tree-optimization/81987
14648 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
14649 insert an initializer in a location not dominated by the stride
14652 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
14654 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
14655 on the entire header of the finally block in the fallthru case.
14657 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
14659 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
14661 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
14663 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
14664 rs6000_emit_move_from_cr and call renamed function.
14665 (rs6000_emit_prologue): Call renamed functions.
14666 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
14667 movesi_from_cr, remove volatile CRs.
14669 2017-08-30 Jon Beniston <jon@beniston.com>
14670 Richard Biener <rguenther@suse.de>
14672 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
14673 of VECTOR_MODE_P check.
14674 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
14675 element vector types.
14677 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14679 * df.h (df_read_modify_subreg_p): Remove in favor of...
14680 * rtl.h (read_modify_subreg_p): ...this new function. Take a
14681 const_rtx instead of an rtx.
14682 * cprop.c (local_cprop_find_used_regs): Update accordingly.
14683 * df-problems.c (df_word_lr_mark_ref): Likewise.
14684 * ira-lives.c (mark_pseudo_reg_live): Likewise.
14685 (mark_pseudo_reg_dead): Likewise.
14686 (mark_ref_dead): Likewise.
14687 * reginfo.c (init_subregs_of_mode): Likewise.
14688 * sched-deps.c (sched_analyze_1): Likewise.
14689 * df-scan.c (df_def_record_1): Likewise.
14690 (df_uses_record): Likewise.
14691 (df_read_modify_subreg_p): Remove in favor of...
14692 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
14693 const_rtx instead of an rtx.
14695 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14696 Alan Hayward <alan.hayward@arm.com>
14697 David Sherwood <david.sherwood@arm.com>
14699 * rtl.h (partial_subreg_p): New function.
14700 * caller-save.c (save_call_clobbered_regs): Use it.
14701 * calls.c (expand_call): Likewise.
14702 * combine.c (combinable_i3pat): Likewise.
14703 (simplify_set): Likewise.
14704 (make_extraction): Likewise.
14705 (make_compound_operation_int): Likewise.
14706 (gen_lowpart_or_truncate): Likewise.
14707 (force_to_mode): Likewise.
14708 (make_field_assignment): Likewise.
14709 (reg_truncated_to_mode): Likewise.
14710 (record_truncated_value): Likewise.
14711 (move_deaths): Likewise.
14712 * cse.c (record_jump_cond): Likewise.
14713 (cse_insn): Likewise.
14714 * cselib.c (cselib_lookup_1): Likewise.
14715 * expmed.c (extract_bit_field_using_extv): Likewise.
14716 * function.c (assign_parm_setup_reg): Likewise.
14717 * ifcvt.c (noce_convert_multiple_sets): Likewise.
14718 * ira-build.c (create_insn_allocnos): Likewise.
14719 * lra-coalesce.c (merge_pseudos): Likewise.
14720 * lra-constraints.c (match_reload): Likewise.
14721 (simplify_operand_subreg): Likewise.
14722 (curr_insn_transform): Likewise.
14723 * lra-lives.c (process_bb_lives): Likewise.
14724 * lra.c (new_insn_reg): Likewise.
14725 (lra_substitute_pseudo): Likewise.
14726 * regcprop.c (mode_change_ok): Likewise.
14727 (maybe_mode_change): Likewise.
14728 (copyprop_hardreg_forward_1): Likewise.
14729 * reload.c (push_reload): Likewise.
14730 (find_reloads): Likewise.
14731 (find_reloads_subreg_address): Likewise.
14732 * reload1.c (alter_reg): Likewise.
14733 (eliminate_regs_1): Likewise.
14734 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14736 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
14738 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
14739 back to if statements, including unpack.
14741 2017-08-30 Martin Liska <mliska@suse.cz>
14743 PR inline-asm/82001
14744 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
14746 (func_checker::compare_asm_inputs_outputs): ... this function.
14747 (func_checker::compare_gimple_asm): Use the function to compare
14748 also ASM constrains.
14749 * ipa-icf-gimple.h: Rename the function.
14751 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14752 Alan Hayward <alan.hayward@arm.com>
14753 David Sherwood <david.sherwood@arm.com>
14755 * coretypes.h (complex_mode): New type.
14756 * gdbhooks.py (build_pretty_printer): Handle it.
14757 * machmode.h (complex_mode): New class.
14758 (complex_mode::includes_p): New function.
14759 (is_complex_int_mode): Likewise.
14760 (is_complex_float_mode): Likewise.
14761 * genmodes.c (get_mode_class): Handle complex mode classes.
14762 * function.c (expand_function_end): Use is_complex_int_mode.
14764 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14765 Alan Hayward <alan.hayward@arm.com>
14766 David Sherwood <david.sherwood@arm.com>
14768 * coretypes.h (scalar_mode_pod): New typedef.
14769 * gdbhooks.py (build_pretty_printer): Handle it.
14770 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
14771 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
14772 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
14773 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
14774 as_a <scalar_mode>.
14776 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14777 Alan Hayward <alan.hayward@arm.com>
14778 David Sherwood <david.sherwood@arm.com>
14780 * machmode.h (mode_for_vector): Take a scalar_mode instead
14782 * stor-layout.c (mode_for_vector): Likewise.
14783 * explow.c (promote_mode): Use as_a <scalar_mode>.
14784 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
14786 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14787 Alan Hayward <alan.hayward@arm.com>
14788 David Sherwood <david.sherwood@arm.com>
14790 * target.def (preferred_simd_mode): Take a scalar_mode
14791 instead of a machine_mode.
14792 * targhooks.h (default_preferred_simd_mode): Likewise.
14793 * targhooks.c (default_preferred_simd_mode): Likewise.
14794 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
14795 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
14796 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
14797 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
14798 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
14799 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
14800 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
14801 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
14803 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
14804 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
14805 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
14806 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
14807 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
14808 * doc/tm.texi: Regenerate.
14809 * optabs-query.c (can_vec_mask_load_store_p): Return false for
14812 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14813 Alan Hayward <alan.hayward@arm.com>
14814 David Sherwood <david.sherwood@arm.com>
14816 * target.def (scalar_mode_supported_p): Take a scalar_mode
14817 instead of a machine_mode.
14818 * targhooks.h (default_scalar_mode_supported_p): Likewise.
14819 * targhooks.c (default_scalar_mode_supported_p): Likewise.
14820 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
14821 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
14822 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
14823 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
14824 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
14825 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
14826 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
14827 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
14828 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
14829 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
14830 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
14831 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
14833 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
14834 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
14835 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
14836 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
14837 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
14839 * doc/tm.texi: Regenerate.
14841 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14842 Alan Hayward <alan.hayward@arm.com>
14843 David Sherwood <david.sherwood@arm.com>
14845 * coretypes.h (opt_scalar_mode): New typedef.
14846 * gdbhooks.py (build_pretty_printers): Handle it.
14847 * machmode.h (mode_iterator::get_2xwider): Add overload for
14849 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
14851 * expr.c (convert_mode_scalar): Likewise.
14852 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14853 * optabs.c (expand_float): Likewise.
14854 (expand_fix): Likewise.
14855 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14857 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14858 Alan Hayward <alan.hayward@arm.com>
14859 David Sherwood <david.sherwood@arm.com>
14861 * optabs.c (expand_float): Explicitly check for scalars before
14862 using a branching expansion.
14863 (expand_fix): Likewise.
14865 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14866 Alan Hayward <alan.hayward@arm.com>
14867 David Sherwood <david.sherwood@arm.com>
14869 * expr.c (convert_mode): Split scalar handling out into...
14870 (convert_mode_scalar): ...this new function. Treat the modes
14873 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14874 Alan Hayward <alan.hayward@arm.com>
14875 David Sherwood <david.sherwood@arm.com>
14877 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
14879 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
14881 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14882 Alan Hayward <alan.hayward@arm.com>
14883 David Sherwood <david.sherwood@arm.com>
14885 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
14886 rather than a machine_mode.
14887 (fixed_from_string): Likewise.
14888 (fixed_convert): Likewise.
14889 (fixed_convert_from_int): Likewise.
14890 (fixed_convert_from_real): Likewise.
14891 (real_convert_from_fixed): Likewise.
14892 * fixed-value.c (fixed_from_double_int): Likewise.
14893 (fixed_from_string): Likewise.
14894 (fixed_convert): Likewise.
14895 (fixed_convert_from_int): Likewise.
14896 (fixed_convert_from_real): Likewise.
14897 (real_convert_from_fixed): Likewise.
14898 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
14900 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14901 Alan Hayward <alan.hayward@arm.com>
14902 David Sherwood <david.sherwood@arm.com>
14904 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
14905 of separate mode class checks. Do not allow vector modes here.
14906 (immed_wide_int_const): Use as_a <scalar_mode>.
14907 * explow.c (trunc_int_for_mode): Likewise.
14908 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
14909 (wi::shwi): Likewise.
14910 (wi::min_value): Likewise.
14911 (wi::max_value): Likewise.
14912 * dwarf2out.c (loc_descriptor): Likewise.
14913 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
14914 for CONST_WIDE_INT.
14916 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14917 Alan Hayward <alan.hayward@arm.com>
14918 David Sherwood <david.sherwood@arm.com>
14920 * tree.h (SCALAR_TYPE_MODE): New macro.
14921 * expr.c (expand_expr_addr_expr_1): Use it.
14922 (expand_expr_real_2): Likewise.
14923 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
14924 (fold_convert_const_fixed_from_int): Likewise.
14925 (fold_convert_const_fixed_from_real): Likewise.
14926 (native_encode_fixed): Likewise
14927 (native_encode_complex): Likewise
14928 (native_encode_vector): Likewise.
14929 (native_interpret_fixed): Likewise.
14930 (native_interpret_real): Likewise.
14931 (native_interpret_complex): Likewise.
14932 (native_interpret_vector): Likewise.
14933 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
14934 (simd_clone_adjust_argument_types): Likewise.
14935 (simd_clone_init_simd_arrays): Likewise.
14936 (simd_clone_adjust): Likewise.
14937 * stor-layout.c (layout_type): Likewise.
14938 * tree.c (build_minus_one_cst): Likewise.
14939 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
14940 * tree-inline.c (estimate_move_cost): Likewise.
14941 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
14942 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
14943 (vectorizable_reduction): Likewise.
14944 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
14945 (vect_recog_mixed_size_cond_pattern): Likewise.
14946 (check_bool_pattern): Likewise.
14947 (adjust_bool_pattern): Likewise.
14948 (search_type_for_mask_1): Likewise.
14949 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
14950 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14951 (vectorizable_load): Likewise.
14952 (vectorizable_store): Likewise.
14953 * ubsan.c (ubsan_encode_value): Likewise.
14954 * varasm.c (output_constant): Likewise.
14956 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14957 Alan Hayward <alan.hayward@arm.com>
14958 David Sherwood <david.sherwood@arm.com>
14960 * coretypes.h (scalar_mode): New class.
14961 * machmode.h (scalar_mode): Likewise.
14962 (scalar_mode::includes_p): New function.
14963 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
14964 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
14965 * genmodes.c (get_mode_class): Handle remaining scalar modes.
14966 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
14967 * expmed.c (store_bit_field_1): Likewise.
14968 (extract_bit_field_1): Likewise.
14969 * expr.c (write_complex_part): Likewise.
14970 (read_complex_part): Likewise.
14971 (emit_move_complex_push): Likewise.
14972 (expand_expr_real_2): Likewise.
14973 * function.c (assign_parm_setup_reg): Likewise.
14974 (assign_parms_unsplit_complex): Likewise.
14975 * optabs.c (expand_binop): Likewise.
14976 * rtlanal.c (subreg_get_info): Likewise.
14977 * simplify-rtx.c (simplify_immed_subreg): Likewise.
14978 * varasm.c (output_constant_pool_2): Likewise.
14980 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14981 Alan Hayward <alan.hayward@arm.com>
14982 David Sherwood <david.sherwood@arm.com>
14984 * expmed.c (extract_high_half): Use scalar_int_mode and remove
14986 (expmed_mult_highpart_optab): Likewise.
14987 (expmed_mult_highpart): Likewise.
14989 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14990 Alan Hayward <alan.hayward@arm.com>
14991 David Sherwood <david.sherwood@arm.com>
14993 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
14994 instead of a machine_mode.
14995 (builtin_memset_read_str): Likewise.
14996 * builtins.c (c_readstr): Likewise.
14997 (builtin_memcpy_read_str): Likewise.
14998 (builtin_strncpy_read_str): Likewise.
14999 (builtin_memset_read_str): Likewise.
15000 (builtin_memset_gen_str): Likewise.
15001 (expand_builtin_signbit): Use scalar_int_mode for local variables.
15002 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
15003 instead of a machine_mode.
15004 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
15006 (make_extraction): Likewise.
15007 (try_widen_shift_mode): Take and return scalar_int_modes instead
15009 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
15010 a scalar_int_mode instead of a machine_mode.
15011 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
15012 (avr_addr_space_pointer_mode): Likewise.
15013 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
15014 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
15015 (msp430_unwind_word_mode): Likewise.
15016 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
15017 (spu_addr_space_pointer_mode): Likewise.
15018 (spu_addr_space_address_mode): Likewise.
15019 (spu_libgcc_cmp_return_mode): Likewise.
15020 (spu_libgcc_shift_count_mode): Likewise.
15021 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
15022 (rl78_addr_space_pointer_mode): Likewise.
15023 (fl78_unwind_word_mode): Likewise.
15024 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
15026 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
15027 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
15028 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
15029 (mips_valid_pointer_mode): Likewise.
15030 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
15031 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
15032 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
15034 (ft32_addr_space_address_mode): Likewise.
15035 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
15036 scalar_int_mode instead of a machine_mode.
15037 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
15039 (m32c_addr_space_address_mode): Likewise.
15040 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
15041 (rs6000_eh_return_filter_mode): Likewise.
15042 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
15043 (rs6000_eh_return_filter_mode): Likewise.
15044 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
15045 (s390_libgcc_shift_count_mode): Likewise.
15046 (s390_unwind_word_mode): Likewise.
15047 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
15049 * target.def (mode_rep_extended): Likewise.
15050 (valid_pointer_mode): Likewise.
15051 (addr_space.valid_pointer_mode): Likewise.
15052 (eh_return_filter_mode): Return a scalar_int_mode rather than
15054 (libgcc_cmp_return_mode): Likewise.
15055 (libgcc_shift_count_mode): Likewise.
15056 (unwind_word_mode): Likewise.
15057 (addr_space.pointer_mode): Likewise.
15058 (addr_space.address_mode): Likewise.
15059 * doc/tm.texi: Regenerate.
15060 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
15062 (do_jump): Use scalar_int_mode for local variables.
15063 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
15064 rather than a machine_mode.
15065 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
15066 (scompare_loc_descriptor_wide): Likewise.
15067 (scompare_loc_descriptor_narrow): Likewise.
15068 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
15070 * except.c (sjlj_emit_dispatch_table): Likewise.
15071 (expand_builtin_eh_copy_values): Likewise.
15072 * explow.c (convert_memory_address_addr_space_1): Likewise.
15073 Take a scalar_int_mode rather than a machine_mode.
15074 (convert_memory_address_addr_space): Take a scalar_int_mode rather
15075 than a machine_mode.
15076 (memory_address_addr_space): Use scalar_int_mode for local variables.
15077 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
15078 rather than a machine_mode.
15079 * expmed.c (mask_rtx): Likewise.
15080 (init_expmed_one_conv): Likewise.
15081 (expand_mult_highpart_adjust): Likewise.
15082 (extract_high_half): Likewise.
15083 (expmed_mult_highpart_optab): Likewise.
15084 (expmed_mult_highpart): Likewise.
15085 (expand_smod_pow2): Likewise.
15086 (expand_sdiv_pow2): Likewise.
15087 (emit_store_flag_int): Likewise.
15088 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
15090 (extract_low_bits): Likewise.
15091 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
15093 * expr.c (pieces_addr::adjust): Likewise.
15094 (can_store_by_pieces): Likewise.
15095 (store_by_pieces): Likewise.
15096 (clear_by_pieces_1): Likewise.
15097 (expand_expr_addr_expr_1): Likewise.
15098 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
15099 (expand_expr_real_1): Likewise.
15100 (try_casesi): Likewise.
15101 * final.c (shorten_branches): Likewise.
15102 * fold-const.c (fold_convert_const_int_from_fixed): Change the
15103 type of "mode" to machine_mode.
15104 * internal-fn.c (expand_arith_overflow_result_store): Take a
15105 scalar_int_mode rather than a machine_mode.
15106 (expand_mul_overflow): Use scalar_int_mode for local variables.
15107 * loop-doloop.c (doloop_modify): Likewise.
15108 (doloop_optimize): Likewise.
15109 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
15110 than a machine_mode.
15111 (expand_doubleword_shift_condmove): Likewise.
15112 (expand_doubleword_shift): Likewise.
15113 (expand_doubleword_clz): Likewise.
15114 (expand_doubleword_popcount): Likewise.
15115 (expand_doubleword_parity): Likewise.
15116 (expand_absneg_bit): Use scalar_int_mode for local variables.
15117 (prepare_float_lib_cmp): Likewise.
15118 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
15119 rather than a machine_mode.
15120 (convert_memory_address_addr_space): Likewise.
15121 (get_mode_bounds): Likewise.
15122 (get_address_mode): Return a scalar_int_mode rather than a
15124 * rtlanal.c (get_address_mode): Likewise.
15125 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
15126 than a machine_mode.
15127 * targhooks.c (default_mode_rep_extended): Likewise.
15128 (default_valid_pointer_mode): Likewise.
15129 (default_addr_space_valid_pointer_mode): Likewise.
15130 (default_eh_return_filter_mode): Return a scalar_int_mode rather
15131 than a machine_mode.
15132 (default_libgcc_cmp_return_mode): Likewise.
15133 (default_libgcc_shift_count_mode): Likewise.
15134 (default_unwind_word_mode): Likewise.
15135 (default_addr_space_pointer_mode): Likewise.
15136 (default_addr_space_address_mode): Likewise.
15137 * targhooks.h (default_eh_return_filter_mode): Likewise.
15138 (default_libgcc_cmp_return_mode): Likewise.
15139 (default_libgcc_shift_count_mode): Likewise.
15140 (default_unwind_word_mode): Likewise.
15141 (default_addr_space_pointer_mode): Likewise.
15142 (default_addr_space_address_mode): Likewise.
15143 (default_mode_rep_extended): Take a scalar_int_mode rather than
15145 (default_valid_pointer_mode): Likewise.
15146 (default_addr_space_valid_pointer_mode): Likewise.
15147 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
15149 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
15150 rather than a machine_mode.
15151 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
15152 for local variables.
15153 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15154 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
15155 than a machine_mode.
15157 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15158 Alan Hayward <alan.hayward@arm.com>
15159 David Sherwood <david.sherwood@arm.com>
15161 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
15162 the mode argument to scalar_int_mode.
15163 (do_jump_by_parts_zero_rtx): Likewise.
15164 (do_jump_by_parts_equality_rtx): Likewise.
15165 (do_jump_by_parts_greater): Take a mode argument.
15166 (do_jump_by_parts_equality): Likewise.
15167 (do_jump_1): Update calls accordingly.
15169 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15170 Alan Hayward <alan.hayward@arm.com>
15171 David Sherwood <david.sherwood@arm.com>
15173 * is-a.h (safe_dyn_cast): New function.
15174 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
15175 (jump_table_for_label): Likewise.
15176 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
15177 instead of an rtx_insn *.
15178 (shorten_branches): Use dyn_cast instead of LABEL_P and
15179 JUMP_TABLE_DATA_P. Use jump_table_for_label and
15180 rtx_jump_table_data::get_data_mode.
15181 (final_scan_insn): Likewise.
15183 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15184 Alan Hayward <alan.hayward@arm.com>
15185 David Sherwood <david.sherwood@arm.com>
15187 * combine.c (try_combine): Use is_a <scalar_int_mode> when
15188 trying to combine a full-register integer set with a subreg
15191 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15192 Alan Hayward <alan.hayward@arm.com>
15193 David Sherwood <david.sherwood@arm.com>
15195 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
15196 that is always either address_mode or pointer_mode.
15198 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15199 Alan Hayward <alan.hayward@arm.com>
15200 David Sherwood <david.sherwood@arm.com>
15202 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
15203 when the two are known to be equal.
15205 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15206 Alan Hayward <alan.hayward@arm.com>
15207 David Sherwood <david.sherwood@arm.com>
15209 * simplify-rtx.c (simplify_const_unary_operation): Use
15210 is_a <scalar_int_mode> instead of checking for a nonzero
15211 precision. Forcibly convert op_mode to a scalar_int_mode
15212 in that case. More clearly differentiate the operand and
15213 result modes and use the former when deciding what the value
15214 of a count-bits operation should be. Use is_int_mode instead
15215 of checking for a MODE_INT. Remove redundant check for whether
15216 this mode has a zero precision.
15218 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15219 Alan Hayward <alan.hayward@arm.com>
15220 David Sherwood <david.sherwood@arm.com>
15222 * optabs.c (widen_leading): Change the type of the mode argument
15223 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
15224 (widen_bswap): Likewise.
15225 (expand_parity): Likewise.
15226 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
15227 (expand_ffs): Likewise.
15228 (epand_unop): Check for scalar integer modes before calling the
15231 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15232 Alan Hayward <alan.hayward@arm.com>
15233 David Sherwood <david.sherwood@arm.com>
15235 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
15237 (expand_expr_real_1): Update call accordingly.
15239 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15240 Alan Hayward <alan.hayward@arm.com>
15241 David Sherwood <david.sherwood@arm.com>
15243 * expmed.c (store_bit_field_using_insv): Add op0_mode and
15244 value_mode arguments. Use scalar_int_mode internally.
15245 (store_bit_field_1): Rename the new integer mode from imode
15246 to op0_mode and use it instead of GET_MODE (op0). Update calls
15247 to store_split_bit_field, store_bit_field_using_insv and
15248 store_fixed_bit_field.
15249 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
15250 Use scalar_int_mode internally. Use a bit count rather than a mode
15251 when calculating the largest bit size for get_best_mode.
15252 Update calls to store_split_bit_field and store_fixed_bit_field_1.
15253 (store_fixed_bit_field_1): Add mode and value_mode arguments.
15254 Remove assertion that OP0 has a scalar integer mode.
15255 (store_split_bit_field): Add op0_mode and value_mode arguments.
15256 Update calls to extract_fixed_bit_field.
15257 (extract_bit_field_using_extv): Add an op0_mode argument.
15258 Use scalar_int_mode internally.
15259 (extract_bit_field_1): Rename the new integer mode from imode to
15260 op0_mode and use it instead of GET_MODE (op0). Update calls to
15261 extract_split_bit_field, extract_bit_field_using_extv and
15262 extract_fixed_bit_field.
15263 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
15264 to extract_split_bit_field and extract_fixed_bit_field_1.
15265 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
15266 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
15267 on the target mode.
15268 (extract_split_bit_field): Add an op0_mode argument. Update call
15269 to extract_fixed_bit_field.
15271 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15272 Alan Hayward <alan.hayward@arm.com>
15273 David Sherwood <david.sherwood@arm.com>
15275 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
15276 * explow.c (hard_function_value): Likewise.
15277 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
15278 convert_to_mode call outside the loop.
15279 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
15280 for the mode iterator. Require the mode specified by max_pieces
15282 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
15284 (copy_blkmode_to_reg): Likewise.
15285 (set_storage_via_setmem): Likewise.
15286 * optabs.c (prepare_cmp_insn): Likewise.
15287 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15288 * stor-layout.c (finish_bitfield_representative): Likewise.
15290 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15291 Alan Hayward <alan.hayward@arm.com>
15292 David Sherwood <david.sherwood@arm.com>
15294 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
15295 * expr.c (convert_move): Use them.
15296 (convert_modes): Likewise.
15297 (store_expr_with_bounds): Likewise.
15299 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15300 Alan Hayward <alan.hayward@arm.com>
15301 David Sherwood <david.sherwood@arm.com>
15303 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
15304 parameter for the mode of "x". Remove the "known_x", "known_mode"
15305 and "known_ret" arguments. Change the type of the mode argument
15306 to scalar_int_mode.
15307 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
15308 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
15309 (reg_num_sign_bit_copies_for_combine): Likewise.
15310 * rtlanal.c (nonzero_bits1): Likewise.
15311 (num_sign_bit_copies1): Likewise.
15312 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
15313 (reg_num_sign_bit_copies_general): Likewise.
15314 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
15315 (reg_nonzero_bits_general): Likewise.
15317 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15318 Alan Hayward <alan.hayward@arm.com>
15319 David Sherwood <david.sherwood@arm.com>
15321 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
15322 than in subroutines. Return 1 for non-integer modes.
15323 (cached_num_sign_bit_copies): Change the type of the mode parameter
15324 to scalar_int_mode.
15325 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
15326 classes. Handle CONST_INT_P first and then check whether X also
15327 has a scalar integer mode. Check the same thing for inner registers
15328 of a SUBREG and for values that are being extended or truncated.
15330 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15331 Alan Hayward <alan.hayward@arm.com>
15332 David Sherwood <david.sherwood@arm.com>
15334 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
15335 in subroutines. Return the mode mask for non-integer modes.
15336 (cached_nonzero_bits): Change the type of the mode parameter
15337 to scalar_int_mode.
15338 (nonzero_bits1): Likewise. Remove early exit for other mode
15339 classes. Handle CONST_INT_P first and then check whether X
15340 also has a scalar integer mode.
15342 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15343 Alan Hayward <alan.hayward@arm.com>
15344 David Sherwood <david.sherwood@arm.com>
15346 * expr.c (widest_int_mode_for_size): Make the comment match the code.
15347 Return a scalar_int_mode and assert that the size is greater than
15349 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
15350 (op_by_pieces_d::op_by_pieces_d): Likewise.
15351 (op_by_pieces_d::run): Likewise.
15352 (can_store_by_pieces): Likewise.
15354 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15355 Alan Hayward <alan.hayward@arm.com>
15356 David Sherwood <david.sherwood@arm.com>
15358 * combine.c (extract_left_shift): Add a mode argument and update
15360 (make_compound_operation_int): Change the type of the mode parameter
15361 to scalar_int_mode and update the call to extract_left_shift.
15363 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15364 Alan Hayward <alan.hayward@arm.com>
15365 David Sherwood <david.sherwood@arm.com>
15367 * combine.c (simplify_and_const_int): Change the type of the mode
15368 parameter to scalar_int_mode.
15369 (simplify_and_const_int_1): Likewise. Update recursive call.
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 * combine.c (simplify_compare_const): Check that the mode is a
15376 scalar_int_mode (rather than VOIDmode) before testing its
15378 (simplify_comparison): Move COMPARISON_P handling out of the
15379 loop and restrict the latter part of the loop to scalar_int_modes.
15380 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
15381 and when considering SUBREG_REGs. Use is_int_mode instead of
15382 checking GET_MODE_CLASS against MODE_INT.
15384 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15385 Alan Hayward <alan.hayward@arm.com>
15386 David Sherwood <david.sherwood@arm.com>
15388 * combine.c (try_widen_shift_mode): Move check for equal modes to...
15389 (simplify_shift_const_1): ...here. Use scalar_int_mode for
15390 shift_unit_mode and for modes involved in scalar shifts.
15392 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15393 Alan Hayward <alan.hayward@arm.com>
15394 David Sherwood <david.sherwood@arm.com>
15396 * combine.c (force_int_to_mode): New function, split out from...
15397 (force_to_mode): ...here. Keep xmode up-to-date and use it
15398 instead of GET_MODE (x).
15400 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15401 Alan Hayward <alan.hayward@arm.com>
15402 David Sherwood <david.sherwood@arm.com>
15404 * optabs-query.h (extraction_insn::struct_mode): Change type to
15405 opt_scalar_int_mode and update comment.
15406 (extraction_insn::field_mode): Change type to scalar_int_mode.
15407 (extraction_insn::pos_mode): Likewise.
15408 * combine.c (make_extraction): Update accordingly.
15409 * optabs-query.c (get_traditional_extraction_insn): Likewise.
15410 (get_optab_extraction_insn): Likewise.
15411 * recog.c (simplify_while_replacing): Likewise.
15412 * expmed.c (narrow_bit_field_mem): Change the type of the mode
15413 parameter to opt_scalar_int_mode.
15415 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15416 Alan Hayward <alan.hayward@arm.com>
15417 David Sherwood <david.sherwood@arm.com>
15419 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
15420 to a scalar_int_mode instead of a machine_mode.
15421 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
15422 (get_best_mode): Return a boolean and use a pointer argument to store
15423 the selected mode. Replace the limit mode parameter with a bit limit.
15424 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
15425 for the values returned by bit_field_mode_iterator::next_mode.
15426 (store_bit_field): Update call to get_best_mode.
15427 (store_fixed_bit_field): Likewise.
15428 (extract_fixed_bit_field): Likewise.
15429 * expr.c (optimize_bitfield_assignment_op): Likewise.
15430 * fold-const.c (optimize_bit_field_compare): Likewise.
15431 (fold_truth_andor_1): Likewise.
15432 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
15433 Update for new type of m_mode.
15434 (get_best_mode): As above.
15436 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15437 Alan Hayward <alan.hayward@arm.com>
15438 David Sherwood <david.sherwood@arm.com>
15440 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
15441 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
15442 (store_bit_field): Check is_a <scalar_int_mode> before calling
15443 strict_volatile_bitfield_p.
15444 (extract_bit_field): Likewise.
15446 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15447 Alan Hayward <alan.hayward@arm.com>
15448 David Sherwood <david.sherwood@arm.com>
15450 * target.def (cstore_mode): Return a scalar_int_mode.
15451 * doc/tm.texi: Regenerate.
15452 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
15453 * targhooks.h (default_cstore_mode): Likewise.
15454 * targhooks.c (default_cstore_mode): Likewise, using a forced
15456 * expmed.c (emit_cstore): Expect the target of the cstore to be
15459 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15460 Alan Hayward <alan.hayward@arm.com>
15461 David Sherwood <david.sherwood@arm.com>
15463 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
15465 (niter_desc): Likewise mode.
15466 (iv_analyze): Add a mode parameter.
15468 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
15469 and change its type to scalar_int_mode.
15470 * loop-iv.c: Update commentary at head of file.
15471 (iv_constant): Pass the mode paraeter before the rtx it describes
15472 and change its type to scalar_int_mode. Remove VOIDmode handling.
15473 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
15474 (iv_extend): Likewise.
15475 (shorten_into_mode): Likewise.
15476 (iv_add): Use scalar_int_mode.
15477 (iv_mult): Likewise.
15478 (iv_shift): Likewise.
15479 (canonicalize_iv_subregs): Likewise.
15480 (get_biv_step_1): Pass the outer_mode parameter before the rtx
15481 it describes and change its mode to scalar_int_mode. Also change
15482 the type of the returned inner_mode to scalar_int_mode.
15483 (get_biv_step): Likewise, turning outer_mode from a pointer
15484 into a direct parameter. Update call to get_biv_step_1.
15485 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
15486 iv_constant and get_biv_step.
15487 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
15488 and change its type to scalar_int_mode. Don't initialise iv->mode
15489 to VOIDmode and remove later checks for its still being VOIDmode.
15490 Update calls to iv_analyze_op and iv_analyze_expr. Check
15491 is_a <scalar_int_mode> when changing the mode under consideration.
15492 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
15493 Update call to iv_analyze_expr.
15494 (iv_analyze_op): Add a mode parameter. Reject subregs whose
15495 inner register is not also a scalar_int_mode. Update call to
15497 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
15498 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
15499 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
15500 separate mode class checks. Update calls to iv_analyze. Remove
15501 fix-up of VOIDmodes after iv_analyze_biv.
15502 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
15503 don't have a scalar_int_mode. Update call to biv_p.
15505 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15506 Alan Hayward <alan.hayward@arm.com>
15507 David Sherwood <david.sherwood@arm.com>
15509 * cfgexpand.c (convert_debug_memory_address): Use
15510 as_a <scalar_int_mode>.
15511 * combine.c (expand_compound_operation): Likewise.
15512 (make_extraction): Likewise.
15513 (change_zero_ext): Likewise.
15514 (simplify_comparison): Likewise.
15515 * cse.c (cse_insn): Likewise.
15516 * dwarf2out.c (minmax_loc_descriptor): Likewise.
15517 (mem_loc_descriptor): Likewise.
15518 (loc_descriptor): Likewise.
15519 * expmed.c (init_expmed_one_mode): Likewise.
15520 (synth_mult): Likewise.
15521 (emit_store_flag_1): Likewise.
15522 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
15523 of a comparison with size.
15524 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
15525 (reduce_to_bit_field_precision): Likewise.
15526 * function.c (expand_function_end): Likewise.
15527 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
15528 * loop-doloop.c (doloop_modify): Likewise.
15529 * optabs.c (expand_binop): Likewise.
15530 (expand_unop): Likewise.
15531 (expand_copysign_absneg): Likewise.
15532 (prepare_cmp_insn): Likewise.
15533 (maybe_legitimize_operand): Likewise.
15534 * recog.c (const_scalar_int_operand): Likewise.
15535 * rtlanal.c (get_address_mode): Likewise.
15536 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15537 (simplify_cond_clz_ctz): Likewise.
15538 * tree-nested.c (get_nl_goto_field): Likewise.
15539 * tree.c (build_vector_type_for_mode): Likewise.
15540 * var-tracking.c (use_narrower_mode): Likewise.
15542 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15543 Alan Hayward <alan.hayward@arm.com>
15544 David Sherwood <david.sherwood@arm.com>
15546 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
15547 * builtins.c (expand_builtin_signbit): Use it.
15548 * cfgexpand.c (expand_debug_expr): Likewise.
15549 * dojump.c (do_jump): Likewise.
15550 (do_compare_and_jump): Likewise.
15551 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
15552 * expmed.c (make_tree): Likewise.
15553 * expr.c (expand_expr_real_2): Likewise.
15554 (expand_expr_real_1): Likewise.
15555 (try_casesi): Likewise.
15556 * fold-const-call.c (fold_const_call_ss): Likewise.
15557 * fold-const.c (unextend): Likewise.
15558 (extract_muldiv_1): Likewise.
15559 (fold_single_bit_test): Likewise.
15560 (native_encode_int): Likewise.
15561 (native_encode_string): Likewise.
15562 (native_interpret_int): Likewise.
15563 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
15564 * internal-fn.c (expand_addsub_overflow): Likewise.
15565 (expand_neg_overflow): Likewise.
15566 (expand_mul_overflow): Likewise.
15567 (expand_arith_overflow): Likewise.
15568 * match.pd: Likewise.
15569 * stor-layout.c (layout_type): Likewise.
15570 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15571 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
15572 * tree-ssanames.c (get_range_info): Likewise.
15573 * tree-switch-conversion.c (array_value_type) Likewise.
15574 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
15575 (vect_recog_divmod_pattern): Likewise.
15576 (vect_recog_mixed_size_cond_pattern): Likewise.
15577 * tree-vrp.c (extract_range_basic): Likewise.
15578 (simplify_float_conversion_using_ranges): Likewise.
15579 * tree.c (int_fits_type_p): Likewise.
15580 * ubsan.c (instrument_bool_enum_load): Likewise.
15581 * varasm.c (mergeable_string_section): Likewise.
15582 (narrowing_initializer_constant_valid_p): Likewise.
15583 (output_constant): Likewise.
15585 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15586 Alan Hayward <alan.hayward@arm.com>
15587 David Sherwood <david.sherwood@arm.com>
15589 * machmode.h (NARROWEST_INT_MODE): New macro.
15590 * expr.c (alignment_for_piecewise_move): Use it instead of
15591 GET_CLASS_NARROWEST_MODE (MODE_INT).
15592 (push_block): Likewise.
15593 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
15595 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15597 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15598 Alan Hayward <alan.hayward@arm.com>
15599 David Sherwood <david.sherwood@arm.com>
15601 * postreload.c (move2add_valid_value_p): Change the type of the
15602 mode parameter to scalar_int_mode.
15603 (move2add_use_add2_insn): Add a mode parameter and use it instead
15605 (move2add_use_add3_insn): Likewise.
15606 (reload_cse_move2add): Update accordingly.
15608 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15609 Alan Hayward <alan.hayward@arm.com>
15610 David Sherwood <david.sherwood@arm.com>
15612 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
15614 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
15615 * optabs.c (expand_unop): Likewise.
15617 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15618 Alan Hayward <alan.hayward@arm.com>
15619 David Sherwood <david.sherwood@arm.com>
15621 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
15622 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
15623 (popcount_loc_descriptor): Likewise.
15624 (bswap_loc_descriptor): Likewise.
15625 (rotate_loc_descriptor): Likewise.
15626 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
15627 calling the functions above.
15629 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15630 Alan Hayward <alan.hayward@arm.com>
15631 David Sherwood <david.sherwood@arm.com>
15633 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
15635 (try_combine): Likewise.
15636 (simplify_if_then_else): Likewise.
15637 * cse.c (cse_insn): Likewise.
15638 * dwarf2out.c (mem_loc_descriptor): Likewise.
15639 * emit-rtl.c (gen_lowpart_common): Likewise.
15640 * simplify-rtx.c (simplify_truncation): Likewise.
15641 (simplify_binary_operation_1): Likewise.
15642 (simplify_const_relational_operation): Likewise.
15643 (simplify_ternary_operation): Likewise.
15644 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
15646 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15647 Alan Hayward <alan.hayward@arm.com>
15648 David Sherwood <david.sherwood@arm.com>
15650 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
15651 * reload.c (push_reload): Likewise.
15652 (find_reloads): Likewise.
15654 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15655 Alan Hayward <alan.hayward@arm.com>
15656 David Sherwood <david.sherwood@arm.com>
15658 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
15659 (make_compound_operation_int): Likewise.
15660 (change_zero_ext): Likewise.
15661 * expr.c (convert_move): Likewise.
15662 (convert_modes): Likewise.
15663 * fwprop.c (forward_propagate_subreg): Likewise.
15664 * loop-iv.c (get_biv_step_1): Likewise.
15665 * optabs.c (widen_operand): Likewise.
15666 * postreload.c (move2add_valid_value_p): Likewise.
15667 * recog.c (simplify_while_replacing): Likewise.
15668 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15669 (simplify_binary_operation_1): Likewise. Remove redundant
15670 mode equality check.
15672 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15673 Alan Hayward <alan.hayward@arm.com>
15674 David Sherwood <david.sherwood@arm.com>
15676 * combine.c (combine_simplify_rtx): Add checks for
15677 is_a <scalar_int_mode>.
15678 (simplify_if_then_else): Likewise.
15679 (make_field_assignment): Likewise.
15680 (simplify_comparison): Likewise.
15681 * ifcvt.c (noce_try_bitop): Likewise.
15682 * loop-invariant.c (canonicalize_address_mult): Likewise.
15683 * simplify-rtx.c (simplify_unary_operation_1): 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 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
15690 is_a <scalar_int_mode> instead of != BLKmode.
15692 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15693 Alan Hayward <alan.hayward@arm.com>
15694 David Sherwood <david.sherwood@arm.com>
15696 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
15697 instead of != VOIDmode.
15698 * combine.c (if_then_else_cond): Likewise.
15699 (change_zero_ext): Likewise.
15700 * dwarf2out.c (mem_loc_descriptor): Likewise.
15701 (loc_descriptor): Likewise.
15702 * rtlanal.c (canonicalize_condition): Likewise.
15703 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
15705 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15706 Alan Hayward <alan.hayward@arm.com>
15707 David Sherwood <david.sherwood@arm.com>
15709 * simplify-rtx.c (simplify_binary_operation_1): Use
15710 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
15712 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15713 Alan Hayward <alan.hayward@arm.com>
15714 David Sherwood <david.sherwood@arm.com>
15716 * wide-int.h (int_traits<unsigned char>) New class.
15717 (int_traits<unsigned short>) Likewise.
15718 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
15719 Use GET_MODE_UNIT_PRECISION and remove redundant test for
15721 * combine.c (set_nonzero_bits_and_sign_copies): Use
15722 is_a <scalar_int_mode>.
15723 (find_split_point): Likewise.
15724 (combine_simplify_rtx): Likewise.
15725 (simplify_logical): Likewise.
15726 (expand_compound_operation): Likewise.
15727 (expand_field_assignment): Likewise.
15728 (make_compound_operation): Likewise.
15729 (extended_count): Likewise.
15730 (change_zero_ext): Likewise.
15731 (simplify_comparison): Likewise.
15732 * dwarf2out.c (scompare_loc_descriptor): Likewise.
15733 (ucompare_loc_descriptor): Likewise.
15734 (minmax_loc_descriptor): Likewise.
15735 (mem_loc_descriptor): Likewise.
15736 (loc_descriptor): Likewise.
15737 * expmed.c (init_expmed_one_mode): Likewise.
15738 * lra-constraints.c (lra_constraint_offset): Likewise.
15739 * optabs.c (prepare_libcall_arg): Likewise.
15740 * postreload.c (move2add_note_store): Likewise.
15741 * reload.c (operands_match_p): Likewise.
15742 * rtl.h (load_extend_op): Likewise.
15743 * rtlhooks.c (gen_lowpart_general): Likewise.
15744 * simplify-rtx.c (simplify_truncation): Likewise.
15745 (simplify_unary_operation_1): Likewise.
15746 (simplify_binary_operation_1): Likewise.
15747 (simplify_const_binary_operation): Likewise.
15748 (simplify_const_relational_operation): Likewise.
15749 (simplify_subreg): Likewise.
15750 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15751 * var-tracking.c (adjust_mems): Likewise.
15752 (prepare_call_arguments): 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 * machmode.h (is_int_mode): New fuction.
15759 * combine.c (find_split_point): Use it.
15760 (combine_simplify_rtx): Likewise.
15761 (simplify_if_then_else): Likewise.
15762 (simplify_set): Likewise.
15763 (simplify_shift_const_1): Likewise.
15764 (simplify_comparison): Likewise.
15765 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
15766 * cse.c (notreg_cost): Likewise.
15767 (cse_insn): Likewise.
15768 * cselib.c (cselib_lookup_1): Likewise.
15769 * dojump.c (do_jump_1): Likewise.
15770 (do_compare_rtx_and_jump): Likewise.
15771 * dse.c (get_call_args): Likewise.
15772 * dwarf2out.c (rtl_for_decl_init): Likewise.
15773 (native_encode_initializer): Likewise.
15774 * expmed.c (emit_store_flag_1): Likewise.
15775 (emit_store_flag): Likewise.
15776 * expr.c (convert_modes): Likewise.
15777 (store_field): Likewise.
15778 (expand_expr_real_1): Likewise.
15779 * fold-const.c (fold_read_from_constant_string): Likewise.
15780 * gimple-ssa-sprintf.c (get_format_string): Likewise.
15781 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
15782 * optabs.c (expand_binop): Likewise.
15783 (expand_unop): Likewise.
15784 (expand_abs_nojump): Likewise.
15785 (expand_one_cmpl_abs_nojump): Likewise.
15786 * simplify-rtx.c (mode_signbit_p): Likewise.
15787 (val_signbit_p): Likewise.
15788 (val_signbit_known_set_p): Likewise.
15789 (val_signbit_known_clear_p): Likewise.
15790 (simplify_relational_operation_1): Likewise.
15791 * tree.c (vector_type_mode): Likewise.
15793 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15794 Alan Hayward <alan.hayward@arm.com>
15795 David Sherwood <david.sherwood@arm.com>
15797 * machmode.h (smallest_mode_for_size): Fix formatting.
15798 (smallest_int_mode_for_size): New function.
15799 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
15800 instead of smallest_mode_for_size.
15801 * combine.c (make_extraction): Likewise.
15802 * config/arc/arc.c (arc_expand_movmem): Likewise.
15803 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
15804 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15805 * config/s390/s390.c (s390_expand_insv): Likewise.
15806 * config/sparc/sparc.c (assign_int_registers): Likewise.
15807 * config/spu/spu.c (spu_function_value): Likewise.
15808 (spu_function_arg): Likewise.
15809 * coverage.c (get_gcov_type): Likewise.
15810 (get_gcov_unsigned_t): Likewise.
15811 * dse.c (find_shift_sequence): Likewise.
15812 * expmed.c (store_bit_field_1): Likewise.
15813 * expr.c (convert_move): Likewise.
15814 (store_field): Likewise.
15815 * internal-fn.c (expand_arith_overflow): Likewise.
15816 * optabs-query.c (get_best_extraction_insn): Likewise.
15817 * optabs.c (expand_twoval_binop_libfunc): Likewise.
15818 * stor-layout.c (layout_type): Likewise.
15819 (initialize_sizetypes): Likewise.
15820 * targhooks.c (default_get_mask_mode): Likewise.
15821 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
15823 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15824 Alan Hayward <alan.hayward@arm.com>
15825 David Sherwood <david.sherwood@arm.com>
15827 * machmode.h (opt_mode::else_blk): New function.
15828 (int_mode_for_mode): Declare.
15829 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
15830 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
15832 * cfgexpand.c (expand_debug_expr): Likewise.
15833 * combine.c (gen_lowpart_or_truncate): Likewise.
15834 (gen_lowpart_for_combine): Likewise.
15835 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
15836 * config/avr/avr.c (avr_to_int_mode): Likewise.
15837 (avr_out_plus_1): Likewise.
15838 (avr_out_plus): Likewise.
15839 (avr_out_round): Likewise.
15840 * config/i386/i386.c (ix86_split_to_parts): Likewise.
15841 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15842 (s390_expand_vcond): Likewise.
15843 * config/spu/spu.c (spu_split_immediate): Likewise.
15844 (spu_expand_mov): Likewise.
15845 * dse.c (get_stored_val): Likewise.
15846 * expmed.c (store_bit_field_1): Likewise.
15847 (convert_extracted_bit_field): Use int_mode_for_mode instead of
15849 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
15850 (extract_low_bits): Likewise.
15851 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
15852 handling rather than repeating the check.
15853 (emit_group_store): Likewise.
15854 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
15855 * optabs.c (expand_absneg_bit): Likewise.
15856 (expand_copysign_absneg): Likewise.
15857 (expand_copysign_bit): Likewise.
15858 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
15859 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
15860 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
15861 * var-tracking.c (prepare_call_arguments): Likewise.
15862 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
15863 int_mode_for_mode instead of mode_for_size.
15864 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15866 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15867 Alan Hayward <alan.hayward@arm.com>
15868 David Sherwood <david.sherwood@arm.com>
15870 * machmode.h (int_mode_for_size): New function.
15871 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
15872 instead of mode_for_size.
15873 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
15875 * combine.c (expand_field_assignment): Use int_mode_for_size
15876 instead of mode_for_size.
15877 (make_extraction): Likewise.
15878 (simplify_shift_const_1): Likewise.
15879 (simplify_comparison): Likewise.
15880 * dojump.c (do_jump): Likewise.
15881 * dwarf2out.c (mem_loc_descriptor): Likewise.
15882 * emit-rtl.c (init_derived_machine_modes): Likewise.
15883 * expmed.c (flip_storage_order): Likewise.
15884 (convert_extracted_bit_field): Likewise.
15885 * expr.c (copy_blkmode_from_reg): Likewise.
15886 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
15887 * internal-fn.c (expand_mul_overflow): Likewise.
15888 * lower-subreg.c (simple_move): Likewise.
15889 * optabs-libfuncs.c (init_optabs): Likewise.
15890 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15891 * tree.c (vector_type_mode): Likewise.
15892 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
15893 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
15894 * tree-vect-generic.c (expand_vector_parallel): Likewise.
15895 * tree-vect-stmts.c (vectorizable_load): Likewise.
15896 (vectorizable_store): Likewise.
15898 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15899 Alan Hayward <alan.hayward@arm.com>
15900 David Sherwood <david.sherwood@arm.com>
15902 * coretypes.h (pod_mode): New type.
15903 (scalar_int_mode_pod): New typedef.
15904 * machmode.h (pod_mode): New class.
15905 (int_n_data_t::m): Change type to scalar_int_mode_pod.
15906 * genmodes.c (emit_mode_int_n): Update accordingly.
15907 * lower-subreg.h (target_lower_subreg): Change type to
15908 scalar_int_mode_pod.
15909 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
15910 scalar_int_mode_pod.
15912 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15913 Alan Hayward <alan.hayward@arm.com>
15914 David Sherwood <david.sherwood@arm.com>
15916 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
15917 machine_mode to scalar_int_mode.
15918 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
15919 (rs6000_option_override_internal): Remove cast to int.
15920 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
15921 machine_mode to scalar_int_mode.
15922 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
15923 (rs6000_option_override_internal): Remove cast to int.
15924 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
15925 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
15927 * config/s390/s390.c (s390_expand_builtin): Likewise.
15928 * coretypes.h (scalar_int_mode): New type.
15929 (opt_scalar_int_mode): New typedef.
15930 * machmode.h (scalar_int_mode): New class.
15931 (scalar_int_mode::includes_p): New function.
15932 (byte_mode): Change type to scalar_int_mode.
15933 (word_mode): Likewise.
15934 (ptr_mode): Likewise.
15935 * emit-rtl.c (byte_mode): Likewise.
15936 (word_mode): Likewise.
15937 (ptr_mode): Likewise.
15938 (init_derived_machine_modes): Update accordingly.
15939 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
15940 and MODE_PARTIAL_INT.
15941 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
15942 opt_scalar_int_mode.
15944 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15945 Alan Hayward <alan.hayward@arm.com>
15946 David Sherwood <david.sherwood@arm.com>
15948 * target.def (libgcc_floating_mode_supported_p): Take a
15950 * doc/tm.texi: Regenerate.
15951 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
15953 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
15954 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
15957 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15958 Alan Hayward <alan.hayward@arm.com>
15959 David Sherwood <david.sherwood@arm.com>
15961 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
15962 * doc/tm.texi: Regenerate.
15963 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
15964 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
15965 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
15966 * targhooks.h (default_floatn_mode): Likewise.
15967 * targhooks.c (default_floatn_mode): Likewise.
15968 * tree.c (build_common_tree_nodes): Update accordingly.
15970 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15971 Alan Hayward <alan.hayward@arm.com>
15972 David Sherwood <david.sherwood@arm.com>
15974 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
15975 (mode_iterator::iterate_p): Likewise.
15976 (mode_iterator::get_wider): Likewise.
15977 * expr.c (init_expr_target): Use opt_scalar_float_mode.
15979 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15980 Alan Hayward <alan.hayward@arm.com>
15981 David Sherwood <david.sherwood@arm.com>
15983 * coretypes.h (opt_scalar_float_mode): New typedef.
15984 * machmode.h (float_mode_for_size): New function.
15985 * emit-rtl.c (double_mode): Delete.
15986 (init_emit_once): Use float_mode_for_size.
15987 * stor-layout.c (layout_type): Likewise.
15988 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
15990 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15991 Alan Hayward <alan.hayward@arm.com>
15992 David Sherwood <david.sherwood@arm.com>
15994 * output.h (assemble_real): Take a scalar_float_mode.
15995 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
15996 * config/arm/arm.md (consttable_4): Likewise.
15997 (consttable_8): Likewise.
15998 (consttable_16): Likewise.
15999 * config/mips/mips.md (consttable_float): Likewise.
16000 * config/s390/s390.c (s390_output_pool_entry): Likewise.
16001 * varasm.c (assemble_real): Take a scalar_float_mode.
16002 (output_constant_pool_2): Update accordingly.
16003 (output_constant): Likewise.
16005 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16006 Alan Hayward <alan.hayward@arm.com>
16007 David Sherwood <david.sherwood@arm.com>
16009 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
16010 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
16011 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
16012 (native_encode_real): Likewise.
16013 (native_interpret_real): Likewise.
16014 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
16015 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16017 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16018 Alan Hayward <alan.hayward@arm.com>
16019 David Sherwood <david.sherwood@arm.com>
16021 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
16022 <scalar_float_mode>. Simplify.
16023 (gen_extend_conv_libfunc): Likewise.
16025 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16026 Alan Hayward <alan.hayward@arm.com>
16027 David Sherwood <david.sherwood@arm.com>
16029 * coretypes.h (scalar_float_mode): New type.
16030 * machmode.h (mode_traits::from_int): Use machine_mode if
16031 USE_ENUM_MODES is defined.
16032 (is_a): New function.
16034 (dyn_cast): Likewise.
16035 (scalar_float_mode): New class.
16036 (scalar_float_mode::includes_p): New function.
16037 (is_float_mode): Likewise.
16038 * gdbhooks.py (MachineModePrinter): New class.
16039 (build_pretty_printer): Use it for scalar_float_mode.
16040 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
16041 (format_helper::format_helper): Turn into a template.
16042 * genmodes.c (get_mode_class): New function.
16043 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
16044 or machine_mode if none.
16045 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
16046 as_a <scalar_float_mode>.
16047 * dwarf2out.c (mem_loc_descriptor): Likewise.
16048 (insert_float): Likewise.
16049 (add_const_value_attribute): Likewise.
16050 * simplify-rtx.c (simplify_immed_subreg): Likewise.
16051 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
16052 (expand_unop): Update accordingly.
16053 (expand_abs_nojump): Likewise.
16054 (expand_copysign_absneg): Take a scalar_float_mode.
16055 (expand_copysign_bit): Likewise.
16056 (expand_copysign): Update accordingly.
16058 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16059 Alan Hayward <alan.hayward@arm.com>
16060 David Sherwood <david.sherwood@arm.com>
16062 * coretypes.h (opt_mode): New class.
16063 * machmode.h (opt_mode): Likewise.
16064 (opt_mode::else_void): New function.
16065 (opt_mode::require): Likewise.
16066 (opt_mode::exists): Likewise.
16067 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
16068 (GET_MODE_2XWIDER_MODE): Likewise.
16069 (mode_iterator::get_wider): Update accordingly.
16070 (mode_iterator::get_2xwider): Likewise.
16071 (mode_iterator::get_known_wider): Likewise, turning into a template.
16072 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
16073 forcing a wider mode to exist.
16074 * config/cr16/cr16.h (LONG_REG_P): Likewise.
16075 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16076 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
16077 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
16078 * lower-subreg.c (init_lower_subreg): Likewise.
16079 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
16080 on the final iteration.
16081 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
16082 a wider mode exists before asking for a move pattern.
16083 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
16084 forcing a wider mode to exist.
16085 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
16086 returning false if no such mode exists.
16087 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
16088 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
16089 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
16090 Avoid checking for a MODE_INT if we already know the mode is not a
16092 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
16093 forcing a wider mode to exist.
16094 (expmed_mult_highpart_optab): Likewise.
16095 (expmed_mult_highpart): Likewise.
16096 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
16098 * lto-streamer-in.c (lto_input_mode_table): Likewise.
16099 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
16100 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16101 * internal-fn.c (expand_mul_overflow): Update use of
16102 GET_MODE_2XWIDER_MODE.
16103 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16104 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
16105 GET_MODE_WIDER_MODE.
16106 (convert_plusminus_to_widen): Likewise.
16107 * tree-switch-conversion.c (array_value_type): Likewise.
16108 * var-tracking.c (emit_note_insn_var_location): Likewise.
16109 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16110 Return false inside rather than outside the loop if no wider mode
16112 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
16113 and GET_MODE_2XWIDER_MODE
16114 (can_compare_p): Use else_void.
16115 * gdbhooks.py (OptMachineModePrinter): New class.
16116 (build_pretty_printer): Use it for opt_mode.
16118 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16119 Alan Hayward <alan.hayward@arm.com>
16120 David Sherwood <david.sherwood@arm.com>
16122 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
16123 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
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 * machmode.h (mode_traits): New structure.
16130 (get_narrowest_mode): New function.
16131 (mode_iterator::start): Likewise.
16132 (mode_iterator::iterate_p): Likewise.
16133 (mode_iterator::get_wider): Likewise.
16134 (mode_iterator::get_known_wider): Likewise.
16135 (mode_iterator::get_2xwider): Likewise.
16136 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
16137 (FOR_EACH_MODE): Likewise.
16138 (FOR_EACH_MODE_FROM): Likewise.
16139 (FOR_EACH_MODE_UNTIL): Likewise.
16140 (FOR_EACH_WIDER_MODE): Likewise.
16141 (FOR_EACH_2XWIDER_MODE): Likewise.
16142 * builtins.c (expand_builtin_strlen): Use new mode iterators.
16143 * combine.c (simplify_comparison): Likewise
16144 * config/i386/i386.c (type_natural_mode): Likewise.
16145 * cse.c (cse_insn): Likewise.
16146 * dse.c (find_shift_sequence): Likewise.
16147 * emit-rtl.c (init_derived_machine_modes): Likewise.
16148 (init_emit_once): Likewise.
16149 * explow.c (hard_function_value): Likewise.
16150 * expmed.c (extract_fixed_bit_field_1): Likewise.
16151 (extract_bit_field_1): Likewise.
16152 (expand_divmod): Likewise.
16153 (emit_store_flag_1): Likewise.
16154 * expr.c (init_expr_target): Likewise.
16155 (convert_move): Likewise.
16156 (alignment_for_piecewise_move): Likewise.
16157 (widest_int_mode_for_size): Likewise.
16158 (emit_block_move_via_movmem): Likewise.
16159 (copy_blkmode_to_reg): Likewise.
16160 (set_storage_via_setmem): Likewise.
16161 (compress_float_constant): Likewise.
16162 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16163 * optabs-query.c (get_best_extraction_insn): Likewise.
16164 * optabs.c (expand_binop): Likewise.
16165 (expand_twoval_unop): Likewise.
16166 (expand_twoval_binop): Likewise.
16167 (widen_leading): Likewise.
16168 (widen_bswap): Likewise.
16169 (expand_parity): Likewise.
16170 (expand_unop): Likewise.
16171 (prepare_cmp_insn): Likewise.
16172 (prepare_float_lib_cmp): Likewise.
16173 (expand_float): Likewise.
16174 (expand_fix): Likewise.
16175 (expand_sfix_optab): Likewise.
16176 * postreload.c (move2add_use_add2_insn): Likewise.
16177 * reg-stack.c (reg_to_stack): Likewise.
16178 * reginfo.c (choose_hard_reg_mode): Likewise.
16179 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16180 * stor-layout.c (mode_for_size): Likewise.
16181 (smallest_mode_for_size): Likewise.
16182 (mode_for_vector): Likewise.
16183 (finish_bitfield_representative): Likewise.
16184 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
16185 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
16186 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16187 * var-tracking.c (prepare_call_arguments): Likewise.
16189 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16190 Alan Hayward <alan.hayward@arm.com>
16191 David Sherwood <david.sherwood@arm.com>
16193 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
16194 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
16195 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
16196 * machmode.h (mode_size): Move earlier in file.
16197 (mode_precision): Likewise.
16198 (mode_inner): Likewise.
16199 (mode_nunits): Likewise.
16200 (mode_unit_size): Likewise.
16201 (unit_unit_precision): Likewise.
16202 (mode_wider): Likewise.
16203 (mode_2xwider): Likewise.
16204 (machine_mode): New class.
16205 (mode_to_bytes): New function.
16206 (mode_to_bits): Likewise.
16207 (mode_to_precision): Likewise.
16208 (mode_to_inner): Likewise.
16209 (mode_to_unit_size): Likewise.
16210 (mode_to_unit_precision): Likewise.
16211 (mode_to_nunits): Likewise.
16212 (GET_MODE_SIZE): Use mode_to_bytes.
16213 (GET_MODE_BITSIZE): Use mode_to_bits.
16214 (GET_MODE_PRECISION): Use mode_to_precision.
16215 (GET_MODE_INNER): Use mode_to_inner.
16216 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
16217 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
16218 (GET_MODE_NUNITS): Use mode_to_nunits.
16219 * system.h (ALWAYS_INLINE): New macro.
16220 * config/powerpcspe/powerpcspe-c.c
16221 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
16222 int for arg1_mode and arg2_mode.
16224 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16225 Alan Hayward <alan.hayward@arm.com>
16226 David Sherwood <david.sherwood@arm.com>
16228 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
16229 Prefix mode names with E_ in case statements.
16230 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16231 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
16232 (aarch64_split_simd_move): Likewise.
16233 (aarch64_gen_storewb_pair): Likewise.
16234 (aarch64_gen_loadwb_pair): Likewise.
16235 (aarch64_gen_store_pair): Likewise.
16236 (aarch64_gen_load_pair): Likewise.
16237 (aarch64_get_condition_code_1): Likewise.
16238 (aarch64_constant_pool_reload_icode): Likewise.
16239 (get_rsqrte_type): Likewise.
16240 (get_rsqrts_type): Likewise.
16241 (get_recpe_type): Likewise.
16242 (get_recps_type): Likewise.
16243 (aarch64_gimplify_va_arg_expr): Likewise.
16244 (aarch64_simd_container_mode): Likewise.
16245 (aarch64_emit_load_exclusive): Likewise.
16246 (aarch64_emit_store_exclusive): Likewise.
16247 (aarch64_expand_compare_and_swap): Likewise.
16248 (aarch64_gen_atomic_cas): Likewise.
16249 (aarch64_emit_bic): Likewise.
16250 (aarch64_emit_atomic_swap): Likewise.
16251 (aarch64_emit_atomic_load_op): Likewise.
16252 (aarch64_evpc_trn): Likewise.
16253 (aarch64_evpc_uzp): Likewise.
16254 (aarch64_evpc_zip): Likewise.
16255 (aarch64_evpc_ext): Likewise.
16256 (aarch64_evpc_rev): Likewise.
16257 (aarch64_evpc_dup): Likewise.
16258 (aarch64_gen_ccmp_first): Likewise.
16259 (aarch64_gen_ccmp_next): Likewise.
16260 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
16261 (alpha_emit_xfloating_libcall): Likewise.
16262 (emit_insxl): Likewise.
16263 (alpha_arg_type): Likewise.
16264 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
16265 (arc_preferred_simd_mode): Likewise.
16266 (arc_secondary_reload): Likewise.
16267 (get_arc_condition_code): Likewise.
16268 (arc_print_operand): Likewise.
16269 (arc_legitimate_constant_p): Likewise.
16270 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16271 * config/arc/arc.md (casesi_load): Likewise.
16272 (casesi_compact_jump): Likewise.
16273 * config/arc/predicates.md (proper_comparison_operator): Likewise.
16274 (cc_use_register): Likewise.
16275 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16276 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
16277 (arm_init_iwmmxt_builtins): Likewise.
16278 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
16279 (neon_expand_vector_init): Likewise.
16280 (arm_attr_length_move_neon): Likewise.
16281 (maybe_get_arm_condition_code): Likewise.
16282 (arm_emit_vector_const): Likewise.
16283 (arm_preferred_simd_mode): Likewise.
16284 (arm_output_iwmmxt_tinsr): Likewise.
16285 (thumb1_output_casesi): Likewise.
16286 (thumb2_output_casesi): Likewise.
16287 (arm_emit_load_exclusive): Likewise.
16288 (arm_emit_store_exclusive): Likewise.
16289 (arm_expand_compare_and_swap): Likewise.
16290 (arm_evpc_neon_vuzp): Likewise.
16291 (arm_evpc_neon_vzip): Likewise.
16292 (arm_evpc_neon_vrev): Likewise.
16293 (arm_evpc_neon_vtrn): Likewise.
16294 (arm_evpc_neon_vext): Likewise.
16295 (arm_validize_comparison): Likewise.
16296 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
16297 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
16298 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
16299 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
16300 (c6x_preferred_simd_mode): Likewise.
16301 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
16302 (epiphany_rtx_costs): Likewise.
16303 * config/epiphany/predicates.md (proper_comparison_operator):
16305 * config/frv/frv.c (condexec_memory_operand): Likewise.
16306 (frv_emit_move): Likewise.
16307 (output_move_single): Likewise.
16308 (output_condmove_single): Likewise.
16309 (frv_hard_regno_mode_ok): Likewise.
16310 (frv_matching_accg_mode): Likewise.
16311 * config/h8300/h8300.c (split_adds_subs): Likewise.
16312 (h8300_rtx_costs): Likewise.
16313 (h8300_print_operand): Likewise.
16314 (compute_mov_length): Likewise.
16315 (output_logical_op): Likewise.
16316 (compute_logical_op_length): Likewise.
16317 (compute_logical_op_cc): Likewise.
16318 (h8300_shift_needs_scratch_p): Likewise.
16319 (output_a_shift): Likewise.
16320 (compute_a_shift_length): Likewise.
16321 (compute_a_shift_cc): Likewise.
16322 (expand_a_rotate): Likewise.
16323 (output_a_rotate): Likewise.
16324 * config/i386/i386.c (classify_argument): Likewise.
16325 (function_arg_advance_32): Likewise.
16326 (function_arg_32): Likewise.
16327 (function_arg_64): Likewise.
16328 (function_value_64): Likewise.
16329 (ix86_gimplify_va_arg): Likewise.
16330 (ix86_legitimate_constant_p): Likewise.
16331 (put_condition_code): Likewise.
16332 (split_double_mode): Likewise.
16333 (ix86_avx256_split_vector_move_misalign): Likewise.
16334 (ix86_expand_vector_logical_operator): Likewise.
16335 (ix86_split_idivmod): Likewise.
16336 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
16337 (ix86_build_const_vector): Likewise.
16338 (ix86_build_signbit_mask): Likewise.
16339 (ix86_match_ccmode): Likewise.
16340 (ix86_cc_modes_compatible): Likewise.
16341 (ix86_expand_branch): Likewise.
16342 (ix86_expand_sse_cmp): Likewise.
16343 (ix86_expand_sse_movcc): Likewise.
16344 (ix86_expand_int_sse_cmp): Likewise.
16345 (ix86_expand_vec_perm_vpermi2): Likewise.
16346 (ix86_expand_vec_perm): Likewise.
16347 (ix86_expand_sse_unpack): Likewise.
16348 (ix86_expand_int_addcc): Likewise.
16349 (ix86_split_to_parts): Likewise.
16350 (ix86_vectorize_builtin_gather): Likewise.
16351 (ix86_vectorize_builtin_scatter): Likewise.
16352 (avx_vpermilp_parallel): Likewise.
16353 (inline_memory_move_cost): Likewise.
16354 (ix86_tieable_integer_mode_p): Likewise.
16355 (x86_maybe_negate_const_int): Likewise.
16356 (ix86_expand_vector_init_duplicate): Likewise.
16357 (ix86_expand_vector_init_one_nonzero): Likewise.
16358 (ix86_expand_vector_init_one_var): Likewise.
16359 (ix86_expand_vector_init_concat): Likewise.
16360 (ix86_expand_vector_init_interleave): Likewise.
16361 (ix86_expand_vector_init_general): Likewise.
16362 (ix86_expand_vector_set): Likewise.
16363 (ix86_expand_vector_extract): Likewise.
16364 (emit_reduc_half): Likewise.
16365 (ix86_emit_i387_round): Likewise.
16366 (ix86_mangle_type): Likewise.
16367 (ix86_expand_round_sse4): Likewise.
16368 (expand_vec_perm_blend): Likewise.
16369 (canonicalize_vector_int_perm): Likewise.
16370 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
16371 (expand_vec_perm_1): Likewise.
16372 (expand_vec_perm_interleave3): Likewise.
16373 (expand_vec_perm_even_odd_pack): Likewise.
16374 (expand_vec_perm_even_odd_1): Likewise.
16375 (expand_vec_perm_broadcast_1): Likewise.
16376 (ix86_vectorize_vec_perm_const_ok): Likewise.
16377 (ix86_expand_vecop_qihi): Likewise.
16378 (ix86_expand_mul_widen_hilo): Likewise.
16379 (ix86_expand_sse2_abs): Likewise.
16380 (ix86_expand_pextr): Likewise.
16381 (ix86_expand_pinsr): Likewise.
16382 (ix86_preferred_simd_mode): Likewise.
16383 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
16384 * config/i386/sse.md (*andnot<mode>3): Likewise.
16385 (<mask_codefor><code><mode>3<mask_name>): Likewise.
16386 (*<code><mode>3): Likewise.
16387 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
16388 (ia64_expand_atomic_op): Likewise.
16389 (ia64_arg_type): Likewise.
16390 (ia64_mode_to_int): Likewise.
16391 (ia64_scalar_mode_supported_p): Likewise.
16392 (ia64_vector_mode_supported_p): Likewise.
16393 (expand_vec_perm_broadcast): Likewise.
16394 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
16395 (iq2000_function_arg_advance): Likewise.
16396 (iq2000_function_arg): Likewise.
16397 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
16398 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
16399 (m68k_libcall_value): Likewise.
16400 (m68k_function_value): Likewise.
16401 (sched_attr_op_type): Likewise.
16402 * config/mcore/mcore.c (mcore_output_move): Likewise.
16403 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
16405 (microblaze_function_arg): Likewise.
16406 * config/mips/mips.c (mips16_build_call_stub): Likewise.
16407 (mips_print_operand): Likewise.
16408 (mips_mode_ok_for_mov_fmt_p): Likewise.
16409 (mips_vector_mode_supported_p): Likewise.
16410 (mips_preferred_simd_mode): Likewise.
16411 (mips_expand_vpc_loongson_even_odd): Likewise.
16412 (mips_expand_vec_unpack): Likewise.
16413 (mips_expand_vi_broadcast): Likewise.
16414 (mips_expand_vector_init): Likewise.
16415 (mips_expand_vec_reduc): Likewise.
16416 (mips_expand_msa_cmp): Likewise.
16417 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
16418 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
16419 (cc_flags_for_mode): Likewise.
16420 * config/msp430/msp430.c (msp430_print_operand): Likewise.
16421 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
16422 (nds32_output_casesi_pc_relative): Likewise.
16423 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16424 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
16425 (nvptx_gen_unpack): Likewise.
16426 (nvptx_gen_pack): Likewise.
16427 (nvptx_gen_shuffle): Likewise.
16428 (nvptx_gen_wcast): Likewise.
16429 (nvptx_preferred_simd_mode): Likewise.
16430 * config/pa/pa.c (pa_secondary_reload): Likewise.
16431 * config/pa/predicates.md (base14_operand): Likewise.
16432 * config/powerpcspe/powerpcspe-c.c
16433 (altivec_resolve_overloaded_builtin): Likewise.
16434 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
16436 (rs6000_preferred_simd_mode): Likewise.
16437 (output_vec_const_move): Likewise.
16438 (rs6000_expand_vector_extract): Likewise.
16439 (rs6000_split_vec_extract_var): Likewise.
16440 (reg_offset_addressing_ok_p): Likewise.
16441 (rs6000_legitimate_offset_address_p): Likewise.
16442 (rs6000_legitimize_address): Likewise.
16443 (rs6000_emit_set_const): Likewise.
16444 (rs6000_const_vec): Likewise.
16445 (rs6000_emit_move): Likewise.
16446 (spe_build_register_parallel): Likewise.
16447 (rs6000_darwin64_record_arg_recurse): Likewise.
16448 (swap_selector_for_mode): Likewise.
16449 (spe_init_builtins): Likewise.
16450 (paired_init_builtins): Likewise.
16451 (altivec_init_builtins): Likewise.
16452 (do_load_for_compare): Likewise.
16453 (rs6000_generate_compare): Likewise.
16454 (rs6000_expand_float128_convert): Likewise.
16455 (emit_load_locked): Likewise.
16456 (emit_store_conditional): Likewise.
16457 (rs6000_output_function_epilogue): Likewise.
16458 (rs6000_handle_altivec_attribute): Likewise.
16459 (rs6000_function_value): Likewise.
16460 (emit_fusion_gpr_load): Likewise.
16461 (emit_fusion_p9_load): Likewise.
16462 (emit_fusion_p9_store): Likewise.
16463 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
16464 (fusion_gpr_mem_load): Likewise.
16465 (fusion_addis_mem_combo_load): Likewise.
16466 (fusion_addis_mem_combo_store): Likewise.
16467 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
16468 (fusion_gpr_mem_load): Likewise.
16469 (fusion_addis_mem_combo_load): Likewise.
16470 (fusion_addis_mem_combo_store): Likewise.
16471 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16473 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
16474 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16475 (rs6000_preferred_simd_mode): Likewise.
16476 (output_vec_const_move): Likewise.
16477 (rs6000_expand_vector_extract): Likewise.
16478 (rs6000_split_vec_extract_var): Likewise.
16479 (reg_offset_addressing_ok_p): Likewise.
16480 (rs6000_legitimate_offset_address_p): Likewise.
16481 (rs6000_legitimize_address): Likewise.
16482 (rs6000_emit_set_const): Likewise.
16483 (rs6000_const_vec): Likewise.
16484 (rs6000_emit_move): Likewise.
16485 (rs6000_darwin64_record_arg_recurse): Likewise.
16486 (swap_selector_for_mode): Likewise.
16487 (paired_init_builtins): Likewise.
16488 (altivec_init_builtins): Likewise.
16489 (rs6000_expand_float128_convert): Likewise.
16490 (emit_load_locked): Likewise.
16491 (emit_store_conditional): Likewise.
16492 (rs6000_output_function_epilogue): Likewise.
16493 (rs6000_handle_altivec_attribute): Likewise.
16494 (rs6000_function_value): Likewise.
16495 (emit_fusion_gpr_load): Likewise.
16496 (emit_fusion_p9_load): Likewise.
16497 (emit_fusion_p9_store): Likewise.
16498 * config/rx/rx.c (rx_gen_move_template): Likewise.
16499 (flags_from_mode): Likewise.
16500 * config/s390/predicates.md (s390_alc_comparison): Likewise.
16501 (s390_slb_comparison): Likewise.
16502 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
16503 (s390_vector_mode_supported_p): Likewise.
16504 (s390_cc_modes_compatible): Likewise.
16505 (s390_match_ccmode_set): Likewise.
16506 (s390_canonicalize_comparison): Likewise.
16507 (s390_emit_compare_and_swap): Likewise.
16508 (s390_branch_condition_mask): Likewise.
16509 (s390_rtx_costs): Likewise.
16510 (s390_secondary_reload): Likewise.
16511 (__SECONDARY_RELOAD_CASE): Likewise.
16512 (s390_expand_cs): Likewise.
16513 (s390_preferred_simd_mode): Likewise.
16514 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
16515 * config/sh/sh.c (sh_print_operand): Likewise.
16516 (dump_table): Likewise.
16517 (sh_secondary_reload): Likewise.
16518 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16519 * config/sh/sh.md (casesi_worker_1): Likewise.
16520 (casesi_worker_2): Likewise.
16521 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
16522 (fcc_comparison_operator): Likewise.
16523 * config/sparc/sparc.c (sparc_expand_move): Likewise.
16524 (emit_soft_tfmode_cvt): Likewise.
16525 (sparc_preferred_simd_mode): Likewise.
16526 (output_cbranch): Likewise.
16527 (sparc_print_operand): Likewise.
16528 (sparc_expand_vec_perm_bmask): Likewise.
16529 (vector_init_bshuffle): Likewise.
16530 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
16531 (spu_vector_mode_supported_p): Likewise.
16532 (spu_expand_insv): Likewise.
16533 (spu_emit_branch_or_set): Likewise.
16534 (spu_handle_vector_attribute): Likewise.
16535 (spu_builtin_splats): Likewise.
16536 (spu_builtin_extract): Likewise.
16537 (spu_builtin_promote): Likewise.
16538 (spu_expand_sign_extend): Likewise.
16539 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
16540 (tilegx_simd_int): Likewise.
16541 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
16542 (tilepro_simd_int): Likewise.
16543 * config/v850/v850.c (const_double_split): Likewise.
16544 (v850_print_operand): Likewise.
16545 (ep_memory_offset): Likewise.
16546 * config/vax/vax.c (vax_rtx_costs): Likewise.
16547 (vax_output_int_move): Likewise.
16548 (vax_output_int_add): Likewise.
16549 (vax_output_int_subtract): Likewise.
16550 * config/visium/predicates.md (visium_branch_operator): Likewise.
16551 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
16552 (visium_print_operand_address): Likewise.
16553 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16554 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
16555 (xtensa_expand_conditional_branch): Likewise.
16556 (xtensa_copy_incoming_a7): Likewise.
16557 (xtensa_output_literal): Likewise.
16558 * dfp.c (decimal_real_maxval): Likewise.
16559 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16561 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16562 Alan Hayward <alan.hayward@arm.com>
16563 David Sherwood <david.sherwood@arm.com>
16565 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
16566 (mode_nunits_inline): Likewise.
16567 (mode_inner_inline): Likewise.
16568 (mode_unit_size_inline): Likewise.
16569 (mode_unit_precision_inline): Likewise.
16570 (emit_insn_modes_h): Likewise. Also emit a #define of the
16572 (emit_mode_wider): Add an E_ prefix to mode names.
16573 (emit_mode_complex): Likewise.
16574 (emit_mode_inner): Likewise.
16575 (emit_mode_adjustments): Likewise.
16576 (emit_mode_int_n): Likewise.
16577 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
16578 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
16579 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
16580 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
16581 (CRC32_BUILTIN, ENTRY): Likewise.
16582 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
16583 (aarch64_pop_regs): Likewise.
16584 (aarch64_process_components): Likewise.
16585 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
16586 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
16587 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
16588 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
16589 * config/arm/arm.c (arm_init_libfuncs): Likewise.
16590 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
16592 * config/i386/i386-builtin.def (pcmpestr): Likewise.
16593 (pcmpistr): Likewise.
16594 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
16595 * config/mmix/mmix.c (mmix_output_condition): Likewise.
16596 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
16598 * config/rl78/rl78.c (mduc_regs): Likewise.
16599 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16600 (htm_expand_builtin): Likewise.
16601 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
16602 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
16603 * config/xtensa/xtensa.c (print_operand): Likewise.
16604 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
16605 (NUM_MODE_VECTOR_INT): Likewise.
16606 * genoutput.c (null_operand): Likewise.
16607 (output_operand_data): Likewise.
16608 * genrecog.c (print_parameter_value): Likewise.
16609 * lra.c (debug_operand_data): Likewise.
16611 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16613 * dwarf2out.c (scompare_loc_descriptor_wide)
16614 (scompare_loc_descriptor_narrow): New functions, split out from...
16615 (scompare_loc_descriptor): ...here.
16616 * expmed.c (emit_store_flag_int): New function, split out from...
16617 (emit_store_flag): ...here.
16619 2017-08-30 Richard Biener <rguenther@suse.de>
16621 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
16622 (dwarf2out_early_finish): Move setting of AT_pubnames from
16623 early debug output to early finish.
16625 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16627 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
16628 and -mdata-region to the assembler.
16630 2017-08-30 Richard Biener <rguenther@suse.de>
16632 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
16634 (gen_subprogram_die): Add DW_AT_object_pointer only early.
16635 (dwarf2out_early_global_decl): Only generate a DIE for the
16636 abstract origin if it doesn't already exist or is a declaration DIE.
16637 (resolve_addr): Do not add the linkage name twice when
16638 generating a stub DIE for the DW_TAG_GNU_call_site target.
16640 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16642 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16643 Use machine_mode rather than int for arg1_mode.
16645 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
16648 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
16649 that the second argument of the built-in functions to unpack
16650 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
16651 switch statement instead a lot of if statements.
16652 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
16653 Allow 64-bit values to be in Altivec registers as well as
16654 traditional floating point registers.
16655 (pack<mode>, FMOVE128_VSX iterator): Likewise.
16657 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
16659 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
16660 MAX_REGS_PER_ADDRESS == 1.
16662 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
16664 * config/i386/i386.opt (flag_fentry): Do not init to -1.
16665 * config/i386/i386.c (ix86_option_override_internal): Simplify
16666 setting of opts->x_flag_entry.
16668 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16669 Jakub Jelinek <jakub@redhat.com>
16670 Richard Biener <rguenther@suse.de>
16672 PR tree-optimization/81503
16673 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
16674 folded constant fits in the target type; reorder tests for clarity.
16676 2017-08-29 Martin Liska <mliska@suse.cz>
16678 * passes.def: Include pass_lower_switch.
16679 * stmt.c (dump_case_nodes): Remove and move to
16680 tree-switch-conversion.
16681 (case_values_threshold): Likewise.
16682 (expand_switch_as_decision_tree_p): Likewise.
16683 (emit_case_decision_tree): Likewise.
16684 (expand_case): Likewise.
16685 (balance_case_nodes): Likewise.
16686 (node_has_low_bound): Likewise.
16687 (node_has_high_bound): Likewise.
16688 (node_is_bounded): Likewise.
16689 (emit_case_nodes): Likewise.
16690 (struct simple_case_node): New struct.
16691 (add_case_node): Remove.
16692 (emit_case_dispatch_table): Use vector instead of case_list.
16693 (reset_out_edges_aux): Remove.
16694 (compute_cases_per_edge): Likewise.
16695 (expand_case): Build list of simple_case_node.
16696 (expand_sjlj_dispatch_table): Use it.
16697 * tree-switch-conversion.c (struct case_node): Moved from
16698 stmt.c and adjusted.
16699 (emit_case_nodes): Likewise.
16700 (node_has_low_bound): Likewise.
16701 (node_has_high_bound): Likewise.
16702 (node_is_bounded): Likewise.
16703 (case_values_threshold): Likewise.
16704 (reset_out_edges_aux): Likewise.
16705 (compute_cases_per_edge): Likewise.
16706 (add_case_node): Likewise.
16707 (dump_case_nodes): Likewise.
16708 (balance_case_nodes): Likewise.
16709 (expand_switch_as_decision_tree_p): Likewise.
16710 (emit_jump): Likewise.
16711 (emit_case_decision_tree): Likewise.
16712 (try_switch_expansion): Likewise.
16713 (do_jump_if_equal): Likewise.
16714 (emit_cmp_and_jump_insns): Likewise.
16715 (fix_phi_operands_for_edge): New function.
16716 (record_phi_operand_mapping): Likewise.
16717 (class pass_lower_switch): New pass.
16718 (pass_lower_switch::execute): New function.
16719 (make_pass_lower_switch): Likewise.
16720 (conditional_probability):
16721 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
16722 * tree-pass.h: Add make_pass_lower_switch.
16724 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16727 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
16730 2017-08-29 Richard Biener <rguenther@suse.de>
16732 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
16733 we do not add a DW_AT_inline attribute twice.
16734 (gen_subprogram_die): Remove code setting DW_AT_inline on
16735 DECL_ABSTRACT_P nodes.
16737 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
16739 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
16740 calls to internal functions.
16741 (gimplify_modify_expr): Likewise.
16742 * tree-call-cdce.c (use_internal_fn): Likewise.
16743 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
16744 (convert_to_divmod): Set the nothrow flag.
16745 * tree-if-conv.c (predicate_mem_writes): Likewise.
16746 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16747 (vectorizable_call): Likewise.
16748 (vectorizable_store): Likewise.
16749 (vectorizable_load): Likewise.
16750 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
16751 (vect_recog_mask_conversion_pattern): Likewise.
16753 2017-08-29 Martin Liska <mliska@suse.cz>
16756 * gcc.c (driver_handle_option): Add new argument.
16757 * opts-common.c (handle_option): Pass
16758 target_option_override_hook.
16759 * opts-global.c (lang_handle_option): Add new option.
16760 (set_default_handlers): Add new argument.
16761 (decode_options): Likewise.
16762 * opts.c (target_handle_option): Likewise.
16763 (common_handle_option): Call target_option_override_hook.
16764 * opts.h (struct cl_option_handler_func): Add hook for
16765 target option override.
16766 (struct cl_option_handlers): Likewise.
16767 (set_default_handlers): Add new argument.
16768 (decode_options): Likewise.
16769 (common_handle_option): Likewise.
16770 (target_handle_option): Likewise.
16771 * toplev.c (toplev::main): Pass targetm.target_option.override
16774 2017-08-29 Richard Biener <rguenther@suse.de>
16775 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
16777 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
16778 life to the active subtree.
16780 2017-08-28 Jeff Law <law@redhat.com>
16782 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
16783 derive_equivalences.
16784 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
16785 Code moved into....
16786 (edge_info::derive_equivalences): New private member function
16788 * tree-ssa-dom.c (class edge_info): Changed from a struct
16789 to a class. Add ctor/dtor, methods and data members.
16790 (edge_info::edge_info): Renamed from allocate_edge_info.
16791 Initialize additional members.
16792 (edge_info::~edge_info): New.
16793 (free_dom_edge_info): Delete the edge info.
16794 (record_edge_info): Use new class & associated member functions.
16795 Tighten forms for testing for edge equivalences.
16796 (record_temporary_equivalences): Iterate over the simple
16797 equivalences rather than assuming there's only one per edge.
16798 (cprop_into_successor_phis): Iterate over the simple
16799 equivalences rather than assuming there's only one per edge.
16800 (optimize_stmt): Use operand_equal_p rather than pointer
16801 equality for mini-DSE code.
16803 2017-08-28 Nathan Sidwell <nathan@acm.org>
16805 * gcc.c (execute): Fold SIGPIPE handling into switch
16806 statement. Adjust internal error message.
16808 2017-08-28 Richard Biener <rguenther@suse.de>
16811 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
16812 Do nothing for removed DIEs.
16814 2017-08-28 Richard Biener <rguenther@suse.de>
16816 PR tree-optimization/81977
16817 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
16820 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
16823 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
16824 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
16825 using targetm.gen_mem_thread_fence.
16827 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
16830 * config/i386/i386.md (*<btsc><mode>): Change operand 2
16831 predicate to register_operand. Reorder operands.
16832 (*btr<mode>): Ditto.
16833 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
16834 (*btr<mode>_mask): Ditto.
16836 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
16838 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
16839 * config/rs6000/xmmintrin.h: New file.
16840 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
16842 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16845 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
16846 parameter and_insn and return it.
16847 (recombine_lvx_pattern): Insert a copy to ensure availability of
16848 the base register of the copied masking operation at the point of
16849 the instruction replacement.
16850 (recombine_stvx_pattern): Likewise.
16852 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
16854 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
16855 undocumented switches.
16856 (-mpower9-dform-vector): Likewise.
16857 (-mpower9-dform): Likewise.
16858 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
16859 comments to delete references to -mpower9-dform* switches.
16860 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
16861 Delete reference to -mpower9-dform* switches, test for
16862 -mpower9-vector instead.
16863 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
16864 (OTHER_P9_VECTOR_MASKS): Likewise.
16865 (POWERPC_MASKS): Likewise.
16866 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
16867 tests against -mpower9-dform* to -mpower9-vector. Delete code
16868 that checked for -mpower9-dform* consistancy with other options.
16869 Add test for -mpower9-misc to enable other power9 switches.
16870 (rs6000_init_hard_regno_mode_ok): Likewise.
16871 (rs6000_option_override_internal): Likewise.
16872 (rs6000_emit_prologue): Likewise.
16873 (rs6000_emit_epilogue): Likewise.
16874 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
16875 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
16876 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
16878 (emit_fusion_p9_store): Likewise.
16879 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
16880 resetting these macros if the assembler does not support ISA 3.0
16882 (TARGET_P9_DFORM_VECTOR): Likewise.
16883 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
16884 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
16886 2017-08-25 Alan Modra <amodra@gmail.com>
16888 PR rtl-optimization/81747
16889 * cse.c (cse_extended_basic_block): Don't attempt to record
16890 equivalences for degenerate conditional branches that branch
16891 to their fall-through.
16893 2017-08-24 Martin Sebor <msebor@redhat.com>
16895 PR middle-end/81908
16896 * gimple-fold.c (size_must_be_zero_p): New function.
16897 (gimple_fold_builtin_memory_op): Call it.
16899 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
16901 * config/rs6000/mm_malloc.h: New file.
16903 2017-08-24 Bin Cheng <bin.cheng@arm.com>
16905 PR tree-optimization/81913
16906 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
16907 analysis when either IVs in condition can wrap.
16909 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
16911 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
16912 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
16914 2017-08-24 Richard Biener <rguenther@suse.de>
16917 * targhooks.c (default_target_can_inline_p): Properly
16918 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
16919 is present and always compare.
16920 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
16921 infer -mfpmath=sse from TARGET_SSE_P.
16922 (ix86_can_inline_p): Properly use target_option_default_node when
16923 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
16925 2017-08-24 Richard Biener <rguenther@suse.de>
16928 * dwarf2out.c (output_die): Handle flag_generate_offload like
16930 (output_comp_unit): Likewise.
16931 (gen_array_type_die): Likewise.
16932 (dwarf2out_early_finish): Likewise.
16933 (note_variable_value_in_expr): Likewise.
16934 (dwarf2out_finish): Likewise. Adjust assert.
16935 * cgraphunit.c (symbol_table::compile): Move setting of
16936 flag_generate_offload earlier ...
16937 (symbol_table::finalize_compilation_unit): ... here, before
16938 early debug finalization.
16940 2017-08-24 Richard Biener <rguenther@suse.de>
16942 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
16943 and ipa-fnsummary.h.
16944 (ix86_can_inline_p): When ix86_fpmath flags do not match
16945 check whether the callee uses FP math at all.
16947 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
16949 PR middle-end/81931
16950 * tree-ssanames.c (get_nonzero_bits): Use element_precision
16951 instead of TYPE_PRECISION.
16953 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
16954 Alan Hayward <alan.hayward@arm.com>
16955 David Sherwood <david.sherwood@arm.com>
16957 * combine.c (make_extraction): Use subreg_offset_from_lsb.
16959 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
16961 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
16963 (ix86_frame::stack_realign_allocate): New field.
16964 (struct machine_frame_state): Modify comments.
16965 (machine_frame_state::sp_realigned_fp_end): New field.
16966 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
16967 layout calculation.
16968 (sp_valid_at): Add assertion to assure no attempt to access invalid
16969 offset of a realigned stack.
16970 (fp_valid_at): Likewise.
16971 (choose_baseaddr): Modify comments.
16972 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
16973 ix86_expand_prologue.
16974 (ix86_expand_prologue): Modify stack realignment and allocation.
16975 (ix86_expand_epilogue): Modify comments.
16976 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
16977 avx2_runtime, avx512f, and avx512f_runtime.
16979 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
16981 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
16982 (mstackrealign): Do not init to -1.
16983 * config/i386/i386.c (ix86_option_override_internal):
16984 Check opts_set, not opts when setting default value of
16985 opts->x_ix86_force_align_arg_pointer.
16987 2017-08-23 Richard Biener <rguenther@suse.de>
16989 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
16990 lang_hooks.decl_printable_name.
16991 * print-rtl-function.c (print_rtx_function): Likewise.
16992 * tree-pretty-print.c (dump_function_header): Likewise.
16994 2017-08-23 Richard Biener <rguenther@suse.de>
16997 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
16998 -g0 at compile-time.
17000 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17002 PR middle-end/19706
17003 * doc/sourcebuild.texi (Other hardware attributes):
17006 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17008 PR middle-end/19706
17009 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
17010 Fix single-use check.
17012 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
17014 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
17016 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17018 * doc/install.texi: Modify to add more details on running selected
17021 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17023 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
17024 is combined with -mabi=ms.
17025 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
17028 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17030 PR tree-optimization/81488
17031 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
17032 and cached_basis fields.
17033 (MAX_SPREAD): New constant.
17034 (alloc_cand_and_find_basis): Initialize new fields.
17035 (clear_visited): New function.
17036 (create_phi_basis_1): Rename from create_phi_basis, set visited
17037 and cached_basis fields.
17038 (create_phi_basis): New wrapper function.
17039 (phi_add_costs_1): Rename from phi_add_costs, add spread
17040 parameter, set visited field, short-circuit when limits reached.
17041 (phi_add_costs): New wrapper function.
17042 (record_phi_increments_1): Rename from record_phi_increments, set
17044 (record_phi_increments): New wrapper function.
17045 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
17046 (phi_incr_cost): New wrapper function.
17047 (all_phi_incrs_profitable_1): Rename from
17048 all_phi_incrs_profitable, set visited field.
17049 (all_phi_incrs_profitable): New wrapper function.
17051 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17052 Alan Hayward <alan.hayward@arm.com>
17053 David Sherwood <david.sherwood@arm.com>
17055 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
17056 that takes the outer and inner modes.
17057 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
17058 comparison as the canonical test for a paradoxical subreg.
17059 * combine.c (simplify_set): Use paradoxical_subreg_p.
17060 (make_extraction): Likewise.
17061 (force_to_mode): Likewise.
17062 (rtx_equal_for_field_assignment_p): Likewise.
17063 (gen_lowpart_for_combine): Likewise.
17064 (simplify_comparison): Likewise.
17065 * cse.c (equiv_constant): Likewise.
17066 * expmed.c (store_bit_field_1): Likewise.
17067 * final.c (alter_subreg): Likewise.
17068 * fwprop.c (propagate_rtx): Likewise.
17069 (forward_propagate_subreg): Likewise.
17070 * ira-conflicts.c (ira_build_conflicts): Likewise.
17071 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
17072 * lra-constraints.c (curr_insn_transform): Likewise.
17073 (split_reg): Likewise.
17074 * lra-eliminations.c (move_plus_up): Likewise.
17075 (lra_eliminate_regs_1): Likewise.
17076 * recog.c (general_operand): Likewise.
17077 * ree.c (combine_reaching_defs): Likewise.
17078 * reload.c (push_reload): Likewise.
17079 (find_reloads): Likewise.
17080 * reload1.c (elimination_effects): Likewise.
17081 (compute_reload_subreg_offset): Likewise.
17082 (choose_reload_regs): Likewise.
17083 * rtlanal.c (subreg_lsb_1): Likewise.
17084 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17085 (simplify_subreg): Likewise.
17086 * var-tracking.c (track_loc_p): Likewise.
17087 * emit-rtl.c (byte_lowpart_offset): Likewise.
17088 (paradoxical_subreg_p): Delete out-of-line definition.
17090 2017-08-22 Jeff Law <law@redhat.com>
17092 PR tree-optimization/81741
17093 PR tree-optimization/71947
17094 * tree-ssa-dom.c: Include tree-inline.h.
17095 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
17096 equivalences if one is more expensive to compute than the other.
17097 * tree-ssa-scopedtables.h (class const_or_copies): Make
17098 record_const_or_copy_raw method private.
17099 (class avail_exprs_stack): New method simplify_binary_operation.
17100 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
17101 avail_exprs_stack::simplify_binary_operation as needed.
17102 (avail_exprs_stack::simplify_binary_operation): New function.
17104 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
17106 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
17107 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
17108 (DOT_SYMBOLS): Likewise.
17109 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
17110 (RELOCATABLE_NEEDS_FIXUP): Likewise.
17111 (RS6000_ABI_NAME): Likewise.
17112 (TARGET_CMODEL): Likewise.
17113 (TOC_SECTION_ASM_OP): Likewise.
17114 (SET_CMODEL): New macro.
17115 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
17117 2017-08-22 Richard Biener <rguenther@suse.de>
17119 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
17120 to allow for free-lang-data replacements similar to verify_type_variant.
17122 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17123 Alan Hayward <alan.hayward@arm.com>
17124 David Sherwood <david.sherwood@arm.com>
17126 * config/aarch64/aarch64.md (casesi): Use DImode rather than
17127 VOIDmode for the LABEL_REF.
17129 2017-08-22 Richard Biener <rguenther@suse.de>
17131 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
17133 2017-08-22 Richard Biener <rguenther@suse.de>
17135 * common.opt (feliminate-dwarf2-dups): Ignore.
17136 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
17137 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
17138 same_die_p_wrap, compute_section_prefix,
17139 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
17140 (comdat_symbol_id, comdat_symbol_number): Likewise.
17141 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
17143 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
17144 (output_die): Mark unreachable path unreachable.
17145 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
17146 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
17147 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
17148 (dwarf2out_early_finish): Likewise.
17150 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
17152 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
17154 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
17157 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
17158 not VAR_P. Filter attribute warnings with OPT_Wattributes.
17159 (avr_attribute_table) <io, io_low, address>: Initialize
17160 .decl_required with true.
17162 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
17164 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
17165 undocumented debugging options.
17166 (-mvsx-scalar-double): Likewise.
17167 (-mallow-df-permute): Likewise.
17168 (-mvectorize-builtins): Likewise.
17169 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17170 (rs6000_builtin_vectorized_function): Likewise.
17171 (rs6000_builtin_md_vectorized_function): Likewise.
17172 (rs6000_opt_vars): Likewise.
17174 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
17177 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
17178 (*btrq_imm): Rename from *btrq.
17179 (*btcq_imm): Rename from *btcq.
17180 (btsc): New code attribute.
17181 (*<btsc><mode>): New insn pattern.
17182 (*btr<mode>): Ditto.
17183 (*<btsc><mode>_mask): New insn_and_split pattern.
17184 (*btr<mode>_mask): Ditto.
17186 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17187 Alan Hayward <alan.hayward@arm.com>
17188 David Sherwood <david.sherwood@arm.com>
17190 * function.c (pad_below): Simplify padding calculation.
17192 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17193 Alan Hayward <alan.hayward@arm.com>
17194 David Sherwood <david.sherwood@arm.com>
17196 * target.def (function_prologue): Remove frame size argument.
17197 (function_epilogue): Likewise.
17198 * doc/tm.texi: Regenerate.
17199 * final.c (final_start_function): Update call to function_prologue.
17200 (final_end_function): Update call to function_epilogue.
17201 (default_function_pro_epilogue): Remove frame size argument.
17202 * output.h (default_function_pro_epilogue): Likewise.
17203 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
17204 (arm_output_function_prologue): Likewise.
17205 * config/frv/frv.c (frv_function_prologue): Likewise.
17206 (frv_function_epilogue): Likewise.
17207 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
17208 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
17209 (ia64_output_function_epilogue): Likewise.
17210 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
17211 (m32r_output_function_epilogue): Likewise.
17212 * config/microblaze/microblaze.c (microblaze_function_prologue)
17213 (microblaze_function_epilogue): Likewise.
17214 * config/mips/mips.c (mips_output_function_prologue): Likewise.
17215 (mips_output_function_epilogue): Likewise.
17216 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
17217 (mmix_target_asm_function_epilogue): Likewise.
17218 * config/msp430/msp430.c (msp430_start_function): Likewise.
17219 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
17220 (nds32_asm_function_epilogue): Likewise.
17221 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
17222 * config/pa/pa.c (pa_output_function_prologue): Likewise.
17223 (pa_output_function_epilogue): Likewise.
17224 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
17225 (rs6000_output_function_epilogue): Likewise.
17226 * config/rl78/rl78.c (rl78_start_function): Likewise.
17227 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
17228 (rs6000_output_function_epilogue): Likewise.
17229 * config/rx/rx.c (rx_output_function_prologue): Likewise.
17230 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
17231 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
17232 (sparc_asm_function_epilogue): Likewise.
17234 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17236 * tree.h (type_has_mode_precision_p): New function.
17237 * convert.c (convert_to_integer_1): Use it.
17238 * expr.c (expand_expr_real_2): Likewise.
17239 (expand_expr_real_1): Likewise.
17240 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
17241 * match.pd: Likewise.
17242 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
17243 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
17244 * tree-tailcall.c (process_assignment): Likewise.
17245 * tree-vect-loop.c (vectorizable_reduction): Likewise.
17246 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
17247 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
17248 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17249 (vectorizable_assignment): Likewise.
17250 (vectorizable_shift): Likewise.
17251 (vectorizable_operation): Likewise.
17252 * tree-vrp.c (register_edge_assert_for_2): Likewise.
17254 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
17256 * match.pd: Add pow (C, x) simplification.
17258 2017-08-21 Richard Biener <rguenther@suse.de>
17260 PR tree-optimization/81900
17261 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
17262 for blocks with abnormal predecessors.
17263 (compute_antic): Do not set visited flag prematurely.
17265 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
17268 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
17270 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17272 * stor-layout.h (vector_type_mode): Move to...
17273 * tree.h (vector_type_mode): ...here.
17274 * stor-layout.c (vector_type_mode): Move to...
17275 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
17277 2017-08-21 Richard Biener <rguenther@suse.de>
17279 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
17280 register_external_die hooks.
17281 (debug_false_tree_charstarstar_uhwistar): Declare.
17282 (debug_nothing_tree_charstar_uhwi): Likewise.
17283 * debug.c (do_nothing_debug_hooks): Adjust.
17284 (debug_false_tree_charstarstar_uhwistar): New do nothing.
17285 (debug_nothing_tree_charstar_uhwi): Likewise.
17286 * dbxout.c (dbx_debug_hooks): Adjust.
17287 (xcoff_debug_hooks): Likewise.
17288 * sdbout.c (sdb_debug_hooks): Likewise.
17289 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
17290 * dwarf2out.c (macinfo_label_base): New global.
17291 (dwarf2out_register_external_die): New function for the
17292 register_external_die hook.
17293 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
17294 (dwarf2_debug_hooks): Use them.
17295 (dwarf2_lineno_debug_hooks): Adjust.
17296 (struct die_struct): Add with_offset flag.
17297 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
17298 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
17299 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
17300 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
17301 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
17302 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
17303 defining section names for the early LTO debug variants.
17304 (reset_indirect_string): New helper.
17305 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
17306 (print_dw_val): Add support for offsetted symbol references.
17307 (get_ultimate_context): Split out from is_cxx.
17308 (is_cxx): Use get_ultimate_context.
17309 (is_fortran): Add decl overload.
17310 (compute_comp_unit_symbol): Split out worker from
17311 compute_section_prefix.
17312 (compute_section_prefix): Call compute_comp_unit_symbol and
17313 set comdat_type_p here.
17314 (output_die): Skip DIE symbol output for the LTO added one.
17315 Handle DIE symbol references with offset.
17316 (output_comp_unit): Guard section name mangling properly.
17317 For LTO debug sections emit a symbol at the section beginning
17318 which we use to refer to its DIEs.
17319 (add_abstract_origin_attribute): For DIEs registered via
17320 dwarf2out_register_external_die directly refer to the early
17321 DIE rather than indirectly through the shadow one we created.
17322 Remove obsolete call to dwarf2out_abstract_function for
17323 non-function/block origins.
17324 (gen_array_type_die): When generating early LTO debug do
17325 not emit DW_AT_string_length.
17326 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
17327 late when in LTO. As suggested place a gcc_unreachable for
17328 the DECL_ABSTRACT_P case.
17329 (gen_subprogram_die): Avoid another specification DIE
17330 for early built declarations/definitions for the late LTO case.
17331 (gen_variable_die): Add type references for late duplicated VLA dies
17333 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
17334 we have the abstract instance already.
17335 (process_scope_var): Adjust decl DIE contexts in LTO which
17336 first puts them in limbo.
17337 (gen_decl_die): Do not generate type DIEs late apart from
17338 types for VLAs or for decls we do not yet have a DIE. Do not
17339 call dwarf2out_abstract_function late.
17340 (dwarf2out_early_global_decl): Make sure to create DIEs
17341 for abstract instances of a decl first.
17342 (dwarf2out_late_global_decl): Adjust comment.
17343 (output_macinfo_op): With multiple macro sections use
17344 macinfo_label_base to distinguish labels.
17345 (output_macinfo): Likewise. Update macinfo_label_base.
17346 Pass in the line info label.
17347 (note_variable_value_in_expr): When generating LTO resolve
17348 all variable values here by generating DIEs as needed.
17349 (init_sections_and_labels): Add early LTO debug flag parameter
17350 and generate different sections and names if set. Add generation
17351 counter for the labels so we can have multiple of them.
17352 (reset_dies): Helper to allow DIEs to be output multiple times.
17353 (dwarf2out_finish): When outputting DIEs to the fat part of an
17354 LTO object first reset DIEs.
17355 (dwarf2out_early_finish): Output early DIEs when generating LTO.
17356 (modified_type_die): Check for decl_ultimate_origin being self
17358 (gen_type_die_with_usage): Likewise.
17359 (gen_typedef_die): Allow decl_ultimate_origin being self.
17360 (set_decl_abstract_flags): Remove.
17361 (set_block_abstract_flags): Likewise.
17362 (dwarf2out_abstract_function): Treat the early generated DIEs
17363 as the abstract copy and only add DW_AT_inline and
17364 DW_AT_artificial here and call set_decl_origin_self.
17365 If the DIE has an abstract origin don't do anything.
17366 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
17367 if we have none yet (Go fails to build one, PR78628).
17368 (variably_modified_type_p): Prevent endless recursion for Ada
17369 cyclic pointer types.
17370 * lto-streamer-in.c: Include debug.h.
17371 (dref_queue): New global.
17372 (lto_read_tree_1): Stream in DIE references.
17373 (lto_input_tree): Register DIE references.
17374 (input_function): Stream DECL_DEBUG_ARGS.
17375 * lto-streamer-out.c: Include debug.h.
17376 (lto_write_tree_1): Output DIE references.
17377 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
17378 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
17379 (output_function): Stream DECL_DEBUG_ARGS.
17380 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
17381 Stream DECL_ABSTRACT_ORIGIN.
17382 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
17383 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
17384 DECL_CONTEXT for file-scope decls.
17385 * lto-streamer.h (struct dref_entry): Declare.
17386 (dref_queue): Likewise.
17387 * cfgexpand.c (pass_expand::execute): Do not call the
17388 outlining_inline_function hook here.
17389 * lto-wrapper.c (debug_obj): New global.
17390 (tool_cleanup): Unlink it if required.
17391 (debug_objcopy): New function.
17392 (run_gcc): Handle early debug sections in the IL files by
17393 extracting them to separate files, partially linkin them and
17394 feeding the result back as result to the linker.
17395 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
17396 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
17397 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
17398 sections into a separate segment.
17399 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
17401 (darwin_asm_dwarf_section): Likewise.
17402 (darwin_asm_output_dwarf_offset): Likewise.
17403 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
17405 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17406 Alan Hayward <alan.hayward@arm.com>
17407 David Sherwood <david.sherwood@arm.com>
17409 * read-md.h (md_reader::record_potential_iterator_use): Replace
17410 pointer argument with an rtx and an index.
17411 * read-rtl.c (iterator_group::apply_iterator): Likewise.
17412 (apply_mode_iterator): Likewise.
17413 (apply_code_iterator): Likewise.
17414 (apply_int_iterator): Likewise.
17415 (apply_subst_iterator): Likewise.
17416 (record_iterator_use): Likewise.
17417 (record_attribute_use): Likewise.
17418 (md_reader::record_potential_iterator_use): Likewise. Update calls
17419 to record_iterator_use and apply_iterator.
17420 (iterator_use): Replace ptr with x and index.
17421 (attribute_use): Likewise.
17422 (apply_attribute_uses): Update calls to apply_iterator.
17423 (apply_iterators): Likewise. Update initialization of iterator_use.
17424 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
17425 and record_potential_iterator_use.
17426 (rtx_reader::read_rtx_operand): Likewise.
17428 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17429 Alan Hayward <alan.hayward@arm.com>
17430 David Sherwood <david.sherwood@arm.com>
17432 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
17435 2017-08-21 Richard Biener <rguenther@suse.de>
17437 PR middle-end/81884
17438 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
17439 at struct end conservatively when comparing common bases.
17441 2017-08-21 Richard Biener <rguenther@suse.de>
17443 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
17444 (mem_ref_in_stmt): Remove.
17445 (determine_max_movement): Use ref index to get at the reference.
17446 (invariantness_dom_walker::before_dom_children): Deal with
17447 lim data already initialized.
17448 (gather_mem_refs_stmt): Initialize lim data and record ref index.
17450 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
17452 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
17453 (TARGET_ISA_ROUND): Ditto.
17454 (TARGET_ROUND): Ditto.
17455 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
17456 * config/i386/i386.md: Ditto.
17457 * config/i386/sse.md: Ditto.
17458 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
17459 with OPTION_MASK_ISA_SSE4_1.
17461 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
17464 * doc/extend.texi (x86 Built-in Functions): Correct the name of
17465 __builtin_ia32_lzcnt_u16.
17467 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
17470 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
17471 (rs6000_set_current_function): Rewrite function to use it.
17473 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
17476 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
17477 and TYPE_WARN_IF_NOT_ALIGN.
17478 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
17479 (handle_warn_if_not_align): New.
17480 (place_union_field): Call handle_warn_if_not_align.
17481 (place_field): Call handle_warn_if_not_align.
17482 Copy TYPE_WARN_IF_NOT_ALIGN.
17483 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
17484 (layout_type): Likewise.
17485 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
17487 (tree_decl_common): Add warn_if_not_align.
17488 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
17489 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
17490 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
17491 (DECL_WARN_IF_NOT_ALIGN): Likewise.
17492 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
17493 * doc/extend.texi: Document warn_if_not_aligned attribute.
17494 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
17496 2017-08-17 Martin Liska <mliska@suse.cz>
17499 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
17500 (get_data_dependence): Use it as pointer type.
17501 (distribute_loop): Likewise.
17503 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17505 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
17506 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
17507 (p8_vmrgow_<mode>_direct): New define_insn.
17508 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
17509 handle endianness for vmrgew and vmrgow permute patterns.
17511 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
17513 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
17514 * config/rs6000/rs6000-cpus.def: Remove comment.
17515 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
17516 (POWERPC_MASKS): Likewise.
17517 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
17518 use of TARGET_VSX_TIMODE.
17519 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
17520 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
17521 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
17522 (rs6000_option_override_internal): Remove dead code.
17523 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
17524 (rs6000_legitimize_reload_address): Likewise.
17525 (rs6000_legitimate_address_p): Likewise.
17526 (rs6000_opt_masks): Delete "vsx-timode".
17527 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
17528 from function comment.
17529 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
17530 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
17531 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
17533 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
17534 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
17535 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
17536 (VSX_TI): Likewise.
17538 (define_peephole2): Likewise.
17540 2017-08-17 Martin Sebor <msebor@redhat.com>
17543 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
17544 past the end of an array.
17545 (test_pp_format): Add test cases.
17547 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
17549 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
17550 missing ECF_NOTHROW flags.
17552 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
17555 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
17556 operands residing in integer registers.
17557 (*vsx_le_perm_load_<mode>): Likewise.
17558 (*vsx_le_perm_store_<mode>): Likewise.
17559 (define_peephole2): Add peepholes to optimize the above.
17561 2017-08-17 Marek Polacek <polacek@redhat.com>
17563 PR middle-end/81814
17564 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
17565 to mimic what shorten_compare did. Change the return type to bool.
17566 (fold_cond_expr_with_comparison): Update call to
17567 operand_equal_for_comparison_p.
17568 (fold_ternary_loc): Likewise.
17570 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
17572 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
17574 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
17576 2017-08-17 Richard Biener <rguenther@suse.de>
17578 * tree-ssa-structalias.c (solve_graph): When propagating
17579 to successors update the graphs succ edges and avoid duplicate work.
17581 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
17584 * config/i386/i386.c (ix86_option_override_internal): Save target
17585 specific options after ix86_stack_protector_guard_reg was changed.
17587 2017-08-17 Richard Biener <rguenther@suse.de>
17589 PR tree-optimization/81827
17590 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
17591 (new_var_info): Initialize it conservatively.
17592 (get_call_vi): Mark register vars.
17593 (new_scalar_tmp_constraint_exp): Likewise.
17594 (handle_rhs_call): Likewise.
17595 (handle_const_call): Likewise.
17596 (create_function_info_for): Likewise.
17597 (solve_constraints): Sort varinfos to separate register from
17598 non-register vars to pack points-to solution bitmaps during
17601 2017-08-17 Marek Polacek <polacek@redhat.com>
17603 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
17605 2017-08-17 Richard Biener <rguenther@suse.de>
17607 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
17608 to true when overflow is undefined and we saturated the result.
17610 2017-08-17 Alan Modra <amodra@gmail.com>
17613 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
17614 Don't use store multiple if only one reg needs saving.
17615 (interesting_frame_related_regno): New function.
17616 (rs6000_frame_related): Don't emit frame info for regs that
17618 (rs6000_emit_epilogue): Likewise.
17620 2017-08-16 Nathan Sidwell <nathan@acm.org>
17622 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
17623 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
17624 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
17625 * tree.c (free_lang_data_in_type): Use else-if chain. Always
17626 clear TYPE_LANG_1. Remove obsolete member-function stripping.
17627 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
17628 (verify_type): Adjust for TYPE_BINFO move.
17629 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
17630 process TYPE_BINFO directly.
17631 (hash_tree): Likewise.
17632 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17634 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17637 2017-08-16 David Malcolm <dmalcolm@redhat.com>
17639 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
17641 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
17644 * config/i386/i386.md (*anddi_1_btr): Change predicates of
17645 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
17646 Add ix86_binary_operator_ok to insn constraint.
17647 (*iordi_1_bts): Ditto.
17648 (*xordi_1_btc): Ditto.
17649 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
17650 Update corresponding peephole2 pattern.
17654 2017-08-16 Bin Cheng <bin.cheng@arm.com>
17656 PR tree-optimization/81832
17657 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
17658 copy loop header which has IFN_LOOP_DIST_ALIAS call.
17660 2017-08-16 Marek Polacek <polacek@redhat.com>
17663 * fold-const.c (fold_indirect_ref_1): Restore original behavior
17664 regarding size_zero_node.
17666 2017-08-16 Martin Liska <mliska@suse.cz>
17669 * config.gcc: Respect previously set extra_objs in case
17672 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
17674 PR tree-optimization/81835
17675 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
17676 the phi SSA_NAME. Check that the condition in a COND_EXPR does
17677 not depend on the phi.
17679 2017-08-16 Alan Modra <amodra@gmail.com>
17681 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
17684 2017-08-16 Alan Modra <amodra@gmail.com>
17686 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
17687 (save_reg_p): ..into this. Update all callers.
17688 (first_reg_to_save): Simplify.
17690 2017-08-16 Alan Modra <amodra@gmail.com>
17692 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
17695 2017-08-15 Joseph Myers <joseph@codesourcery.com>
17699 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
17700 constant count if that count is less than 32.
17702 2017-08-15 Nathan Sidwell <nathan@acm.org>
17704 * gcc.c (execute): Emit friendlier message if inferior is killed
17705 by an external cause.
17707 2017-08-15 Richard Biener <rguenther@suse.de>
17709 PR tree-optimization/81790
17710 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
17711 CONSTRUCTORs from simplifying and VN.
17713 2017-08-14 Martin Sebor <msebor@redhat.com>
17715 * builtin-attrs.def: Add comments.
17717 2017-08-14 Martin Sebor <msebor@redhat.com>
17720 * doc/extend.texi (attribute nonstring): Document new attribute.
17722 2017-08-14 Martin Sebor <msebor@redhat.com>
17725 * tree-diagnostic.c (default_tree_printer): Handle %G.
17726 * gimple-pretty-print.h (percent_G_format): Declare new function.
17727 * gimple-pretty-print.c (percent_G_format): Define.
17728 * tree-pretty-print.c (percent_K_format): Add argument.
17730 2017-08-14 Martin Sebor <msebor@redhat.com>
17732 PR translation/79998
17733 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
17734 Remove a stray space.
17736 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
17739 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
17740 (*iordi_1_bts): Ditto.
17741 (*xordi_1_btc): Ditto.
17743 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17746 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
17747 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17749 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
17750 quoted strings, and make more translator-friendly.
17751 (darwin_rs6000_override_options): Likewise.
17752 (rs6000_option_override_internal): Likewise.
17753 (rs6000_return_in_memory): Fix overlong line.
17754 (init_cmulative_args): Use quoted strings, and make more
17755 translator-friendly.
17756 (rs6000_pass_by_reference): Fix overlong line.
17757 (def_builtin): Use quoted strings.
17758 (altivec_expand_predicate_builtin): Use quoted strings, and make
17759 more translator-friendly.
17760 (htm_expand_builtin): Use quoted strings.
17761 (cpu_expand_builtin): Use quoted strings, and make more
17762 translator-friendly.
17763 (altivec_expand_builtin): Likewise.
17764 (paired_expand_predicate_builtin): Likewise.
17765 (rs6000_invalid_builtin): Likewise.
17766 (builtin_function_type): Use quoted strings.
17767 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
17768 more translator-friendly.
17769 (rs6000_trampoline_init): Likewise.
17770 (rs6000_handle_altivec_attribute): Likewise.
17771 (rs6000_inner_target_options): Use quoted strings.
17772 (rs6000_disable_incompatible_switches): Likewise.
17773 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
17774 strings, and make more translator-friendly.
17775 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
17777 2017-08-14 Bin Cheng <bin.cheng@arm.com>
17779 PR tree-optimization/81799
17780 * tree-loop-distribution.c (version_loop_by_alias_check): Force
17781 cond_expr to simple gimple operand.
17783 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
17785 PR middle-end/46932
17786 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
17788 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
17792 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
17793 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
17794 TARGET_GASISR_PROLOGUES.
17795 * config/avr/avr.c (avr_option_override): Same.
17796 (avr_pass_pre_proep::execute): Same.
17798 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
17801 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
17802 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
17804 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
17806 * config/i386/i386.md (*load_tp_<mode>): Redefine as
17807 define_insn_and_split. Split to a memory load from 0 in
17808 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
17809 using PTR mode iterator.
17810 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
17811 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
17812 (*add_tp_<mode>): Redefine as define_insn_and_split.
17813 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
17814 address space. Merge with *add_tp_x32 using PTR mode iterator.
17815 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
17816 Split to an add with a memory load from 0 in
17817 DEFAULT_TLS_SEG_REG address space.
17819 2017-08-12 Andrew Pinski <apinski@cavium.com>
17821 * config/aarch64/aarch64-option-extensions.def (rdma):
17822 Fix feature string to what Linux prints out in /proc/cpuinfo.
17824 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
17827 * dwarf2out.c (modified_type_die): For C typedef types that have
17828 an ultimate origin, process the ultimate origin instead of the
17830 (gen_typedef_die): Assert that input DECLs have no ultimate
17832 (gen_type_die_with_usage): For typedef variants that have an
17833 ultimate origin, just call gen_decl_die on the original DECL.
17834 (process_scope_var): Avoid creating DIEs for local typedefs and
17835 concrete static variables.
17837 2017-08-12 Alan Modra <amodra@gmail.com>
17841 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
17842 match gnu-user.h startfile.
17843 (ENDFILE_LINUX_SPEC): Similarly.
17845 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
17848 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
17850 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
17852 2017-08-11 Tamar Christina <tamar.christina@arm.com>
17853 * config/aarch64/aarch64.md (mov<mode>): Change.
17854 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
17855 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
17856 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
17858 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
17860 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
17861 for storage order barriers.
17863 2017-08-11 Martin Liska <mliska@suse.cz>
17866 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
17867 variables of void type.
17869 2017-08-11 Martin Liska <mliska@suse.cz>
17871 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
17872 TARGET_SUPPORTS_ALIASES.
17873 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
17874 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
17875 (optimize_weakref): Likewise.
17876 * symtab.c (symtab_node::noninterposable_alias): Likewise.
17877 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
17878 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
17880 2017-08-11 Martin Liska <mliska@suse.cz>
17883 * config/i386/i386.c (make_resolver_func): Do complete
17884 refactoring of the function.
17886 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17889 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
17890 * config/i386/i386.c (ix86_stack_protect_guard): Use
17891 ix86_stack_protect_guard_symbol_str to generate varible declaration.
17892 * doc/invoke.texi (x86 Options): Document
17893 -mstack-protector-guard-symbol= option.
17895 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17897 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
17898 * config/i386/i386.c (ix86_split_stack_guard): New function.
17899 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
17900 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
17901 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
17902 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
17903 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
17904 (split_stack_space_check): Call ix86_split_stack_guard.
17906 2017-08-10 Martin Sebor <msebor@redhat.com>
17908 * print-tree.c (print_node): Print location using the established
17910 Replace spaces with colons.
17911 (debug_raw, debug): Ditto.
17913 2017-08-10 Martin Sebor <msebor@redhat.com>
17916 * pretty-print.c (pp_format): Correct the handling of %s precision.
17918 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
17921 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
17923 (ix86_finalize_stack_frame_flags): This. Also clear
17924 frame_pointer_needed if -fno-omit-frame-pointer is used without
17926 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
17927 with ix86_finalize_stack_frame_flags.
17928 (ix86_expand_epilogue): Likewise.
17929 (ix86_expand_split_stack_prologue): Likewise.
17930 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
17932 2017-08-10 Martin Liska <mliska@suse.cz>
17935 * c-attribs.c (handle_target_attribute):
17936 Report warning for an empty string argument of target attribute.
17938 2017-08-09 Jakub Jelinek <jakub@redhat.com>
17941 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
17943 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
17944 or DECL_NONLOCAL labels.
17945 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
17946 or DECL_NONLOCAL labels here.
17948 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
17950 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
17951 to indicate when early gimple folding has been disabled.
17952 (rs6000_gimple_fold_builtin): Add debug content.
17953 (rs6000_invalid_builtin): Fix whitespace.
17954 (rs6000_expand_builtin): Fix whitespace.
17955 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
17957 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
17960 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
17961 SAVE_MULTIPLE if not all the registers that saves, should be saved.
17963 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
17965 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
17966 (qdf24xx): Likewise.
17967 * config/aarch64/aarch64.md: Include falkor.md.
17968 * config/aarch64/falkor.md: New.
17970 2017-08-09 Marek Polacek <polacek@redhat.com>
17973 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
17974 * diagnostic.c (emit_diagnostic): Add a comment.
17975 (emit_diagnostic_valist): New function.
17977 2017-08-09 Marek Polacek <polacek@redhat.com>
17980 * input.c (make_location): New overload.
17981 * input.h (make_location): Declare.
17983 2017-08-08 Alan Modra <amodra@gmail.com>
17984 H.J. Lu <hongjiu.lu@intel.com>
17987 * gcc.c (NO_PIE_SPEC): Delete.
17988 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
17990 (LINK_PIE_SPEC): ..to here.
17991 (LINK_COMMAND_SPEC): Support -no-pie.
17992 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
17993 chain of crtbegin*.o selection, update for PIE_SPEC changes and
17995 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
17996 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
17997 (ENDFILE_CRTEND_SPEC): Similarly.
17999 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
18002 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
18003 (mstack-protector-guard-offset=): Ditto.
18004 * config/i386/i386.c (ix86_option_override): Handle
18005 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
18007 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
18008 ix86_stack_protect_guard_offset variables.
18009 (TARGET_STACK_PROTECT_GUARD): Always define.
18010 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
18011 and -mstack-protector-guard-offset= options.
18013 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18015 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
18016 boundary case for the last candidate.
18018 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18020 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
18021 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
18023 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18025 PR middle-end/19706
18026 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
18027 * config/aarch64/aarch64-builtins.c
18028 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
18029 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
18030 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
18032 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18033 Andrew Pinski <pinskia@gmail.com>
18035 PR middle-end/19706
18036 * internal-fn.def (XORSIGN): New.
18037 * optabs.def (xorsign_optab): New.
18038 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
18039 (convert_expand_mult_copysign): New.
18040 (pass_optimize_widening_mul::execute): Call
18041 convert_expand_mult_copysign.
18043 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18045 PR tree-optimization/81354
18046 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
18047 Insert on edges rather than explicitly creating landing pads.
18048 (analyze_candidates_and_replace): Commit edge inserts.
18050 2017-08-08 Richard Biener <rguenther@suse.de>
18052 PR middle-end/81719
18053 * tree-ssa-loop-niter.c: Include tree-dfa.h.
18054 (expand_simple_operations): Also look through ADDR_EXPRs with
18055 MEM_REF bases treating them as POINTER_PLUS_EXPR.
18057 2017-08-08 Richard Biener <rguenther@suse.de>
18059 PR tree-optimization/81723
18060 * tree-vect-slp.c (struct bst_traits): New hash traits.
18061 (bst_fail): New global.
18062 (vect_build_slp_tree_2): New worker, split out from ...
18063 (vect_build_slp_tree): ... this now wrapping it with using
18064 bst_fail set to cache SLP tree build fails. Properly handle
18066 (vect_analyze_slp_instance): Allocate and free bst_fail.
18068 2017-08-08 Martin Liska <mliska@suse.cz>
18071 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
18072 LABEL_DECLs that can be from a different function.
18074 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18076 PR tree-optimization/81744
18077 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
18078 loop's number of iterations.
18080 2017-08-08 Martin Liska <mliska@suse.cz>
18082 * asan.c: Include header files.
18083 * attribs.c (build_decl_attribute_variant): New function moved
18085 (build_type_attribute_qual_variant): Likewise.
18086 (cmp_attrib_identifiers): Likewise.
18087 (simple_cst_list_equal): Likewise.
18088 (omp_declare_simd_clauses_equal): Likewise.
18089 (attribute_value_equal): Likewise.
18090 (comp_type_attributes): Likewise.
18091 (build_type_attribute_variant): Likewise.
18092 (lookup_ident_attribute): Likewise.
18093 (remove_attribute): Likewise.
18094 (merge_attributes): Likewise.
18095 (merge_type_attributes): Likewise.
18096 (merge_decl_attributes): Likewise.
18097 (merge_dllimport_decl_attributes): Likewise.
18098 (handle_dll_attribute): Likewise.
18099 (attribute_list_equal): Likewise.
18100 (attribute_list_contained): Likewise.
18101 * attribs.h (lookup_attribute): New function moved from tree.[ch].
18102 (lookup_attribute_by_prefix): Likewise.
18103 * bb-reorder.c: Include header files.
18104 * builtins.c: Likewise.
18105 * calls.c: Likewise.
18106 * cfgexpand.c: Likewise.
18107 * cgraph.c: Likewise.
18108 * cgraphunit.c: Likewise.
18109 * convert.c: Likewise.
18110 * dwarf2out.c: Likewise.
18111 * final.c: Likewise.
18112 * fold-const.c: Likewise.
18113 * function.c: Likewise.
18114 * gimple-expr.c: Likewise.
18115 * gimple-fold.c: Likewise.
18116 * gimple-pretty-print.c: Likewise.
18117 * gimple.c: Likewise.
18118 * gimplify.c: Likewise.
18119 * hsa-common.c: Likewise.
18120 * hsa-gen.c: Likewise.
18121 * internal-fn.c: Likewise.
18122 * ipa-chkp.c: Likewise.
18123 * ipa-cp.c: Likewise.
18124 * ipa-devirt.c: Likewise.
18125 * ipa-fnsummary.c: Likewise.
18126 * ipa-inline.c: Likewise.
18127 * ipa-visibility.c: Likewise.
18129 * lto-cgraph.c: Likewise.
18130 * omp-expand.c: Likewise.
18131 * omp-general.c: Likewise.
18132 * omp-low.c: Likewise.
18133 * omp-offload.c: Likewise.
18134 * omp-simd-clone.c: Likewise.
18135 * opts-global.c: Likewise.
18136 * passes.c: Likewise.
18137 * predict.c: Likewise.
18138 * sancov.c: Likewise.
18139 * sanopt.c: Likewise.
18140 * symtab.c: Likewise.
18141 * toplev.c: Likewise.
18142 * trans-mem.c: Likewise.
18143 * tree-chkp.c: Likewise.
18144 * tree-eh.c: Likewise.
18145 * tree-into-ssa.c: Likewise.
18146 * tree-object-size.c: Likewise.
18147 * tree-parloops.c: Likewise.
18148 * tree-profile.c: Likewise.
18149 * tree-ssa-ccp.c: Likewise.
18150 * tree-ssa-live.c: Likewise.
18151 * tree-ssa-loop.c: Likewise.
18152 * tree-ssa-sccvn.c: Likewise.
18153 * tree-ssa-structalias.c: Likewise.
18154 * tree-ssa.c: Likewise.
18155 * tree-streamer-in.c: Likewise.
18156 * tree-vectorizer.c: Likewise.
18157 * tree-vrp.c: Likewise.
18158 * tsan.c: Likewise.
18159 * ubsan.c: Likewise.
18160 * varasm.c: Likewise.
18161 * varpool.c: Likewise.
18162 * tree.c: Remove functions moved to attribs.[ch].
18163 * tree.h: Likewise.
18164 * config/aarch64/aarch64.c: Add attrs.h header file.
18165 * config/alpha/alpha.c: Likewise.
18166 * config/arc/arc.c: Likewise.
18167 * config/arm/arm.c: Likewise.
18168 * config/avr/avr.c: Likewise.
18169 * config/bfin/bfin.c: Likewise.
18170 * config/c6x/c6x.c: Likewise.
18171 * config/cr16/cr16.c: Likewise.
18172 * config/cris/cris.c: Likewise.
18173 * config/darwin.c: Likewise.
18174 * config/epiphany/epiphany.c: Likewise.
18175 * config/fr30/fr30.c: Likewise.
18176 * config/frv/frv.c: Likewise.
18177 * config/ft32/ft32.c: Likewise.
18178 * config/h8300/h8300.c: Likewise.
18179 * config/i386/winnt.c: Likewise.
18180 * config/ia64/ia64.c: Likewise.
18181 * config/iq2000/iq2000.c: Likewise.
18182 * config/lm32/lm32.c: Likewise.
18183 * config/m32c/m32c.c: Likewise.
18184 * config/m32r/m32r.c: Likewise.
18185 * config/m68k/m68k.c: Likewise.
18186 * config/mcore/mcore.c: Likewise.
18187 * config/microblaze/microblaze.c: Likewise.
18188 * config/mips/mips.c: Likewise.
18189 * config/mmix/mmix.c: Likewise.
18190 * config/mn10300/mn10300.c: Likewise.
18191 * config/moxie/moxie.c: Likewise.
18192 * config/msp430/msp430.c: Likewise.
18193 * config/nds32/nds32-isr.c: Likewise.
18194 * config/nds32/nds32.c: Likewise.
18195 * config/nios2/nios2.c: Likewise.
18196 * config/nvptx/nvptx.c: Likewise.
18197 * config/pa/pa.c: Likewise.
18198 * config/pdp11/pdp11.c: Likewise.
18199 * config/powerpcspe/powerpcspe.c: Likewise.
18200 * config/riscv/riscv.c: Likewise.
18201 * config/rl78/rl78.c: Likewise.
18202 * config/rx/rx.c: Likewise.
18203 * config/s390/s390.c: Likewise.
18204 * config/sh/sh.c: Likewise.
18205 * config/sol2.c: Likewise.
18206 * config/sparc/sparc.c: Likewise.
18207 * config/spu/spu.c: Likewise.
18208 * config/stormy16/stormy16.c: Likewise.
18209 * config/tilegx/tilegx.c: Likewise.
18210 * config/tilepro/tilepro.c: Likewise.
18211 * config/v850/v850.c: Likewise.
18212 * config/vax/vax.c: Likewise.
18213 * config/visium/visium.c: Likewise.
18214 * config/xtensa/xtensa.c: Likewise.
18216 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18219 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
18220 constraints since the -mupper-regs-* switches have been
18222 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
18223 into a vector from a double word element that was extracted from
18224 another vector, and eliminate extra XXPERMDI instructions.
18225 (vsx_concat_<mode>_2): Likewise.
18226 (vsx_concat_<mode>_3): Likewise.
18227 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
18228 concat to allow optimizing inserts from previous extracts.
18230 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
18232 * config/i386/i386.c (ix86_stack_protect_guard): Generate
18233 memory reference to a SSP offset in TLS address space.
18234 (ix86_print_operand) <case '@'>: Remove.
18235 (ix86_print_operand_punct_valid_p): Remove '@' code.
18236 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
18237 UNSPEC_SP_TLS_TEST.
18238 (stack_tls_protect_set_<mode>): Remove.
18239 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
18240 (stack_tls_protect_test_<mode>): Remove.
18241 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
18243 2017-08-07 Olivier Hainque <hainque@adacore.com>
18246 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
18248 2017-08-07 Douglas Rupp <rupp@adacore.com>
18250 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
18251 variable was referenced as multidir in command.
18253 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18256 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
18259 2017-08-07 Martin Liska <mliska@suse.cz>
18261 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
18262 * config/rl78/rl78.c: Add include of attribs.h.
18263 * config/sh/sh.c: Likewise.
18264 * config/v850/v850.c: Likewise.
18266 2017-08-07 Tom de Vries <tom@codesourcery.com>
18268 PR middle-end/78266
18269 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
18271 2017-08-07 Martin Liska <mliska@suse.cz>
18273 * config/mips/mips.c: Include attribs.h.
18275 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
18278 * doc/invoke.texi: Document change in behvaior for -Ofast for
18281 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
18283 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
18285 (aarch64_pop_regs): Likewise.
18286 (aarch64_gen_load_pair): Likewise.
18287 (aarch64_save_callee_saves): Likewise.
18288 (aarch64_restore_callee_saves): Likewise.
18290 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18292 * config/i386/i386.c: Revert the last change.
18294 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18297 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18299 (ix86_finalize_stack_frame_flags): This. Also clear
18300 frame_pointer_needed if -fno-omit-frame-pointer is used without
18302 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18303 with ix86_finalize_stack_frame_flags.
18304 (ix86_expand_epilogue): Likewise.
18305 (ix86_expand_split_stack_prologue): Likewise.
18307 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18310 * config/i386/i386.c (get_builtin_code_for_version): Set priority
18311 to P_AES for Westmere.
18313 2017-08-07 Jonathan Yong <10walls@gmail.com>
18315 * config/i386/mingw.opt (fset-stack-executable): Removed.
18316 * config/i386/cygming.opt (fset-stack-executable): Moved
18318 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
18320 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
18322 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
18324 2017-08-07 Marek Polacek <polacek@redhat.com>
18326 PR middle-end/81737
18327 * fold-const.c (fold_indirect_ref_1): Check type_domain.
18329 2017-08-07 Martin Liska <mliska@suse.cz>
18331 * attribs.h (canonicalize_attr_name): New function.
18332 (cmp_attribs): Move from c-format.c and adjusted.
18333 (is_attribute_p): Moved from tree.h.
18334 * tree-inline.c: Add new includes.
18335 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
18336 (private_is_attribute_p): Remove.
18337 (private_lookup_attribute): Likewise.
18338 (private_lookup_attribute_by_prefix): Simplify.
18339 (remove_attribute): Use is_attribute_p.
18340 * tree.h: Remove removed declarations.
18342 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18344 PR middle-end/81698
18345 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
18346 instead of computing it in the function. Formatting fix.
18347 (expand_case): Don't rely on default_edge being the first edge,
18348 clear it if removing it, pass default_edge to
18349 emit_case_dispatch_table.
18350 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
18353 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
18355 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
18356 insn in the function, emit NOP after the insn.
18358 2017-08-06 Tom de Vries <tom@codesourcery.com>
18360 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
18363 2017-08-06 Tom de Vries <tom@codesourcery.com>
18365 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
18368 2017-08-04 Yury Gribov <tetra2005@gmail.com>
18370 PR tree-optimization/57371
18371 * match.pd: New pattern.
18373 2017-08-04 Marek Polacek <polacek@redhat.com>
18375 PR middle-end/81695
18376 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
18377 perform the computation in offset_int.
18379 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18381 PR tree-optimization/81136
18382 * tree-vectorizer.h: Include tree-hash-traits.h.
18383 (vec_base_alignments): New typedef.
18384 (vec_info): Add a base_alignments field.
18385 (vect_record_base_alignments): Declare.
18386 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
18388 (DR_IS_CONDITIONAL_IN_STMT): New macro.
18389 (create_data_ref): Add an is_conditional_in_stmt argument.
18390 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
18391 the is_conditional_in_stmt field.
18392 (data_ref_loc): Add an is_conditional_in_stmt field.
18393 (get_references_in_stmt): Set the is_conditional_in_stmt field.
18394 (find_data_references_in_stmt): Update call to create_data_ref.
18395 (graphite_find_data_references_in_stmt): Likewise.
18396 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
18397 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
18398 (vect_record_base_alignment): New function.
18399 (vect_record_base_alignments): Likewise.
18400 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
18401 for nested statements even if we fail to compute a misalignment.
18402 Use pooled base alignments for unconditional references.
18403 (vect_find_same_alignment_drs): Compare base addresses instead
18405 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
18406 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
18408 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18410 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
18411 Add an explicit name for the enum. Use auto_vec for slp_instances
18412 and grouped_stores.
18413 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
18415 (_bb_vec_info): Add a constructor and destructor.
18416 (vinfo_for_stmt): Return NULL for uids of -1 as well.
18417 (destroy_loop_vec_info): Delete.
18418 (vect_destroy_datarefs): Likewise.
18419 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
18420 (vec_info::vec_info): New function.
18421 (vec_info::~vec_info): Likewise.
18422 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
18423 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
18424 destroy_loop_vec_info.
18425 * tree-vect-loop.c (new_loop_vec_info): Replace with...
18426 (_loop_vec_info::_loop_vec_info): ...this.
18427 (destroy_loop_vec_info): Replace with...
18428 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
18429 the stmt_vec_infos. Leave handling of vec_info information to its
18430 destructor. Remove explicit vector releases.
18431 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
18432 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
18433 * tree-vect-slp.c (new_bb_vec_info): Replace with...
18434 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
18435 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
18436 (destroy_bb_vec_info): Replace with...
18437 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
18438 information to its destructor.
18439 (vect_slp_analyze_bb_1): Use new and delete instead of
18440 new_bb_vec_info and destroy_bb_vec_info.
18441 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
18444 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18446 * tree-data-ref.h (subscript): Add access_fn field.
18447 (data_dependence_relation): Add could_be_independent_p.
18448 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
18449 (same_access_functions): Move to tree-data-ref.c.
18450 * tree-data-ref.c (ref_contains_union_access_p): New function.
18451 (access_fn_component_p): Likewise.
18452 (access_fn_components_comparable_p): Likewise.
18453 (dr_analyze_indices): Add a reference to access_fn_component_p.
18454 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
18456 (constant_access_functions): Likewise.
18457 (add_other_self_distances): Likewise.
18458 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
18459 (initialize_data_dependence_relation): Use XCNEW and remove
18460 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
18461 of access functions that have the same type. Allow the
18462 subsequence to end with different bases in some circumstances.
18463 Record the chosen access functions in SUB_ACCESS_FN.
18464 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
18465 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
18466 (subscript_dependence_tester_1): Likewise dra and drb.
18467 (build_classic_dist_vector): Update calls accordingly.
18468 (subscript_dependence_tester): Likewise.
18469 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
18470 DDR_COULD_BE_INDEPENDENT_P.
18471 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
18472 comp_alias_ddrs instead of may_alias_ddrs.
18473 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
18475 (vect_analyze_data_ref_dependence): Use it if
18476 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
18477 distance vectors if that fails.
18478 (dependence_distance_ge_vf): New function.
18479 (vect_prune_runtime_alias_test_list): Use it. Don't clear
18480 LOOP_VINFO_MAY_ALIAS_DDRS.
18482 2017-08-04 Richard Biener <rguenther@suse.de>
18484 PR middle-end/81705
18485 * fold-const.c (fold_binary_loc): Properly restrict
18486 minus_var0 && minus_var1 case when associating undefined overflow
18489 2017-08-04 Tom de Vries <tom@codesourcery.com>
18491 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
18493 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18495 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18496 Don't start diagnostic messages with a capital letter.
18497 * config/rs6000/rs6000.c (rs6000_option_override_internal):
18499 (rs6000_invalid_builtin): Likewise.
18500 (rs6000_trampoline_init): Likewise.
18502 2017-08-03 Jakub Jelinek <jakub@redhat.com>
18505 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
18506 after setting changeable df flags.
18508 2017-08-03 Richard Biener <rguenther@suse.de>
18510 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
18511 up if the use is in USE - X.
18513 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
18515 * toplev.c (dumpfile.h): New include.
18516 (internal_error_reentered): New static function. Use it...
18517 (internal_error_function): ...here to handle reentered internal_error.
18519 2017-08-03 Richard Biener <rguenther@suse.de>
18521 PR middle-end/81148
18522 * fold-const.c (split_tree): Add minus_var and minus_con
18523 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
18524 here but always use minus_*.
18525 (associate_trees): Assert we never associate with MINUS_EXPR
18526 and NULL first operand. Do not recurse for PLUS_EXPR operands
18527 when associating as MINUS_EXPR either.
18528 (fold_binary_loc): Track minus_var and minus_con.
18530 2017-08-03 Tom de Vries <tom@codesourcery.com>
18533 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
18534 ACCEL_COMPILER, apply finish_options on
18535 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
18537 2017-08-03 Tom de Vries <tom@codesourcery.com>
18540 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
18541 function_entry_patch_area_size > 0.
18543 2017-08-03 Jakub Jelinek <jakub@redhat.com>
18546 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
18547 instead of 10??LU, perform unit multiplication in wide_int,
18548 don't change alloc_object_size_limit if the limit is larger
18551 PR tree-optimization/81655
18552 PR tree-optimization/81588
18553 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
18554 the case when ranges[i].low and high are 1 for unsigned type with
18557 PR middle-end/81052
18558 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
18559 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
18561 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18563 * tree-vrp.h: Add include guard.
18565 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
18568 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
18569 (ud2): New insn pattern.
18570 * config/i386/i386.c (ix86_expand_epilogue):
18571 For naked functions, generate ud2 instead of trap insn.
18573 2017-08-02 Marek Polacek <polacek@redhat.com>
18576 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
18578 2017-08-02 Tom de Vries <tom@codesourcery.com>
18579 Cesar Philippidis <cesar@codesourcery.com>
18581 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
18582 Add missing edge probabilities.
18584 2017-08-02 Tamar Christina <tamar.christina@arm.com>
18586 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
18587 Correct endianness.
18589 2017-08-02 Jakub Jelinek <jakub@redhat.com>
18591 PR middle-end/79499
18592 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
18593 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
18594 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
18595 prologue_seq sequences - if any.
18597 2017-08-02 Richard Biener <rguenther@suse.de>
18599 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
18600 via vectors if supported, integer extracts via punning if supported
18601 or otherwise vector extracts.
18603 2017-08-02 Richard Biener <rguenther@suse.de>
18605 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
18607 (bitmap_insert_into_set): ... this.
18609 2017-08-02 Richard Biener <rguenther@suse.de>
18611 PR tree-optimization/81633
18613 2015-08-17 Alan Hayward <alan.hayward@arm.com>
18615 PR tree-optimization/71752
18616 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
18618 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
18620 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
18621 (machine_function::call_ms2sysv_pad_out): Remove field.
18622 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
18623 (ix86_compute_frame_layout): Likewise.
18625 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
18628 * config/i386/i386.c (ix86_set_func_type): Disallow naked
18629 attribute with interrupt attribute.
18631 2017-08-01 Andrew Pinski <apinski@cavium.com>
18633 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
18634 BIT_INSERT_EXPR's operand 1
18635 to see if the types precision matches.
18637 2017-08-01 Martin Liska <mliska@suse.cz>
18639 PR middle-end/70140
18640 * builtins.c (expand_builtin_memcpy_args): Remove.
18641 (expand_builtin_memcpy): Call newly added function
18642 expand_builtin_memory_copy_args.
18643 (expand_builtin_memcpy_with_bounds): Likewise.
18644 (expand_builtin_mempcpy): Remove last argument.
18645 (expand_builtin_mempcpy_with_bounds): Likewise.
18646 (expand_builtin_memory_copy_args): New function created from
18647 expand_builtin_mempcpy_args with small modifications.
18648 (expand_builtin_mempcpy_args): Remove.
18649 (expand_builtin_stpcpy): Remove unused argument.
18650 (expand_builtin): Likewise.
18651 (expand_builtin_with_bounds): Likewise.
18653 2017-08-01 Martin Liska <mliska@suse.cz>
18656 Make mempcpy more optimal (PR middle-end/70140).
18658 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18661 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
18662 __builtin_vec_cmpne verify both arguments are compatible vectors
18663 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
18664 verify arg1_type is a pointer or array type. For __builtin_vec_st,
18665 move computation of aligned to after checking the argument types.
18669 * config/rs6000/vsx.md (vextract_fp_from_shorth,
18670 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
18673 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
18674 Doug Rupp <rupp@adacore.com>
18675 Olivier Hainque <hainque@adacore.com>
18677 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
18678 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
18680 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
18681 for TARGET_OS_CPP_BUILTIN.
18682 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
18683 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
18685 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
18686 passing required abi options to the assembler for EABI configurations.
18687 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
18688 of .text.hot and .text.unlikely sections for kernel modules when
18689 using ARM style exceptions.
18690 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
18691 options. Add EXTRA_CC1_SPEC.
18692 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
18694 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
18696 (ARM_TARGET2_DWARF_FORMAT): Define.
18697 * config/arm/t-vxworks: Adjust multilib control to removal of the
18698 Diab command line options.
18700 2017-08-01 Martin Liska <mliska@suse.cz>
18702 PR gcov-profile/81561
18703 * gcov.c (unblock): Make unblocking safe as we need to preserve
18704 index correspondence of blocks and block_lists.
18706 2017-08-01 Richard Biener <rguenther@suse.de>
18708 PR tree-optimization/81181
18709 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
18710 (compute_antic): ... end of iteration here.
18712 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
18714 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
18715 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
18716 (ftree-slp-vectorize): Likewise.
18717 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
18718 can no longer be set independent of flag_tree_loop_vectorize.
18719 * omp-general.c (emp_max_vf): Likewise.
18720 * opts.c (enable_fdo_optimizations): Remove references to
18721 flag_tree_vectorize, these are now implicit.
18722 (common_handle_option): Remove handling for OPT_ftree_vectorize,
18723 and leave it for the options machinery.
18725 2017-08-01 Martin Liska <mliska@suse.cz>
18727 PR middle-end/70140
18728 * builtins.c (expand_builtin_memcpy_args): Remove.
18729 (expand_builtin_memcpy): Call newly added function
18730 expand_builtin_memory_copy_args.
18731 (expand_builtin_memcpy_with_bounds): Likewise.
18732 (expand_builtin_mempcpy): Remove last argument.
18733 (expand_builtin_mempcpy_with_bounds): Likewise.
18734 (expand_builtin_memory_copy_args): New function created from
18735 expand_builtin_mempcpy_args with small modifications.
18736 (expand_builtin_mempcpy_args): Remove.
18737 (expand_builtin_stpcpy): Remove unused argument.
18738 (expand_builtin): Likewise.
18739 (expand_builtin_with_bounds): Likewise.
18741 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18744 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
18745 print "ds:" only for immediates in generic address space.
18747 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18750 * config/i386/i386.c (ix86_funciton_naked): New prototype.
18751 (ix86_function_ok_for_sibcall): Return false for naked functions.
18753 2017-08-01 Richard Biener <rguenther@suse.de>
18755 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
18756 (compute_antic): Seed worklist with exit block predecessors.
18757 * cfganal.c (dfs_find_deadend): For a cycle return the source
18758 of the edge closing it.
18760 2017-08-01 Tamar Christina <tamar.christina@arm.com>
18762 * config/aarch64/aarch64.c
18763 (aarch64_can_const_movi_rtx_p): Move 0 check.
18765 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18767 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
18768 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
18770 * match.pd: Ditto in address comparison pattern.
18772 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18774 PR tree-optimization/81627
18775 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
18776 closed ssa form for store-store chain.
18778 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18780 PR tree-optimization/81620
18781 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
18782 for store-store chain.
18784 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18786 PR tree-optimization/81588
18787 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
18788 ranges[i].in_p, invert comparison code ccode. For >/>=,
18789 swap rhs1 and rhs2 and comparison code unconditionally,
18790 for </<= don't do that. Don't swap rhs1/rhs2 again if
18791 ranges[i].in_p, instead invert comparison code ccode if
18792 opcode or oe->rank is BIT_IOR_EXPR.
18795 * optabs.def (vec_extract_optab, vec_init_optab): Change from
18796 a direct optab to conversion optab.
18797 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
18798 with GET_MODE_INNER as last argument instead of optab_handler.
18799 * expmed.c (extract_bit_field_1): Likewise. Use vector from
18800 vector extraction if possible and optab is available.
18801 * expr.c (store_constructor): Use convert_optab_handler instead
18802 of optab_handler. Use vector initialization from smaller
18803 vectors if possible and optab is available.
18804 * tree-vect-stmts.c (vectorizable_load): Likewise.
18805 * doc/md.texi (vec_extract, vec_init): Document that the optabs
18806 now have two modes.
18807 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
18808 of vec_init from half-sized vectors with the same element mode.
18809 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
18810 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
18811 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
18812 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
18813 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
18814 after mode in gen_vec_extract* calls.
18815 (vec_extract<mode>): Renamed to ...
18816 (vec_extract<mode><ssescalarmodelower>): ... this.
18817 (vec_extract<mode><ssehalfvecmodelower>): New expander.
18818 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
18819 element mode after mode in gen_vec_init* calls.
18820 (VEC_INIT_HALF_MODE): New mode iterator.
18821 (vec_init<mode>): Renamed to ...
18822 (vec_init<mode><ssescalarmodelower>): ... this.
18823 (vec_init<mode><ssehalfvecmodelower>): New expander.
18824 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
18825 (vec_extractv2sfsf): ... this.
18826 (vec_initv2sf): Renamed to ...
18827 (vec_initv2sfsf): ... this.
18828 (vec_extractv2si): Renamed to ...
18829 (vec_extractv2sisi): ... this.
18830 (vec_initv2si): Renamed to ...
18831 (vec_initv2sisi): ... this.
18832 (vec_extractv4hi): Renamed to ...
18833 (vec_extractv4hihi): ... this.
18834 (vec_initv4hi): Renamed to ...
18835 (vec_initv4hihi): ... this.
18836 (vec_extractv8qi): Renamed to ...
18837 (vec_extractv8qiqi): ... this.
18838 (vec_initv8qi): Renamed to ...
18839 (vec_initv8qiqi): ... this.
18840 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
18841 (vec_init<mode>): Renamed to ...
18842 (vec_init<mode><VEC_base_l>): ... this.
18843 (vec_extract<mode>): Renamed to ...
18844 (vec_extract<mode><VEC_base_l>): ... this.
18845 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
18846 (vec_initv2sfsf): ... this.
18847 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
18848 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
18849 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
18850 element mode after mode in gen_vec_init* calls.
18851 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
18852 (vec_init<mode><Vel>): ... this.
18853 (vec_extract<mode>): Renamed to ...
18854 (vec_extract<mode><Vel>): ... this.
18855 * config/aarch64/iterators.md (Vel): New mode attribute.
18856 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
18857 Add element mode after mode in gen_vec_extract* calls.
18858 * config/s390/vector.md (non_vec_l): New mode attribute.
18859 (vec_extract<mode>): Renamed to ...
18860 (vec_extract<mode><non_vec_l>): ... this.
18861 (vec_init<mode>): Renamed to ...
18862 (vec_init<mode><non_vec_l>): ... this.
18863 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
18864 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
18866 * config/arm/iterators.md (V_elem_l): New mode attribute.
18867 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
18868 (vec_extract<mode><V_elem_l>): ... this.
18869 (vec_extractv2di): Renamed to ...
18870 (vec_extractv2didi): ... this.
18871 (vec_init<mode>): Renamed to ...
18872 (vec_init<mode><V_elem_l>): ... this.
18873 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
18874 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
18875 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
18876 Add element mode after gen_vec_extract* calls.
18877 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
18878 (vec_init<mode><unitmode>): ... this.
18879 (vec_extract<mode>): Renamed to ...
18880 (vec_extract<mode><unitmode>): ... this.
18881 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
18882 (vec_init<mode><unitmode>): ... this.
18883 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
18884 (vec_initv2sfsf): ... this.
18885 (vec_extractv2sf): Renamed to ...
18886 (vec_extractv2sfsf): ... this.
18887 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
18888 Add element mode after gen_vec_extract* calls.
18889 * config/mips/mips.md (unitmode): New mode iterator.
18890 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
18891 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
18892 * config/spu/spu.md (inner_l): New mode attribute.
18893 (vec_init<mode>): Renamed to ...
18894 (vec_init<mode><inner_l>): ... this.
18895 (vec_extract<mode>): Renamed to ...
18896 (vec_extract<mode><inner_l>): ... this.
18897 * config/sparc/sparc.md (veltmode): New mode iterator.
18898 (vec_init<VMALL:mode>): Renamed to ...
18899 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
18900 * config/ia64/vect.md (vec_initv2si): Renamed to ...
18901 (vec_initv2sisi): ... this.
18902 (vec_initv2sf): Renamed to ...
18903 (vec_initv2sfsf): ... this.
18904 (vec_extractv2sf): Renamed to ...
18905 (vec_extractv2sfsf): ... this.
18906 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
18907 (vec_init<mode>): Renamed to ...
18908 (vec_init<mode><VEC_base_l>): ... this.
18909 (vec_extract<mode>): Renamed to ...
18910 (vec_extract<mode><VEC_base_l>): ... this.
18911 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
18912 (vec_initv2sfsf): ... this.
18913 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
18914 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
18915 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
18916 gen_vec_init* calls.
18918 2017-08-01 Richard Biener <rguenther@suse.de>
18920 PR tree-optimization/81297
18921 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
18922 TREE_OVERFLOW from INTEGER_CSTs.
18924 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
18926 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
18928 2017-07-31 Carl Love <cel@us.ibm.com>
18930 * config/rs6000/rs6000-c: Add support for built-in functions
18931 vector signed char vec_xl_be (signed long long, signed char *);
18932 vector unsigned char vec_xl_be (signed long long, unsigned char *);
18933 vector signed int vec_xl_be (signed long long, signed int *);
18934 vector unsigned int vec_xl_be (signed long long, unsigned int *);
18935 vector signed long long vec_xl_be (signed long long, signed long long *);
18936 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
18937 vector signed short vec_xl_be (signed long long, signed short *);
18938 vector unsigned short vec_xl_be (signed long long, unsigned short *);
18939 vector double vec_xl_be (signed long long, double *);
18940 vector float vec_xl_be (signed long long, float *);
18941 * config/rs6000/altivec.h (vec_xl_be): Add #define.
18942 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
18943 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
18945 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
18946 (altivec_expand_builtin): Add switch statement to call
18947 altivec_expand_xl_be for each builtin.
18948 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
18949 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
18950 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
18951 __builtin_vsx_le_be_v16qi.
18952 * doc/extend.texi: Update the built-in documentation file for the
18953 new built-in functions.
18955 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
18958 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
18960 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
18962 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18964 * config.gcc: Add z14.
18965 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
18966 CPU model numbers for z13s and z14.
18967 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
18969 * config/s390/s390-opts.h (enum processor_type): Rename
18970 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
18971 * config/s390/s390.c (processor_table): Add field for CPU name to
18972 be passed to Binutils.
18973 (s390_asm_output_machine_for_arch): Use the new field in
18974 processor_table for Binutils.
18975 (s390_expand_builtin): Replace arch12 with z14.
18976 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
18977 (s390_get_sched_attrmask): Likewise.
18978 (s390_get_unit_mask): Likewise.
18979 * config/s390/s390.opt: Add z14 to processor_type enum.
18981 2017-07-31 Martin Jambor <mjambor@suse.cz>
18984 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
18985 regardless of optimization level.
18987 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
18988 Martin Liska <mliska@suse.cz>
18990 * predict.def: Remove old comment and adjust probability.
18991 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
18992 PREDICT statements.
18994 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
18997 * config/i386/i386.c (ix86_function_naked): New function.
18998 (ix86_can_use_return_insn_p): Return false for naked functions.
18999 (ix86_expand_prologue): Skip prologue for naked functions.
19000 (ix86_expand_epilogue): Skip epilogue for naked functions
19001 and emit trap instruction.
19002 (ix86_warn_func_return): New function.
19003 (ix86_attribute_table): Add "naked" attribute specification.
19004 (TARGET_WARN_FUNC_RETURN): Define.
19005 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
19007 2017-07-31 Martin Liska <mliska@suse.cz>
19009 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
19010 (dump_gimple_bb_header): Always dump BB info.
19011 (pp_cfg_jump): Do not append info about BB when dumping a jump.
19013 2017-07-31 Martin Liska <mliska@suse.cz>
19016 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
19017 also with current_function_decl non-null equality.
19019 2017-07-31 Jakub Jelinek <jakub@redhat.com>
19022 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
19023 change type to the element type, instead add eltype variable and
19024 use it where we are interested in the element type.
19026 PR tree-optimization/81603
19027 * ipa-polymorphic-call.c
19028 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
19029 offset arithmetic in offset_int, bail out if the resulting bit offset
19030 doesn't fit into shwi.
19032 2017-07-31 Martin Liska <mliska@suse.cz>
19034 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
19035 (gimplify_save_expr): Fix comment.
19037 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
19040 * config/i386/i386.c (ix86_function_arg): Update arguments for
19042 (ix86_compute_frame_layout): Set the initial stack offset to
19043 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
19044 INCOMING_FRAME_SP_OFFSET.
19045 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
19046 stack before exception handler returns.
19047 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
19048 the 'ERROR_CODE' for exception handler.
19050 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
19052 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
19053 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
19054 (ASM_OUTPUT_REG_POP): Ditto.
19055 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
19056 instead of asm_fprintf to output pure string.
19058 2017-07-29 Jakub Jelinek <jakub@redhat.com>
19060 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
19061 to imported_module_or_decl hook.
19062 (debug_nothing_tree_tree_tree_bool): Remove.
19063 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
19064 * debug.c (do_nothing_debug_hooks): Use
19065 debug_nothing_tree_tree_tree_bool_bool instead of
19066 debug_nothing_tree_tree_tree_bool.
19067 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19068 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
19069 * sdbout.c (sdb_debug_hooks): Likewise.
19070 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
19071 (gen_namespace_die): Add DW_AT_export_symbols attribute if
19073 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
19074 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
19075 attribute, don't add anything.
19077 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19079 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
19080 (fold_build2_stat_loc): Likewise.
19081 (fold_build3_stat_loc): Likewise.
19082 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
19083 (fold_build1_loc): Remove macro.
19084 (fold_build2_loc): Likewise.
19085 (fold_build3_loc): Likewise.
19087 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19089 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
19090 (gimple_build_debug_bind_source_stat): Likewise.
19091 * gimple.h (gimple_build_debug_bind): Remove macro.
19092 (gimple_build_debug_bind_source): Likewise.
19094 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19096 * bitmap.c (bitmap_alloc): Adjust.
19097 (bitmap_gc_alloc): Likewise.
19098 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
19100 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19102 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
19103 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
19104 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
19105 (bitmap_gc_alloc_stat): Likewise.
19106 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
19108 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19110 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
19111 * rtl.h (shallow_copy_rtx): Remove macro.
19113 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19115 * emit-rtl.c (gen_raw_REG): Adjust.
19116 * gengenrtl.c (gendef): Likewise.
19117 * rtl.c (rtx_alloc_stat): Remove _stat from name.
19118 * rtl.h (rtx_alloc): Remove macro.
19120 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19122 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
19123 (build_tree_list_stat): Likewise.
19124 * tree.h (build_tree_list): Remove macro.
19125 (build_tree_list_vec): Likewise.
19127 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19129 * tree.c (make_vector_stat): Remove _stat from name.
19130 (build_vector_stat): Likewise.
19131 * tree.h (make_vector_stat): Remove macro.
19132 (build_vector_stat): Likewise.
19134 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19136 * tree.h (build_var_debug_value): Remove prototype.
19138 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19140 * tree.c (tree_cons_stat): Remove _stat from name.
19141 * tree.h (tree_cons): Remove macro.
19143 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19145 * tree.c (build_vl_exp_stat): Remove _stat from name.
19146 * tree.h (build_vl_exp): Remove macro.
19148 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19150 * tree.c (build_decl_stat): Remove _stat from name.
19151 * tree.h (build_decl): Remove macro.
19153 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19155 * gimple.c (gimple_build_with_ops_stat): Adjust.
19156 (gimple_alloc_stat): Remove _stat from name.
19157 * gimple.h (gimple_alloc): Remove macro.
19159 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19161 * tree.c (make_tree_vec_stat): Remove _stat from name.
19162 (grow_tree_vec_stat): Likewise.
19163 * tree.h (make_tree_vec_stat): Adjust prototype.
19164 (grow_tree_vec_stat): Likewise.
19165 (make_tree_vec): Remove macro.
19166 (grow_tree_vec): Likewise.
19168 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19170 * fold-const.c (fold_build1_stat_loc): Adjust.
19171 (fold_build2_stat_loc): Likewise.
19172 (fold_build3_stat_loc): Likewise.
19173 * tree.c (build0_stat): Remove _stat from name.
19174 (build1_stat): Likewise.
19175 (build2_stat): Likewise.
19176 (build3_stat): Likewise.
19177 (build4_stat): Likewise.
19178 (build5_stat): Likewise.
19179 * tree.h (build1_loc): Remove macro, and rename _stat function
19181 (build2_loc): Likewise.
19182 (build3_loc): Likewise.
19183 (build4_loc): Likewise.
19184 (build5_loc): Likewise.
19186 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19188 * tree.c (make_int_cst_stat): Remove _stat from name.
19189 * tree.h (make_int_cst_stat): Adjust prototype.
19190 (make_int_cst): Remove macro.
19192 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19194 * tree.c (make_tre_binfo_stat): Remove _stat from name.
19195 * tree.h (make_tree_binfo_stat): Adjust prototype.
19196 (make_tree_binfo): Remove.
19198 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19200 * tree.c (copy_node_stat): Rename to copy_node.
19201 (build_distinct_type_copy): Adjust.
19202 * tree.h (copy_node_stat): Adjust prototype.
19203 (copy_node): Remove macro.
19205 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19207 * tree.c (make_node_stat): rename to make_node.
19208 (build_tree_list_stat): Adjust.
19209 (build0_stat): Likewise.
19210 (build2_stat): Likewise.
19211 (build3_stat): Likewise.
19212 (build4_stat): Likewise.
19213 (build5_stat): Likewise.
19214 (build_decl_stat): Likewise.
19215 * tree.h (make_node_stat): Adjust prototype.
19216 (make_node): remove macro.
19218 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
19220 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
19221 (PPC_FEATURE2_SCV): Likewise.
19222 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
19224 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19226 * config/aarch64/aarch64.c
19227 (aarch64_internal_mov_immediate): Add new special pattern.
19228 * config/aarch64/aarch64.md (*movdi_aarch64):
19229 Add reg/32bit const mov case.
19231 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19232 Richard Sandiford <richard.sandiford@linaro.org>
19234 * config/aarch64/aarch64.md (mov<mode>): Generalize.
19235 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
19236 Add integer and movi cases.
19237 (movi-split-hf-df-sf split, fp16): New.
19238 (enabled): Added TARGET_FP_F16INST.
19239 * config/aarch64/iterators.md (GPF_HF): New.
19240 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
19242 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19244 * config/aarch64/aarch64.c
19245 (aarch64_simd_container_mode): Add prototype.
19246 (aarch64_expand_mov_immediate): Add HI support.
19247 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
19248 (aarch64_can_const_movi_rtx_p): New.
19249 (aarch64_preferred_reload_class):
19250 Remove restrictions of using FP registers for certain SIMD operations.
19251 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
19252 (aarch64_valid_floating_const): Add integer move validation.
19253 (aarch64_simd_imm_scalar_p): Remove.
19254 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
19255 (aarch64_legitimate_constant_p): Expand list of supported cases.
19256 * config/aarch64/aarch64-protos.h
19257 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
19258 (aarch64_reinterpret_float_as_int): New.
19259 (aarch64_simd_imm_scalar_p): Remove.
19260 * config/aarch64/constraints.md (Uvi): New.
19261 (Dd): Split into Ds and new Dd.
19262 * config/aarch64/aarch64.md (*movsi_aarch64):
19264 (*movdi_aarch64): Add SIMD mov case.
19266 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19268 * tree-predcom.c: (struct chain): Handle store-store chain in which
19269 stores for elimination only store loop invariant values.
19270 (execute_pred_commoning_chain): Ditto.
19271 (prepare_initializers_chain_store_elim): Ditto.
19272 (prepare_finalizers): Ditto.
19273 (is_inv_store_elimination_chain): New function.
19274 (initialize_root_vars_store_elim_1): New function.
19276 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19278 * tree-predcom.c: Revise general description of the pass.
19279 (enum chain_type): New enum type for store elimination.
19280 (struct chain): New field supporting store elimination.
19281 (struct component): Ditto.
19282 (dump_chain): Dump store-stores chain.
19283 (release_chain): Release resources.
19284 (split_data_refs_to_components): Compute and create component
19285 contains only stores for elimination.
19286 (get_chain_last_ref_at): New function.
19287 (make_invariant_chain): Initialization.
19288 (make_rooted_chain): Specify chain type in parameter and record it.
19289 (add_looparound_copies): Skip for store-stores chain.
19290 (determine_roots_comp): Compute type of chain and pass it to
19292 (initialize_root_vars_store_elim_2): New function.
19293 (finalize_eliminated_stores): New function.
19294 (remove_stmt): Handle store for elimination.
19295 (execute_pred_commoning_chain): Execute predictive commoning on
19296 store-store chains.
19297 (determine_unroll_factor): Skip unroll for store-stores chain.
19298 (prepare_initializers_chain_store_elim): New function.
19299 (prepare_initializers_chain): Hanlde store-store chain.
19300 (prepare_finalizers_chain, prepare_finalizers): New function.
19301 (tree_predictive_commoning_loop): Return integer value indicating
19302 if loop is unrolled or lcssa form is corrupted.
19303 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
19305 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19307 * tree-predcom.c (initialize_root): Delete.
19308 (execute_pred_commoning_chain): Initialize root vars and replace
19309 reference of non-combined chain directly, rather than call above
19312 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19314 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
19315 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
19317 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19319 * tree-predcom.c (struct chain): New field init_seq.
19320 (release_chain): Release init_seq.
19321 (prepare_initializers_chain): Record intialization stmts in above
19323 (insert_init_seqs): New function.
19324 (tree_predictive_commoning_loop): Call insert_init_seqs.
19326 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19328 * tree-predcom.c (determine_roots_comp): Skip trivial components.
19330 2017-07-28 Richard Biener <rguenther@suse.de>
19332 * match.pd: Remove superfluous :c.
19333 * genmatch.c (simplify::id): Add member.
19334 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
19336 (current_id): New global.
19337 (dt_node::parent): Move from ...
19338 (dt_operand::parent): ... here. Add for_id member.
19339 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
19340 (decision_tree::find_node): Relax order requirement when
19341 merging DT_TRUE nodes to ones inbetween the current simplify
19342 and the one we try to merge with. Add diagnostic whenever
19343 we need to enforce pattern order by not merging.
19344 (decision_tree::insert): Set current_id.
19345 (decision_tree::print_node): Dump parent node and for_id.
19346 (parser::last_id): Add member.
19347 (parser::push_simplify): Assign unique id.
19348 (parser::parser): Initialize last_id.
19350 2017-07-28 Martin Liska <mliska@suse.cz>
19353 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
19354 gimple_build_debug_bind.
19356 2017-07-28 Richard Biener <rguenther@suse.de>
19358 PR tree-optimization/81502
19359 * match.pd: Add pattern combining BIT_INSERT_EXPR with
19361 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
19363 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
19364 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
19365 for BIT_FIELD_REF args.
19366 * fold-const.c (make_bit_field_ref): Likewise.
19367 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
19369 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19372 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
19373 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
19374 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
19375 Or it into SANITIZER_UNDEFINED.
19376 * ubsan.c: Include gimple-fold.h and varasm.h.
19377 (ubsan_expand_ptr_ifn): New function.
19378 (instrument_pointer_overflow): New function.
19379 (maybe_instrument_pointer_overflow): New function.
19380 (instrument_object_size): Formatting fix.
19381 (pass_ubsan::execute): Call instrument_pointer_overflow
19382 and maybe_instrument_pointer_overflow.
19383 * internal-fn.c (expand_UBSAN_PTR): New function.
19384 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
19385 * sanitizer.def (__ubsan_handle_pointer_overflow,
19386 __ubsan_handle_pointer_overflow_abort): New builtins.
19387 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
19388 * internal-fn.def (UBSAN_PTR): New internal function.
19389 * opts.c (sanitizer_opts): Add pointer-overflow.
19390 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
19391 * fold-const.c (build_range_check): Compute pointer range check in
19392 integral type if pointer arithmetics would be needed. Formatting
19395 2017-07-28 Martin Liska <mliska@suse.cz>
19398 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
19399 parameters that are of a variable-length.
19401 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
19403 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
19405 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
19406 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
19407 (CRT_CALL_STATIC_FUNCTION): Likewise.
19408 (ASM_DEFAULT_SPEC): New define.
19409 (ASM_SPEC32): Likewise.
19410 (ASM_SPEC64): Likewise.
19411 (ASM_SPEC_COMMON): Likewise.
19412 (ASM_SPEC): Likewise.
19413 (INVALID_64BIT): Likewise.
19414 (LINK_OS_DEFAULT_SPEC): Likewise.
19415 (LINK_OS_SPEC32): Likewise.
19416 (LINK_OS_SPEC64): Likewise.
19417 (POWERPC_LINUX): Likewise.
19418 (PTRDIFF_TYPE): Likewise.
19419 (RESTORE_FP_PREFIX): Likewise.
19420 (RESTORE_FP_SUFFIX): Likewise.
19421 (SAVE_FP_PREFIX): Likewise.
19422 (SAVE_FP_SUFFIX): Likewise.
19423 (SIZE_TYPE): Likewise.
19424 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
19425 (TARGET_64BIT): Likewise.
19426 (TARGET_64BIT): Likewise.
19427 (TARGET_AIX): Likewise.
19428 (WCHAR_TYPE_SIZE): Likewise.
19429 (WCHAR_TYPE): Undefine.
19430 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
19431 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
19432 (CPP_OS_RTEMS_SPEC): Delete.
19433 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
19434 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
19436 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
19438 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19440 PR tree-optimization/81578
19441 * tree-parloops.c (build_new_reduction): Bail out if
19442 reduction_code isn't one of the standard OpenMP reductions.
19443 Move the details printing after that decision.
19445 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19447 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
19448 related to reload_in_progress.
19449 (splat_input_operand): Likewise.
19450 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
19452 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
19454 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
19455 (TARGET_INSTANTIATE_DECLS): Likewise.
19456 (legitimate_indexed_address_p): Delete reload_in_progress code.
19457 (rs6000_debug_legitimate_address_p): Likewise.
19458 (rs6000_eliminate_indexed_memrefs): Likewise.
19459 (rs6000_emit_le_vsx_store): Likewise.
19460 (rs6000_emit_move_si_sf_subreg): Likewise.
19461 (rs6000_emit_move): Likewise.
19462 (register_to_reg_type): Likewise.
19463 (rs6000_pre_atomic_barrier): Likewise.
19464 (rs6000_machopic_legitimize_pic_address): Likewise.
19465 (rs6000_allocate_stack_temp): Likewise.
19466 (rs6000_address_for_fpconvert): Likewise.
19467 (rs6000_address_for_altivec): Likewise.
19468 (rs6000_secondary_memory_needed_rtx): Delete function.
19469 (rs6000_check_sdmode): Likewise.
19470 (rs6000_alloc_sdmode_stack_slot): Likewise.
19471 (rs6000_instantiate_decls): Likewise.
19472 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
19473 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
19474 Delete reload_in_progress.
19475 (*vec_reload_and_plus_<mptrsize>): Likewise.
19476 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
19477 (vsx_div_v2di): Likewise.
19478 (vsx_udiv_v2di): Likewise.
19480 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19482 * config/rs6000/rs6000.opt (mlra): Replace with stub.
19483 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
19484 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
19485 (rs6000_debug_reg_global): Delete print of LRA status.
19486 (rs6000_option_override_internal): Delete dead LRA related code.
19487 (rs6000_lra_p): Delete function.
19488 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
19490 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
19492 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
19493 * config/riscv/rtems.h: New file.
19495 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19496 Sudakshina Das <sudi.das@arm.com>
19498 * config/aarch64/aarch64.md
19499 (define_split for and<mode>3nr_compare): Move
19500 non aarch64_logical_operand to a register.
19501 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
19502 register immediate operand to a register.
19503 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
19505 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19507 PR middle-end/81564
19508 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
19510 2017-07-27 Richard Biener <rguenther@suse.de>
19512 PR tree-optimization/81573
19513 PR tree-optimization/81494
19514 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
19515 multi defuse cycle case.
19517 2017-07-27 Richard Biener <rguenther@suse.de>
19519 PR tree-optimization/81571
19520 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
19523 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
19525 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
19526 earlier and only if MASK_FPU is set. Adjust formatting.
19528 2017-07-27 Martin Liska <mliska@suse.cz>
19530 * opt-functions.awk: Add validation of value of Init.
19531 * optc-gen.awk: Pass new argument.
19533 2017-07-27 Martin Liska <mliska@suse.cz>
19535 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
19536 Fix wrong condition.
19538 2017-07-27 Martin Liska <mliska@suse.cz>
19540 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
19541 BBs and edges seen by autoFDO.
19543 2017-07-27 Richard Biener <rguenther@suse.de>
19545 PR tree-optimization/81502
19546 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
19547 with incompatible but same sized type.
19548 (execute_update_addresses_taken): Likewise.
19550 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
19552 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
19553 flag_tree_loop_vectorize rather than flag_tree_vectorize.
19555 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19558 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
19559 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
19560 Change s_operand to memory_operand.
19562 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
19564 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
19565 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
19566 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
19567 Emit instructions rather than returning an expression. Handle TFmode
19568 and KFmode by casting to TImode.
19569 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
19570 (rs6000_emit_le_vsx_store): Likewise.
19571 * config/rs6000/vsx.md (VSX_TI): New iterator.
19572 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
19573 (*vsx_le_undo_permute_<mode>): Likewise.
19574 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
19575 emit the split sequence.
19576 (*vsx_le_perm_store_<mode>): Likewise.
19578 2017-07-27 Jakub Jelinek <jakub@redhat.com>
19580 PR tree-optimization/81555
19581 PR tree-optimization/81556
19582 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
19583 if true, force CHANGED for the recursive invocation.
19584 (reassociate_bb): Remember original length of ops array, pass
19585 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
19587 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
19588 attributes for noipa attribute. For naked attribute use
19589 lookup_attribute first before lookup_attribute_spec.
19590 * final.c (rest_of_handle_final): Disable IPA RA for functions with
19592 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
19593 for functions with noipa attribute.
19594 (cgraph_externally_visible_p): Return true for functions with noipa
19596 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
19597 for functions with noipa attribute.
19598 * doc/extend.texi: Document noipa function attribute.
19599 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
19600 also for functions with noipa attribute.
19601 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
19603 2017-07-26 Andrew Pinski <apinski@cavium.com>
19605 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
19606 vec_unalign_load_cost and vec_unalign_store_cost.
19608 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
19610 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
19611 -mvsx-small-integer option.
19612 (ISA_3_0_MASKS_IEEE): Likewise.
19613 (OTHER_VSX_VECTOR_MASKS): Likewise.
19614 (POWERPC_MASKS): Likewise.
19615 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
19616 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
19617 code, only testing for DImode being allowed in non-VSX floating
19619 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
19620 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
19622 (rs6000_option_override_internal): Delete -mvsx-small-integer.
19623 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
19624 TARGET_P8_VECTOR test.
19625 (rs6000_secondary_reload_simple_move): Likewise.
19626 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
19627 since TARGET_P9_VECTOR was already tested.
19628 (rs6000_opt_masks): Remove -mvsx-small-integer.
19629 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
19630 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19632 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
19633 test for TARGET_VEXTRACTUB was used, and that uses
19635 (p9 extract splitter): Likewise.
19636 (vsx_extract_<mode>_di_p9): Likewise.
19637 (vsx_extract_<mode>_store_p9): Likewise.
19638 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
19639 for TARGET_P9_VECTOR was used. Delete code that is now dead with
19640 the elimination of TARGET_VSX_SMALL_INTEGER.
19641 (vsx_extract_<mode>_p8): Likewise.
19642 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
19643 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
19644 (vsx_set_<mode>_p9): Likewise.
19645 (vsx_set_v4sf_p9): Likewise.
19646 (vsx_set_v4sf_p9_zero): Likewise.
19647 (vsx_insert_extract_v4sf_p9): Likewise.
19648 (vsx_insert_extract_v4sf_p9_2): Likewise.
19649 * config/rs6000/rs6000.md (sign extend splitter): Change
19650 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
19651 (floatsi<mode>2_lfiwax_mem): Likewise.
19652 (floatunssi<mode>2_lfiwzx_mem): Likewise.
19653 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
19654 since a test for TARGET_P9_VECTOR was used.
19655 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19656 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
19657 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19658 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
19659 TARGET_P8_VECTOR test.
19660 (fix_trunc<mode>si2_stfiwx): Likewise.
19661 (fix_trunc<mode>si2_internal): Likewise.
19662 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
19663 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19665 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19666 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
19667 TARGET_P8_VECTOR test.
19668 (fixuns_trunc<mode>si2_stfiwx): Likewise.
19669 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
19670 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19672 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19673 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
19674 since a test for TARGET_P9_VECTOR was used.
19675 (splitter for loading small constants): Likewise.
19677 2017-07-26 Andrew Pinski <apinski@cavium.com>
19679 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
19682 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
19685 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
19686 (fp_valid_at): Likewise.
19688 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
19690 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
19691 (qdf24xx_addrcost_table): Likewise.
19692 (cortexa57_tunings): Update to use generic_branch_cost.
19693 (cortexa72_tunings): Likewise.
19694 (cortexa73_tunings): Likewise.
19695 (qdf24xx_tunings): Likewise.
19697 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
19699 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
19700 (thunderx2t99_branch_cost): Likewise.
19701 (cortexa35_tunings): Update to use generic_branch_cost.
19702 (cortexa53_tunings): Likewise.
19703 (cortexa57_tunings): Likewise.
19704 (cortexa72_tunings): Likewise.
19705 (cortexa73_tunings): Likewise.
19706 (thunderx2t99_tunings): Likewise.
19708 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19710 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
19711 (sparc_option_override): Honour MASK_FSMULD.
19712 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
19713 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
19714 * config/sparc/sparc.opt (mfsmuld): New option.
19715 * doc/invoke.texi (mfsmuld): Document option.
19717 2017-07-26 Marek Polacek <polacek@redhat.com>
19719 PR middle-end/70992
19720 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
19722 2017-07-26 Richard Biener <rguenther@suse.de>
19724 * gimple-match-head.c (do_valueize): Return OP if valueize
19726 (get_def): New helper to get at the def stmt of a SSA name
19727 if valueize allows.
19728 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
19729 do_valueize to get at the def stmt.
19730 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
19732 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
19734 PR middle-end/46932
19735 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
19737 2017-07-26 Martin Liska <mliska@suse.cz>
19740 * function.c (expand_function_start): Make expansion of
19741 nonlocal_goto_save_area after parm_birth_insn.
19743 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19745 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
19746 from all CPU target flags enable members.
19748 2017-07-26 Richard Biener <rguenther@suse.de>
19750 * genmatch.c (dt_simplify::gen): Make iterator vars const.
19751 (decision_tree::gen): Make 'type' const.
19752 (write_predicate): Likewise.
19754 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
19756 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
19757 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
19758 (rs6000_option_override_internal): Likewise.
19759 (rs6000_expand_vector_set): Likewise.
19760 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
19761 (TARGET_UPPER_REGS_SF): Likewise.
19762 (TARGET_UPPER_REGS_DI): Likewise.
19763 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
19764 (TARGET_DIRECT_MOVE_64BIT): Likewise.
19765 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
19766 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19767 (Splitters for DI constants in Altivec registers): Likewise.
19768 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
19769 (vsx_set_v4sf_p9): Likewise.
19770 (vsx_set_v4sf_p9_zero): Likewise.
19771 (vsx_insert_extract_v4sf_p9): Likewise.
19772 (vsx_insert_extract_v4sf_p9_2): Likewise.
19774 2017-07-25 Carl Love <cel@us.ibm.com>
19776 * doc/extend.texi: Update the built-in documentation file for the
19777 existing built-in functions
19778 vector signed char vec_cnttz (vector signed char);
19779 vector unsigned char vec_cnttz (vector unsigned char);
19780 vector signed short vec_cnttz (vector signed short);
19781 vector unsigned short vec_cnttz (vector unsigned short);
19782 vector signed int vec_cnttz (vector signed int);
19783 vector unsigned int vec_cnttz (vector unsigned int);
19784 vector signed long long vec_cnttz (vector signed long long);
19785 vector unsigned long long vec_cnttz (vector unsigned long long);
19787 2017-07-25 Andrew Pinski <apinski@cavium.com>
19789 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
19790 accesses where the use is for the first operand of a BIT_INSERT.
19792 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19795 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
19796 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
19798 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19800 * config/i386/gstabs.h: Delete.
19801 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
19803 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
19805 * config/i386/i386.c (ix86_decompose_address): Do not check for
19806 register RTX when looking at index_reg or base_reg.
19807 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
19809 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
19811 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
19812 to update EH info here.
19814 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19816 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
19818 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19820 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
19822 2017-07-25 Torsten Duwe <duwe@suse.de>
19824 * common.opt: Introduce -fpatchable-function-entry
19825 command line option, and its variables function_entry_patch_area_size
19826 and function_entry_patch_area_start.
19827 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
19828 including a two-value parser.
19829 * target.def (print_patchable_function_entry): New target hook.
19830 * targhooks.h (default_print_patchable_function_entry): New function.
19831 * targhooks.c (default_print_patchable_function_entry): Likewise.
19832 * toplev.c (process_options): Switch off IPA-RA if
19833 patchable function entries are being generated.
19834 * varasm.c (assemble_start_function): Look at the
19835 patchable-function-entry command line switch and current
19836 function attributes and maybe generate NOP instructions by
19837 calling the print_patchable_function_entry hook.
19838 * doc/extend.texi: Document patchable_function_entry attribute.
19839 * doc/invoke.texi: Document -fpatchable_function_entry
19840 command line option.
19841 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
19843 * doc/tm.texi: Re-generate.
19845 2017-07-25 Jakub Jelinek <jakub@redhat.com>
19848 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
19849 TARGET_AVX512DQ rather than TARGET_AVX512BW.
19851 2017-07-25 Tamar Christina <tamar.christina@arm.com>
19853 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
19855 2017-07-25 Richard Biener <rguenther@suse.de>
19857 PR tree-optimization/81455
19858 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
19859 not walk in cycles when looking for guards.
19861 2017-07-25 Richard Biener <rguenther@suse.de>
19863 PR tree-optimization/81529
19864 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
19865 when optimizing backedge uses.
19867 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
19869 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
19871 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
19872 to dl_section_ref. On AIX, append an expression to subtract
19873 the size of the section length to dl_section_ref.
19875 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
19877 * configure.ac: If any of the config.* scripts fail, exit 1.
19878 * configure: Regenerate.
19880 2017-07-25 Richard Biener <rguenther@suse.de>
19882 PR middle-end/81546
19883 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
19884 of immediate uses, be more verbose on errors.
19886 2017-07-25 Richard Biener <rguenther@suse.de>
19888 PR tree-optimization/81510
19889 * tree-vect-loop.c (vect_is_simple_reduction): When the
19890 reduction stmt is not inside the loop bail out.
19892 2017-07-25 Richard Biener <rguenther@suse.de>
19894 PR tree-optimization/81303
19895 * tree-vect-loop-manip.c (vect_loop_versioning): Build
19896 profitability check against LOOP_VINFO_NITERSM1.
19898 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19900 * domwalk.c (cmp_bb_postorder): Simplify.
19901 (sort_bbs_postorder): New function. Use it...
19902 (dom_walker::walk): ...here to optimize common cases.
19904 2017-07-25 Martin Liska <mliska@suse.cz>
19907 * ipa-visibility.c (function_and_variable_visibility): Make the
19908 redirection just on target that supports aliasing.
19909 Fix GNU coding style.
19911 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
19914 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
19916 * config/arm/rtems.h: Likewise.
19917 * config/bfin/rtems.h: Likewise.
19918 * config/i386/rtemself.h: Likewise.
19919 * config/lm32/rtems.h: Likewise.
19920 * config/m32c/rtems.h: Likewise.
19921 * config/m68k/rtemself.h: Likewise.
19922 * config/microblaze/rtems.h: Likewise.
19923 * config/mips/rtems.h: Likewise.
19924 * config/moxie/rtems.h: Likewise.
19925 * config/nios2/rtems.h: Likewise.
19926 * config/powerpcspe/rtems.h: Likewise.
19927 * config/rs6000/rtems.h: Likewise.
19928 * config/rtems.h: Likewise.
19929 * config/sh/rtems.h: Likewise.
19930 * config/sh/rtemself.h: Likewise.
19931 * config/sparc/rtemself.h: Likewise.
19933 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
19936 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
19937 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
19938 * tree-ssa-structalias.c (alias_get_name): Same.
19940 2017-07-25 Bin Cheng <bin.cheng@arm.com>
19943 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
19944 instructions if no du chain is found.
19946 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
19948 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
19950 2017-07-25 Richard Biener <rguenther@suse.de>
19952 PR middle-end/81505
19953 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
19956 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
19958 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
19959 upper-regs options.
19960 (ISA_2_7_MASKS_SERVER): Likewise.
19961 (ISA_3_0_MASKS_IEEE): Likewise.
19962 (OTHER_P8_VECTOR_MASKS): Likewise.
19963 (OTHER_VSX_VECTOR_MASKS): Likewise.
19964 (POWERPC_MASKS): Likewise.
19965 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
19966 duplicate list of options.
19967 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
19968 explicit -mupper-regs options.
19969 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
19970 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
19971 alias for -mupper-regs-df.
19972 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19973 (rs6000_init_hard_regno_mode_ok): Likewise.
19974 (rs6000_option_override_internal): Likewise.
19975 (rs6000_opt_masks): Likewise.
19976 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
19977 options in terms of whether -mvsx or -mpower8-vector was used.
19978 (TARGET_UPPER_REGS_DI): Likewise.
19979 (TARGET_UPPER_REGS_SF): Likewise.
19980 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
19981 -mupper-regs-* options.
19983 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
19985 * passes.c (emergency_dump_function): Print some empty lines and a
19986 header before the RTL dump.
19988 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
19990 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
19992 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
19995 * config/aarch64/aarch64.c (aarch64_classify_symbol):
19996 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
19998 2017-07-24 Carl Love <cel@us.ibm.com>
20000 * config/rs6000/rs6000-c.c: Add support for built-in functions
20001 vector float vec_extract_fp32_from_shorth (vector unsigned short);
20002 vector float vec_extract_fp32_from_shortl (vector unsigned short);
20003 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
20004 vec_extract_fp_from_shortl): Add defines for the two builtins.
20005 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
20006 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
20008 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
20009 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
20010 * doc/extend.texi: Update the built-in documentation file for the
20011 new built-in function.
20013 2017-07-24 Jakub Jelinek <jakub@redhat.com>
20016 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
20018 * doc/generic.texi: Likewise.
20019 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
20020 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
20022 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
20024 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
20025 (aarch64_mls_elt_merge<mode>): Likewise.
20027 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
20029 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
20030 having __cxa_atexit.
20032 2017-07-23 Michael Collison <michael.collison@arm.com>
20034 * config/arm/arm.c (arm_option_override): Deprecate
20035 use of -mstructure-size-boundary.
20036 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
20037 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
20039 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20042 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
20043 Reduce cost estimate for direct moves.
20045 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
20048 * config/i386/i386.c (ix86_option_override_internal): Disable
20049 BMI, BMI2 and TBM instructions for -m16.
20051 2017-07-21 Carl Love <cel@us.ibm.com>
20053 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20054 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20055 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20056 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20057 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20058 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20059 VMULOSW): New enum "unspec" values.
20060 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20061 altivec_vmulosw): New patterns.
20062 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20063 VMULOSW): Add definitions.
20065 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
20067 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
20068 (qdf24xx): Likewise.
20069 * config/aarch64/aarch64-options-extensions.def (rdma); New.
20070 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
20071 (AARCH64_FL_V8_1): Renumber.
20072 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
20073 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
20074 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
20075 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
20076 rdma to feature modifiers list.
20078 2017-07-21 Yury Gribov <tetra2005@gmail.com>
20080 PR middle-end/56727
20081 * ipa-visibility (function_and_variable_visibility): Convert
20082 recursive PLT call to direct call if appropriate.
20084 2017-07-21 Andrew Pinski <apinski@cavium.com>
20086 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
20087 operand 1 to see if the types precision matches.
20088 * fold-const.c (operand_equal_p): Likewise.
20090 2017-07-21 Richard Biener <rguenther@suse.de>
20092 PR tree-optimization/81303
20093 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
20094 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
20095 (vect_peeling_hash_get_lowest_cost): Adjust.
20096 (vect_enhance_data_refs_alignment): Likewise. Use
20097 vect_get_peeling_costs_all_drs to compute the penalty for no
20098 peeling to match up costs.
20100 2017-07-21 Richard Biener <rguenther@suse.de>
20102 PR tree-optimization/81500
20103 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
20104 we didn't identify a reduction path.
20106 2017-07-21 Tom de Vries <tom@codesourcery.com>
20107 Cesar Philippidis <cesar@codesourcery.com>
20109 PR gcov-profile/81442
20110 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
20113 2017-07-21 Tom de Vries <tom@codesourcery.com>
20116 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
20118 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
20119 nvptx_override_options_after_change.
20121 2017-07-21 Ulrich Drepper <drepper@redhat.com>
20123 * dwarf2out.c (output_file_names): Avoid double testing for
20124 dwarf_version >= 5.
20126 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
20128 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
20130 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20132 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
20134 (try_crossjump_to_edge): Do not punt on partitioned functions.
20136 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20138 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
20139 Put all BBs reachable only via paths crossing cold region to cold
20141 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
20143 2016-07-21 Richard Biener <rguenther@suse.de>
20145 PR tree-optimization/81303
20146 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
20147 into account prologue and epilogue iterations when raising
20148 min_profitable_iters to sth at least covering one vector iteration.
20150 2017-07-21 Tamar Christina <tamar.christina@arm.com>
20152 * config/arm/arm.c (arm_test_cpu_arch_dat):
20155 2017-07-20 Nathan Sidwell <nathan@acm.org>
20157 Remove TYPE_METHODS.
20158 * tree.h (TYPE_METHODS): Delete.
20159 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
20160 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
20161 (dbxout_type_methods): Scan TYPE_FIELDS.
20162 (dbxout_type): Don't check TYPE_METHODS here.
20163 * function.c (use_register_for_decl): Always ignore register for
20164 class types when not optimizing.
20165 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
20166 * tree.c (free_lang_data_in_type): Stitch out member functions and
20167 templates from TYPE_FIELDS.
20168 (build_distinct_type_copy, verify_type_variant,
20169 verify_type): Member fns are on TYPE_FIELDS.
20170 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
20171 * tree-pretty-print.c (dump_generic_node): Likewise.
20173 2017-07-20 Jakub Jelinek <jakub@redhat.com>
20176 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
20177 V2TImode and V4TImode.
20178 (ix86_expand_vector_extract): Likewise.
20179 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
20180 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
20181 (ssescalarmode): Handle V4TImode and V2TImode.
20182 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
20183 (*vec_extractv2ti, *vec_extractv4ti): New insns.
20184 (VEXTRACTI128_MODE): New mode iterator.
20185 (splitter for *vec_extractv?ti first element): New.
20186 (VEC_INIT_MODE): New mode iterator.
20187 (vec_init<mode>): Consolidate 3 expanders into one using
20188 VEC_INIT_MODE mode iterator.
20190 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20192 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
20193 non_spilled_static_chain_regno_p.
20195 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20197 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
20199 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
20201 * bb-reorder.c (connect_traces): Allow copying of blocks within
20204 2017-07-20 Richard Biener <rguenther@suse.de>
20206 * gimple.h (gimple_phi_result): Add gphi * overload.
20207 (gimple_phi_result_ptr): Likewise.
20208 (gimple_phi_arg): Likewise. Adjust index assert to only
20209 allow actual argument accesses rather than all slots available
20211 (gimple_phi_arg_def): Add gphi * overload.
20212 * tree-phinodes.c (make_phi_node): Initialize only actual
20214 (resize_phi_node): Clear memory not covered by old node,
20215 do not initialize excess argument slots.
20216 (reserve_phi_args_for_new_edge): Initialize new argument slot
20219 2017-07-20 Bin Cheng <bin.cheng@arm.com>
20221 PR tree-optimization/81388
20223 2016-07-21 Bin Cheng <bin.cheng@arm.com>
20225 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
20226 by removing computation of may_be_zero.
20228 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20229 Tom de Vries <tom@codesourcery.com>
20231 PR middle-end/81030
20232 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
20233 when gimple level profile disagrees with what RTL expander did.
20235 2017-07-20 Richard Biener <rguenther@suse.de>
20237 PR tree-optimization/61171
20238 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
20239 (vect_analyze_stmt): Add slp instance parameter.
20240 (vectorizable_reduction): Likewise.
20241 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
20242 (vect_is_simple_reduction): Deal with chains not detected
20243 as SLP reduction chain, specifically not properly associated
20244 chains containing a mix of plus/minus.
20245 (get_reduction_op): Remove.
20246 (get_initial_defs_for_reduction): Simplify, pass in whether
20247 this is a reduction chain, pass in the SLP node for the PHIs.
20248 (vect_create_epilog_for_reduction): Get the SLP instance as
20250 (vectorizable_reduction): Get the SLP instance as arg.
20251 During analysis remember the SLP node with the PHIs in the
20252 instance. Simplify getting at the vectorized reduction PHIs.
20253 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
20254 through SLP instance.
20255 (vect_slp_analyze_operations): Likewise.
20256 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
20257 (vect_transform_stmt): Likewise.
20259 2017-07-20 Tom de Vries <tom@codesourcery.com>
20261 PR tree-optimization/81489
20262 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
20263 read of phi arg location to before loop that modifies phi.
20265 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20267 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
20270 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20272 PR middle-end/81331
20273 * except.c (execute): Fix ordering issue.
20275 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20277 PR rtl-optimization/81423
20278 * combine.c (make_compound_operation_int): Don't try to optimize
20279 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
20281 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20283 PR rtl-optimization/81423
20284 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
20285 with a constant that is -1 in the truncated to mode.
20287 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20289 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
20290 (determine_unlikely_bbs): ... here.
20291 * predict.h (propagate_unlikely_bbs_forward): Declare.
20292 * cfgexpand.c (pass_expand::execute): Use it.
20293 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
20295 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
20296 propagate_unlikely_bbs_forward.
20298 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20300 PR middle-end/81331
20301 * except.c (maybe_add_nop_after_section_switch): New function.
20304 2017-07-19 Tom de Vries <tom@codesourcery.com>
20306 * gimple.h (gimple_phi_set_arg): Make assert more strict.
20308 2017-07-19 Tom de Vries <tom@codesourcery.com>
20310 * gimple.h (gimple_phi_arg): Make assert more strict.
20312 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
20314 * config.gcc (powerpc*-*-*): Add mmintrin.h.
20315 * config/rs6000/mmintrin.h: New file.
20316 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
20318 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20320 PR tree-optimization/81346
20321 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
20323 2017-07-19 Tom de Vries <tom@codesourcery.com>
20325 * config/nvptx/nvptx.md (VECIM): Add V2DI.
20327 2017-07-19 Tom de Vries <tom@codesourcery.com>
20329 * config/nvptx/nvptx-modes.def: Add V2DImode.
20330 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
20331 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
20332 (nvptx_output_mov_insn): Handle lack of mov.b128.
20333 (nvptx_print_operand): Handle 'H' and 'L' codes.
20334 (nvptx_vector_mode_supported): Allow V2DImode.
20335 (nvptx_preferred_simd_mode): New function.
20336 (nvptx_data_alignment): New function.
20337 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
20338 nvptx_preferred_simd_mode.
20339 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
20341 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
20343 2017-07-19 Tom de Vries <tom@codesourcery.com>
20345 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
20346 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
20347 (nvptx_vector_mode_supported): New function. Allow V2SImode.
20348 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
20349 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
20350 (mov<VECIM>_insn): New define_insn.
20351 (define_expand "mov<VECIM>): New define_expand.
20353 2017-07-19 Tom de Vries <tom@codesourcery.com>
20355 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
20357 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20359 PR tree-optimization/81346
20360 * fold-const.h (fold_div_compare, range_check_type): Declare.
20361 * fold-const.c (range_check_type): New function.
20362 (build_range_check): Use range_check_type.
20363 (fold_div_compare): No longer static, rewritten into
20364 a match.pd helper function.
20365 (fold_comparison): Don't call fold_div_compare here.
20366 * match.pd (X / C1 op C2): New optimization using fold_div_compare
20367 as helper function.
20369 2017-07-19 Nathan Sidwell <nathan@acm.org>
20371 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
20372 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
20373 * tree.c (find_decls_types_r, verify_type): Use
20374 TYPE_{MIN,MAX}_VALUE_RAW.
20375 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
20376 (hash_tree): Likewise.
20377 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20379 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20382 2017-07-18 Tom de Vries <tom@codesourcery.com>
20384 PR middle-end/81464
20385 * omp-expand.c (expand_omp_for_static_chunk): Handle
20386 equal-argument loop exit phi.
20388 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
20391 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
20392 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
20393 operand 2 predicate.
20394 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
20395 operand 2 predicate.
20396 (ror,rol -> rorx splitters): Use const_int_operand as
20397 operand 2 predicate.
20399 2017-06-18 Richard Biener <rguenther@suse.de>
20401 PR tree-optimization/81410
20402 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
20403 the gap in the ! slp_perm SLP case after each group.
20405 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20407 PR middle-end/81463
20408 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
20411 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20413 PR middle-end/81462
20414 * predict.c (set_even_probabilities): Cleanup; do not affect
20415 probabilities that are already known.
20416 (combine_predictions_for_bb): Call even when count is set.
20418 2017-07-18 Nathan Sidwell <nathan@acm.org>
20420 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
20423 2017-07-18 Bin Cheng <bin.cheng@arm.com>
20426 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
20427 optimization for loop niter analysis.
20429 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
20432 * config/avr/avr.c (avr_optimize_casesi): Don't use
20433 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
20435 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
20437 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
20438 body_cost_vec from _vect_peel_extended_info.
20439 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
20440 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
20443 2017-07-18 Bin Cheng <bin.cheng@arm.com>
20445 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
20447 2017-07-18 Richard Biener <rguenther@suse.de>
20449 PR tree-optimization/80620
20450 PR tree-optimization/81403
20451 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
20452 info when re-using a VN table entry.
20454 2017-07-18 Richard Biener <rguenther@suse.de>
20456 PR tree-optimization/81418
20457 * tree-vect-loop.c (vectorizable_reduction): Properly compute
20458 vectype_in. Verify that with lane-reducing reduction operations
20459 we have a single def-use cycle.
20461 2017-07-17 Carl Love <cel@us.ibm.com>
20463 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
20465 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20466 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20467 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20468 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20469 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20470 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20471 VMULOSW): New enum "unspec" values.
20472 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20473 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20474 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20475 altivec_vmulosw): New patterns.
20476 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20477 VMULOSW): Add definitions.
20479 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
20481 * config/alpha/alpha.c: Include predict.h.
20483 2017-07-17 Yury Gribov <tetra2005@gmail.com>
20485 * tree-vrp.c (compare_assert_loc): Fix comparison function
20486 to return predictable results.
20488 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20490 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
20492 (subdi3): Likewise.
20493 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
20494 * doc/invoke.texi (mexpand-adddi): Update text.
20496 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20498 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
20499 that also clobbers the CC register. The old expand code is moved
20501 (*arc_clzsi2): ... here.
20502 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
20503 the CC register. The old expand code is moved to ...
20504 (arc_ctzsi2): ... here.
20506 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20508 * config/arc/arc.opt (mindexed-loads): Use initial value
20509 TARGET_INDEXED_LOADS_DEFAULT.
20510 (mauto-modify-reg): Use initial value
20511 TARGET_AUTO_MODIFY_REG_DEFAULT.
20512 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
20513 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
20514 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
20515 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
20517 2017-07-17 Martin Liska <mliska@suse.cz>
20520 * opts.c (finish_options): Do not allow -fgnu-tm
20521 w/ -fsanitize={kernel-,}address. Say sorry.
20523 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20526 * tree-loop-distribution.c (classify_partition): Only assert on
20527 numer of iterations.
20528 (merge_dep_scc_partitions): Delete prameter. Update function call.
20529 (distribute_loop): Remove code handling loop with unknown niters.
20530 (pass_loop_distribution::execute): Skip loop with unknown niters.
20532 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20535 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
20536 function sort_partitions_by_post_order.
20538 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20540 PR tree-optimization/81374
20541 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
20542 the max index of basic blocks, rather than number of basic blocks.
20544 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20546 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
20548 (arc_legitimate_pic_operand_p): Likewise.
20549 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
20551 (arc_needs_pcl_p): Likewise.
20552 (arc_legitimate_pc_offset_p): Likewise.
20553 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
20554 function is also used in constrains.md.
20555 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
20556 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
20557 PLUS. Only return true/false in known cases, otherwise assert.
20558 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
20559 is already called in arc_legitimate_constant_p.
20560 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
20562 (LEGITIMATE_PIC_OPERAND_P): Use
20563 arc_raw_symbolic_reference_mentioned_p function.
20564 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
20569 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20570 Andrew Burgess <andrew.burgess@embecosm.com>
20572 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
20573 (arc_return_address_register): New function.
20574 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
20575 (arc_handle_fndecl_attribute): Add naked attribute.
20576 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
20577 (TARGET_WARN_FUNC_RETURN): Likewise.
20578 (arc_allocate_stack_slots_for_args): New function.
20579 (arc_warn_func_return): Likewise.
20580 (machine_function): Change type fn_type.
20581 (arc_compute_function_type): Consider new naked function type,
20582 change function return type.
20583 (arc_must_save_register): Adapt to handle new
20584 arc_compute_function_type's return type.
20585 (arc_expand_prologue): Likewise.
20586 (arc_expand_epilogue): Likewise.
20587 (arc_return_address_regs): Delete.
20588 (arc_return_address_register): New function.
20589 (arc_epilogue_uses): Use above function.
20590 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
20591 (arc_function_type): Change encoding, add naked type.
20592 (ARC_INTERRUPT_P): Change to handle the new encoding.
20593 (ARC_FAST_INTERRUPT_P): Likewise.
20594 (ARC_NORMAL_P): Define.
20595 (ARC_NAKED_P): Likewise.
20596 (arc_compute_function_type): Delete prototype.
20597 * config/arc/arc.md (in_ret_delay_slot): Use
20598 arc_return_address_register function.
20599 (simple_return): Likewise.
20600 (p_return_i): Likewise.
20602 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20604 PR tree-optimization/81428
20605 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
20606 can't be built for those types.
20608 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
20610 Remove stuff dead since r239246.
20612 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
20613 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
20614 (avr_inform_devices): Remove dead stuff.
20616 2017-07-17 Tamar Christina <tamar.christina@arm.com>
20618 * config/arm/arm_neon.h: Fix softp typo.
20620 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20622 PR tree-optimization/81365
20623 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
20624 aggregate moves onto bb predecessor edges, make sure there are no
20625 loads that could alias the lhs in between the start of bb and the
20628 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
20631 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
20632 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
20633 [LSHIFTRT, outer_code = TRUNCATE]: Same.
20635 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20637 PR tree-optimization/81396
20638 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
20639 (init_symbolic_number): Initialize it to 1.
20640 (perform_symbolic_merge): Add n_ops from both operands into the new
20642 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
20643 without base_addr as useless if they need more than one operation.
20644 (bswap_replace): Handle !bswap case for NULL base_addr.
20646 2017-07-17 Tom de Vries <tom@codesourcery.com>
20649 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
20652 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
20654 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
20655 conditional builtin define __FIX_LEON3FT_B2BST.
20657 2017-07-17 Daniel Cederman <cederman@gaisler.com>
20659 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
20660 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
20663 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20665 PR rtl-optimization/81424
20666 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
20667 to remove potential trapping from operands if -fnon-call-exceptions.
20669 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20671 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
20672 profile_proability for scalling.
20673 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
20675 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20677 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
20679 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20681 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
20682 fixpoint arithmetics.
20684 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20686 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
20687 fixpoint arithmetics.
20689 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20691 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
20692 fixpoint arithmetics.
20694 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20696 * profile-count.h (profile_probability::from_reg_br_prob_note,
20697 profile_probability::to_reg_br_prob_note): New functions.
20698 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
20699 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
20700 * predict.c (probability_reliable_p): Update.
20701 (edge_probability_reliable_p): Update.
20702 (br_prob_note_reliable_p): Update.
20703 (invert_br_probabilities): Update.
20704 (add_reg_br_prob_note): New function.
20705 (combine_predictions_for_insn): Update.
20706 * asan.c (asan_clear_shadow): Update.
20707 * cfgbuild.c (compute_outgoing_frequencies): Update.
20708 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
20709 (update_br_prob_note): Update.
20710 (rtl_verify_edges): Update.
20711 (purge_dead_edges): Update.
20712 (fixup_reorder_chain): Update.
20713 * emit-rtl.c (try_split): Update.
20714 * ifcvt.c (cond_exec_process_insns): Update.
20715 (cond_exec_process_if_block): Update.
20716 (dead_or_predicable): Update.
20717 * internal-fn.c (expand_addsub_overflow): Update.
20718 (expand_neg_overflow): Update.
20719 (expand_mul_overflow): Update.
20720 * loop-doloop.c (doloop_modify): Update.
20721 * loop-unroll.c (compare_and_jump_seq): Update.
20722 * optabs.c (emit_cmp_and_jump_insn_1): Update.
20723 * predict.h: Update.
20724 * reorg.c (mostly_true_jump): Update.
20726 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
20727 * config/alpha/alpha.c (emit_unlikely_jump): Update.
20728 * config/arc/arc.c: (emit_unlikely_jump): Update.
20729 * config/arm/arm.c: (emit_unlikely_jump): Update.
20730 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
20731 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
20732 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
20733 (ix86_print_operand): Update.
20734 (ix86_split_fp_branch): Update.
20735 (predict_jump): Update.
20736 * config/ia64/ia64.c (ia64_print_operand): Update.
20737 * config/mmix/mmix.c (mmix_print_operand): Update.
20738 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
20739 (rs6000_expand_split_stack_prologue): Update.
20740 * config/rs6000/rs6000.c: Update.
20741 * config/s390/s390.c (s390_expand_vec_strlen): Update.
20742 (s390_expand_vec_movstr): Update.
20743 (s390_expand_cs_tdsi): Update.
20744 (s390_expand_split_stack_prologue): Update.
20745 * config/sh/sh.c (sh_print_operand): Update.
20746 (expand_cbranchsi4): Update.
20747 (expand_cbranchdi4): Update.
20748 * config/sparc/sparc.c (output_v9branch): Update.
20749 * config/spu/spu.c (get_branch_target): Update.
20750 (ea_load_store_inline): Update.
20751 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
20752 * config/tilepro/tilepro.c: Update.
20754 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20756 * gimplify.c (mostly_copy_tree_r): Revert latest change.
20757 (gimplify_save_expr): Likewise.
20759 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20761 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
20763 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20765 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
20767 * timevar.def (TV_IPA_FNSUMMARY): Define.
20769 2017-07-16 Daniel Cederman <cederman@gaisler.com>
20771 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
20772 to back store errata sensitive sequence from being generated.
20773 (sqrtdf2_fix): Likewise.
20775 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20777 * tree-ssa-threadupdate.c (compute_path_counts,
20778 update_joiner_offpath_counts): Use profile_probability.
20780 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
20783 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20785 * config/arm/arm-c.c (arm_cpu_builtins): Define
20786 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20788 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
20790 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20791 array entries to represent __ieee128 versions of the
20792 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
20793 scalar_extract_sig, and scalar_insert_exp built-in functions.
20794 (altivec_resolve_overloaded_builtin): Add special case handling
20795 for the __builtin_scalar_insert_exp function, as represented by
20796 the P9V_BUILTIN_VEC_VSIEDP constant.
20797 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
20798 exponent support for __ieee128 argument.
20799 (VSESQP): Add scalar extract signature support for __ieee128
20801 (VSTDCNQP): Add scalar test negative support for __ieee128
20803 (VSIEQP): Add scalar insert exponent support for __int128 argument
20804 with __ieee128 result.
20805 (VSIEQPF): Add scalar insert exponent support for __ieee128
20806 argument with __ieee128 result.
20807 (VSTDCQP): Add scalar test data class support for __ieee128
20809 (VSTDCNQP): Add overload support for scalar test negative with
20810 __ieee128 argument.
20811 (VSTDCQP): Add overload support for scalar test data class
20812 __ieee128 argument.
20813 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
20814 UNSPEC_VSX_SXSIGDP.
20815 (UNSPEC_VSX_SIEXPQP): New constant.
20816 (xsxexpqp): New insn for VSX scalar extract exponent quad
20818 (xsxsigqp): New insn for VSX scalar extract significand quad
20820 (xsiexpqpf): New insn for VSX scalar insert exponent quad
20821 precision with floating point argument.
20822 (xststdcqp): New expand for VSX scalar test data class quad
20824 (xststdcnegqp): New expand for VSX scalar test negative quad
20826 (xststdcqp): New insn to match expansions for VSX scalar test data
20827 class quad precision and VSX scalar test negative quad precision.
20828 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
20829 special case operand checking to enforce that second operand of
20830 VSX scalar test data class with quad precision argument is a 7-bit
20832 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20833 prototypes and descriptions of __ieee128 versions of
20834 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
20835 scalar_test_data_class, and scalar_test_neg built-in functions.
20837 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20839 PR tree-optimization/81162
20840 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
20841 replace a negate with an add.
20843 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
20845 * doc/invoke.texi (arm/-mcpu): Document +crypto.
20847 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20849 * config/arm/arm-c.c (arm_cpu_builtins): Define
20850 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20852 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20854 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
20855 (armv8-r): Set ARM Cortex-R52 as default CPU.
20856 * config/arm/arm-tables.opt: Regenerate.
20857 * config/arm/arm-tune.md: Regenerate.
20858 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
20860 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
20861 extension for -mcpu=cortex-r52.
20863 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20865 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
20866 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
20867 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
20868 (fp-armv8): Define it as FP_ARMv8 only.
20869 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
20870 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
20872 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
20873 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
20874 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
20875 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
20876 than TARGET_FPU_ARMV8.
20877 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
20878 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
20879 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
20881 * config/arm/neon.md (neon_vrint): Likewise.
20882 (neon_vcvt): Likewise.
20883 (neon_<fmaxmin_op><mode>): Likewise.
20884 (<fmaxmin><mode>3): Likewise.
20885 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
20886 * config/arm/predicates.md (arm_cond_move_operator): Check against
20887 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
20889 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
20891 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
20892 to top of function.
20894 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20896 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
20897 loop in comment with memset.
20899 2017-07-14 Martin Liska <mliska@suse.cz>
20901 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
20902 * dwarf2out.c (is_java): Remove the function.
20903 (output_pubname): Remove usage of the function.
20904 (lower_bound_default): Remove usage of DW_LANG_Java.
20905 (gen_compile_unit_die): Likewise.
20906 * gcc.c: Remove compiler defaults for .java and .zip files.
20907 * gimple-expr.c (remove_suffix): Change as there's no longer
20908 extension than 4-letter one.
20909 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
20910 (gimplify_save_expr): Likewise.
20911 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
20912 as it's possible even for other languages than Java.
20913 * langhooks.h (struct lang_hooks): Remove Java from a comment.
20914 * lto-opts.c (lto_write_options): Remove reference to Java.
20915 * opts.c (strip_off_ending): Update file extension handling.
20916 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
20917 * tree-eh.c (lower_resx): Likewise.
20918 * tree.c (free_lang_data_in_type): Remove dead code.
20919 (find_decls_types_r): Likewise.
20920 (build_common_builtin_nodes): Remove Java from a comment.
20921 (verify_type): Remove dead code.
20922 * varasm.c (assemble_external): Remove Java from a comment.
20924 2017-07-14 Martin Liska <mliska@suse.cz>
20926 * opts.c (finish_options): Add quotes.
20927 (common_handle_option): Likewise.
20929 2017-07-14 Martin Liska <mliska@suse.cz>
20931 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
20932 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
20933 Remove N_SO_PASCAL.
20934 * dwarf2out.c (lower_bound_default): Do not handle
20936 (gen_compile_unit_die): Likewise.
20937 * gcc.c: Remove default extension binding for GNU Pascal.
20938 * stmt.c: Remove Pascal language from a comment.
20939 * xcoffout.c: Likewise.
20941 2017-07-13 David Malcolm <dmalcolm@redhat.com>
20944 * diagnostic-show-locus.c (fixit_cmp): New function.
20945 (layout::layout): Sort m_fixit_hints.
20946 (column_range::column_range): Assert that the values are valid.
20947 (struct char_span): New struct.
20948 (correction::overwrite): New method.
20949 (struct source_line): New struct.
20950 (line_corrections::add_hint): Add assertions. Reimplement memcpy
20951 calls in terms of classes source_line and char_span, and
20952 correction::overwrite.
20953 (selftest::test_overlapped_fixit_printing_2): New function.
20954 (selftest::diagnostic_show_locus_c_tests): Call it.
20956 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
20958 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
20959 early if there is no lhs.
20961 2017-07-13 Martin Liska <mliska@suse.cz>
20963 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
20964 (gen_reference_type_die): Likewise.
20965 * stor-layout.c: Remove Pascal-related comment.
20967 2017-07-13 Martin Liska <mliska@suse.cz>
20969 * opts.c (finish_options): Add quotes to error messages.
20970 (parse_sanitizer_options): Likewise.
20972 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
20974 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
20976 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
20978 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
20980 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
20982 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
20984 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
20986 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
20989 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
20990 provides the hardware capability bits, define the macro
20991 __BUILTIN_CPU_SUPPORTS__.
20992 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
20993 if GLIBC does not provide the hardware capability bits. Add a
20994 gcc_unreachable call if the built-in cpu function is neither
20995 __builtin_cpu_is nor __builtin_cpu_supports.
20996 (rs6000_get_function_versions_dispatcher): Change the warning
20997 that an old GLIBC is used which does not export the capability
20998 bits to be an error.
20999 * doc/extend.texi (target_clones attribute): Document the
21000 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
21001 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
21002 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
21003 the macros defined by GCC if the newer GLIBC is available.
21005 2017-07-12 Jeff Law <law@redhat.com>
21007 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
21008 remaining includes slightly.
21009 * config/riscv/riscv-builtins.c: Include profile-count.h.
21011 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21014 * config/avr/avr.c (avr_set_current_function): In diagnostic
21015 messages: Quote keywords and (parts of) identifiers.
21016 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
21019 2017-07-12 Carl Love <cel@us.ibm.com>
21021 * config/rs6000/rs6000-c.c: Add support for built-in functions
21022 vector bool char vec_revb (vector bool char);
21023 vector bool short vec_revb (vector short char);
21024 vector bool int vec_revb (vector bool int);
21025 vector bool long long vec_revb (vector bool long long);
21026 * doc/extend.texi: Update the built-in documentation file for the
21027 new built-in functions.
21029 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21031 * config/s390/s390.md: Remove movcc splitter.
21033 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21035 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
21036 load/store on condition.
21038 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21041 * config/avr/avr.c (avr_encode_section_info)
21042 [progmem && !TREE_READONLY]: Error if progmem object needs
21045 2017-07-11 Michael Collison <michael.collison@arm.com>
21047 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
21050 2017-07-11 Carl Love <cel@us.ibm.com>
21052 * config/rs6000/rs6000-c.c: Add support for builtins
21053 vector unsigned int vec_parity_lsbb (vector signed int);
21054 vector unsigned int vec_parity_lsbb (vector unsigned int);
21055 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
21056 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
21057 vector unsigned long long vec_parity_lsbb (vector signed long long);
21058 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
21059 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
21060 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
21061 * doc/extend.texi: Update the built-in documentation file for the
21062 new built-in functions.
21064 2017-07-11 David Malcolm <dmalcolm@redhat.com>
21066 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
21067 (layout::m_primary_loc): New field.
21068 (layout::layout): Initialize new field. Move location filtering
21069 logic from here to...
21070 (layout::maybe_add_location_range): ...this new method. Add
21071 support for filtering to just the lines already specified by other
21073 (layout::will_show_line_p): New method.
21074 (gcc_rich_location::add_location_if_nearby): New method.
21075 (selftest::test_add_location_if_nearby): New test function.
21076 (selftest::diagnostic_show_locus_c_tests): Call it.
21077 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
21080 2017-07-11 Tom de Vries <tom@codesourcery.com>
21082 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
21083 (bb_first_real_insn): New function.
21084 (nvptx_single): Add extra initialization of broadcasted condition
21087 2017-07-11 Nathan Sidwell <nathan@acm.org>
21089 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
21091 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
21093 * doc/extend.texi (AVR Function Attributes): Remove weblink to
21094 Binutils doc as TEXI will mess them up.
21095 * doc/invoke.texi (AVR Options): Same here.
21097 2017-07-11 Daniel Cederman <cederman@gaisler.com>
21099 * config/sparc/sparc.opt (mfix-ut700): New option.
21100 (mfix-gr712rc): Likewise.
21101 (sparc_fix_b2bst): New variable.
21102 * doc/invoke.texi (SPARC options): Document them.
21103 (ARM options): Fix warnings.
21104 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
21105 instructions to prevent sequences that can trigger the store-store
21106 errata for certain LEON3FT processors.
21107 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
21108 (sparc_option_override): Set sparc_fix_b2bst appropriately.
21109 * config/sparc/sparc.md (fix_b2bst): New attribute.
21110 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
21112 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
21115 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
21117 (*rsqrtsf2_sse): Ditto.
21119 (div<mode>3): Macroize insn from divdf3 and divsf3
21120 using MODEF mode iterator.
21122 2017-07-10 Martin Sebor <msebor@redhat.com>
21124 PR tree-optimization/80397
21125 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
21126 instead of testing for equality to INTEGER_TYPE.
21128 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
21130 * config.gcc: Remove uclibc from arc target spec.
21132 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
21134 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
21136 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21139 * lto-wrapper.c (remove_option): New function.
21140 (merge_and_complain): Merge PIC/PIE options more realistically.
21142 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21144 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
21148 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
21149 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
21150 * config.in: Regenerate.
21151 * configure: Regenerate.
21152 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
21153 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
21154 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
21155 (TARGET_GASISR_PROLOGUES): ...target mask.
21156 * common/config/avr/avr-common.c
21157 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
21158 Set -mgas-isr-prologues.
21159 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
21160 INSERT_PASS_BEFORE for it.
21161 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
21162 * config/avr/avr.c (avr_option_override)
21163 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
21164 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
21165 (avr_attribute_table) <no_gccisr>: Add new function attribute.
21166 (avr_set_current_function) <is_no_gccisr>: Init machine field.
21167 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
21169 (make_avr_pass_pre_proep): New function.
21170 (emit_push_sfr) <treg>: Add argument to function and use it
21171 instead of TMP_REG.
21172 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
21173 and set machine->gasisr.yes.
21174 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
21175 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
21176 __gcc_isr.n_pushed to .L__stack_usage.
21177 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
21178 (avr_asm_final_postscan_insn): ...this new static function.
21179 * config/avr/avr.h (machine_function)
21180 <is_no_gccisr, use_L__stack_usage>: New fields.
21181 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
21182 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
21183 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
21184 (gasisr, *gasisr): New expander and insn.
21185 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21186 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
21187 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
21189 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
21191 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
21194 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21196 Move jump-tables out of .text again.
21199 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
21200 (ASM_OUTPUT_ADDR_VEC): New function.
21201 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
21202 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
21203 INSN_ADDRESSes as asm comment.
21204 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
21205 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
21206 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
21207 * config/avr/avr.md (*tablejump): Adjust comment.
21208 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
21209 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
21211 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
21212 (avr_output_addr_vec): New proto.
21213 (avr_log_t) <insn_addresses>: New field.
21215 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
21218 * config/i386/i386.c (ix86_function_arg_advance): Set
21219 outgoing_args_on_stack to true if there are outgoing arguments
21221 (ix86_function_arg): Likewise.
21222 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
21223 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
21224 * config/i386/i386.h (machine_function): Add
21225 outgoing_args_on_stack.
21227 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
21229 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
21230 supporting pthreds.
21231 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
21233 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
21235 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
21236 (REAL_H): Remove $(MACHMODE_H).
21237 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
21239 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
21240 $(MACHMODE_H) and double-int.h.
21241 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
21243 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
21246 2017-07-07 Andrew Pinski <apinski@cavium.com>
21248 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
21249 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
21251 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21253 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
21254 Add warning if GCC was not configured to link against a GLIBC that
21255 exports the hardware capability bits.
21256 (make_resolver_func): Make resolver function private and not a
21257 COMDAT function. Create the name with clone_function_name instead
21258 of make_unique_name.
21261 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
21262 correct operand in doing the split.
21264 2017-07-07 Carl Love <cel@us.ibm.com>
21266 * config/rs6000/rs6000-c: Add support for built-in function
21267 vector unsigned short vec_pack_to_short_fp32 (vector float,
21269 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
21270 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
21271 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
21272 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
21273 (convert_4f32_8i16): Add define_expand.
21274 * doc/extend.texi: Update the built-in documentation file for the
21275 new built-in function.
21277 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21279 * config/sparc/m8.md: New file.
21280 * config/sparc/sparc.md: Include m8.md.
21282 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21284 * config/sparc/sparc.opt: New option -mvis4b.
21285 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
21286 (sparc_option_override): Handle VIS4B.
21287 (enum sparc_builtins): Define
21288 SPARC_BUILTIN_DICTUNPACK{8,16,32},
21289 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
21290 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
21291 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
21292 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
21293 (check_constant_argument): New function.
21294 (sparc_vis_init_builtins): Define builtins
21295 __builtin_vis_dictunpack{8,16,32},
21296 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
21297 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
21298 __builtin_vis_fpcmpde{8,16,32}shl and
21299 __builtin_vis_fpcmpur{8,16,32}shl.
21300 (sparc_expand_builtin): Check that the constant operands to
21301 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
21302 constant and in range.
21303 * config/sparc/sparc-c.c (sparc_target_macros): Handle
21305 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
21306 (SPARC_IMM5_P): Likewise.
21307 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
21308 (enabled): Handle vis4b.
21309 (UNSPEC_DICTUNPACK): New unspec.
21310 (UNSPEC_FPCMPSHL): Likewise.
21311 (UNSPEC_FPUCMPSHL): Likewise.
21312 (UNSPEC_FPCMPDESHL): Likewise.
21313 (UNSPEC_FPCMPURSHL): Likewise.
21314 (cpu_feature): New CPU feature `vis4b'.
21315 (dictunpack{8,16,32}): New insns.
21316 (FPCSMODE): New mode iterator.
21317 (fpcscond): New code iterator.
21318 (fpcsucond): Likewise.
21319 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
21320 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
21321 (fpcmpde{8,16,32}{si,di}shl): Likewise.
21322 (fpcmpur{8,16,32}{si,di}shl): Likewise.
21323 * config/sparc/constraints.md: Define constraints `q' for unsigned
21324 2-bit integer constants and `t' for unsigned 5-bit integer
21326 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
21328 (imm5_operand_dictunpack16): Likewise.
21329 (imm5_operand_dictunpack32): Likewise.
21330 (imm2_operand): Likewise.
21331 * doc/invoke.texi (SPARC Options): Document -mvis4b.
21332 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
21333 ditunpack* and fpcmp*shl builtins.
21335 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21337 * config.gcc: Handle m8 in --with-{cpu,tune} options.
21338 * config.in: Add HAVE_AS_SPARC6 define.
21339 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
21341 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
21343 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
21344 (CPP_CPU_SPEC): Handle m8.
21345 (ASM_CPU_SPEC): Likewise.
21346 * config/sparc/sparc-opts.h (enum processor_type): Add
21348 * config/sparc/sparc.c (m8_costs): New struct.
21349 (sparc_option_override): Handle TARGET_CPU_m8.
21350 (sparc32_initialize_trampoline): Likewise.
21351 (sparc64_initialize_trampoline): Likewise.
21352 (sparc_issue_rate): Likewise.
21353 (sparc_register_move_cost): Likewise.
21354 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
21355 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
21356 (ASM_CPU64_DEFAULT_SPEC): Likewise.
21357 (CPP_CPU_SPEC): Handle M8.
21358 (ASM_CPU_SPEC): Likewise.
21359 (AS_M8_FLAG): Define.
21360 * config/sparc/sparc.md: Add m8 to the cpu attribute.
21361 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
21362 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
21364 * configure: Regenerate.
21365 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
21368 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21370 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
21372 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
21373 ("*movdi_insn_sp32"): Do not set v3pipe.
21374 ("*movsi_insn"): Likewise.
21375 ("*movdi_insn_sp64"): Likewise.
21376 ("*movsf_insn"): Likewise.
21377 ("*movdf_insn_sp32"): Likewise.
21378 ("*movdf_insn_sp64"): Likewise.
21379 ("*zero_extendsidi2_insn_sp64"): Likewise.
21380 ("*sign_extendsidi2_insn"): Likewise.
21381 ("*mov<VM32:mode>_insn"): Likewise.
21382 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21383 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21384 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21385 ("<vlop:code><VL:mode>3"): Likewise.
21386 ("*not_<vlop:code><VL:mode>3"): Likewise.
21387 ("*nand<VL:mode>_vis"): Likewise.
21388 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
21389 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
21390 ("one_cmpl<VL:mode>2"): Likewise.
21391 ("faligndata<VM64:mode>_vis"): Likewise.
21392 ("alignaddrsi_vis"): Likewise.
21393 ("alignaddrdi_vis"): Likweise.
21394 ("alignaddrlsi_vis"): Likewise.
21395 ("alignaddrldi_vis"): Likewise.
21396 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21397 ("bmaskdi_vis"): Likewise.
21398 ("bmasksi_vis"): Likewise.
21399 ("bshuffle<VM64:mode>_vis"): Likewise.
21400 ("cmask8<P:mode>_vis"): Likewise.
21401 ("cmask16<P:mode>_vis"): Likewise.
21402 ("cmask32<P:mode>_vis"): Likewise.
21403 ("pdistn<P:mode>_vis"): Likewise.
21404 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21406 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21408 * config/sparc/sparc.md ("subtype"): New insn attribute.
21409 ("*wrgsr_sp64"): Set insn subtype.
21410 ("*rdgsr_sp64"): Likewise.
21411 ("alignaddrsi_vis"): Likewise.
21412 ("alignaddrdi_vis"): Likewise.
21413 ("alignaddrlsi_vis"): Likewise.
21414 ("alignaddrldi_vis"): Likewise.
21415 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21416 ("fexpand_vis"): Likewise.
21417 ("fpmerge_vis"): Likewise.
21418 ("faligndata<VM64:mode>_vis"): Likewise.
21419 ("bshuffle<VM64:mode>_vis"): Likewise.
21420 ("cmask8<P:mode>_vis"): Likewise.
21421 ("cmask16<P:mode>_vis"): Likewise.
21422 ("cmask32<P:mode>_vis"): Likewise.
21423 ("fchksm16_vis"): Likewise.
21424 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
21425 ("fmean16_vis"): Likewise.
21426 ("fp<plusminus_insn>64_vis"): Likewise.
21427 ("<plusminus_insn>v8qi3"): Likewise.
21428 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21429 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
21430 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
21431 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
21432 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
21433 ("*movqi_insn"): Likewise.
21434 ("*movhi_insn"): Likewise.
21435 ("*movsi_insn"): Likewise.
21436 ("movsi_pic_gotdata_op"): Likewise.
21437 ("*movdi_insn_sp32"): Likewise.
21438 ("*movdi_insn_sp64"): Likewise.
21439 ("movdi_pic_gotdata_op"): Likewise.
21440 ("*movsf_insn"): Likewise.
21441 ("*movdf_insn_sp32"): Likewise.
21442 ("*movdf_insn_sp64"): Likewise.
21443 ("*zero_extendhisi2_insn"): Likewise.
21444 ("*zero_extendqihi2_insn"): Likewise.
21445 ("*zero_extendqisi2_insn"): Likewise.
21446 ("*zero_extendqidi2_insn"): Likewise.
21447 ("*zero_extendhidi2_insn"): Likewise.
21448 ("*zero_extendsidi2_insn_sp64"): Likewise.
21449 ("ldfsr"): Likewise.
21450 ("prefetch_64"): Likewise.
21451 ("prefetch_32"): Likewise.
21452 ("tie_ld32"): Likewise.
21453 ("tie_ld64"): Likewise.
21454 ("*tldo_ldub_sp32"): Likewise.
21455 ("*tldo_ldub1_sp32"): Likewise.
21456 ("*tldo_ldub2_sp32"): Likewise.
21457 ("*tldo_ldub_sp64"): Likewise.
21458 ("*tldo_ldub1_sp64"): Likewise.
21459 ("*tldo_ldub2_sp64"): Likewise.
21460 ("*tldo_ldub3_sp64"): Likewise.
21461 ("*tldo_lduh_sp32"): Likewise.
21462 ("*tldo_lduh1_sp32"): Likewise.
21463 ("*tldo_lduh_sp64"): Likewise.
21464 ("*tldo_lduh1_sp64"): Likewise.
21465 ("*tldo_lduh2_sp64"): Likewise.
21466 ("*tldo_lduw_sp32"): Likewise.
21467 ("*tldo_lduw_sp64"): Likewise.
21468 ("*tldo_lduw1_sp64"): Likewise.
21469 ("*tldo_ldx_sp64"): Likewise.
21470 ("*mov<VM32:mode>_insn"): Likewise.
21471 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21472 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21474 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21476 * config/sparc/sparc.md ("type"): New insn type viscmp.
21477 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
21479 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
21480 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
21481 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21482 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
21484 ("n7_vis_logical_11cycle"): Likewise.
21485 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
21486 * config/sparc/niagara2.md ("niag3_vis": Likewise.
21487 * config/sparc/niagara.md ("niag_vis"): Likewise.
21488 * config/sparc/ultra3.md ("us3_fga"): Likewise.
21489 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
21491 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21493 * config/sparc/sparc.md: New instruction type `bmask'.
21494 (bmaskdi_vis): Use the `bmask' type.
21495 (bmasksi_vis): Likewise.
21496 * config/sparc/ultra3.md (us3_array): Likewise.
21497 * config/sparc/niagara7.md (n7_array): Likewise.
21498 * config/sparc/niagara4.md (n4_array): Likewise.
21499 * config/sparc/niagara2.md (niag2_vis): Likewise.
21500 (niag3_vis): Likewise.
21501 * config/sparc/niagara.md (niag_vis): Likewise.
21503 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21505 * ipa-comdats.c: Remove optimize check from gate.
21506 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
21507 for functions not optimized.
21508 (ipa_fn_summary_read): Skip optimize check.
21509 (ipa_fn_summary_write): Likewise.
21510 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
21512 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
21514 (can_inline_edge_p): Check flag_pcc_struct_return for match.
21515 (check_callers): Give up on caller which is not optimized.
21516 (inline_small_functions): Likewise.
21517 (ipa_inline): Do not give up when not optimizing.
21518 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
21519 away unoptimizes cdtors.
21520 (whole_program_function_and_variable_visibility): Do
21521 ipa_discover_readonly_nonaddressable_vars in LTO mode.
21522 * ipa.c (process_references): Do not check optimize.
21523 (symbol_table::remove_unreachable_nodes): Update optimize check.
21524 (set_writeonly_bit): Update optimize check.
21525 (pass_ipa_cdtor_merge::gate): Do not check optimize.
21526 (pass_ipa_single_use::gate): Remove.
21528 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21530 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
21531 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
21532 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
21533 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
21534 permute_load, permute_store, adjust_extract, adjust_splat,
21535 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
21536 replace_swap_with_copy, dump_swap_insn_table,
21537 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
21538 recombine_lvx_pattern, recombine_stvx_pattern,
21539 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
21540 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
21541 to file rs6000-p8swap.c.
21542 * config/rs6000/rs6000-p8swap.c: New file.
21543 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
21544 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
21545 and rs6000*-*-* targets.
21547 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21549 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
21551 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21553 * lto-wrapper.c (merge_and_complain): Do not merge
21554 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
21555 fsigned_zeros, ftrapping_math, fwrapv.
21556 (append_compiler_options): Do not track these options.
21557 (append_linker_options): Likewie
21559 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21561 * cgraphunit.c (cgraph_node::finalize_function): When
21562 !flag_toplevel_reorde set no_reorder flag.
21563 (varpool_node::finalize_decl): Likewise.
21564 (symbol_table::compile): Drop no toplevel reorder path.
21566 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21568 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
21569 edges; zero probability is not better than uninitialized.
21571 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
21573 * asan.h (asan_sanitize_allocas_p): Declare.
21574 * asan.c (asan_sanitize_allocas_p): New function.
21575 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
21576 (handle_builtin_alloca): Likewise.
21577 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
21578 if !asan_sanitize_allocas_p.
21579 * params.def (asan-instrument-allocas): Add new option.
21580 * params.h (ASAN_PROTECT_ALLOCAS): Define.
21581 * opts.c (common_handle_option): Disable allocas sanitization for
21584 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
21586 * asan.c: Include gimple-fold.h.
21587 (get_last_alloca_addr): New function.
21588 (handle_builtin_stackrestore): Likewise.
21589 (handle_builtin_alloca): Likewise.
21590 (asan_emit_allocas_unpoison): Likewise.
21591 (get_mem_refs_of_builtin_call): Add new parameter, remove const
21592 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
21593 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
21594 (instrument_builtin_call): Pass gimple iterator to
21595 get_mem_refs_of_builtin_call.
21596 (last_alloca_addr): New global.
21597 * asan.h (asan_emit_allocas_unpoison): Declare.
21598 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
21599 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
21600 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
21601 if function calls alloca.
21602 * gimple-fold.c (replace_call_with_value): Remove static keyword.
21603 * gimple-fold.h (replace_call_with_value): Declare.
21604 * internal-fn.c: Include asan.h.
21605 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
21606 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
21608 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21610 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
21611 (C_SELFTEST_FLAGS): New.
21612 (CPP_SELFTEST_FLAGS): New.
21613 (SELFTEST_DEPS): New, from deps of s-selftest.
21614 (C_SELFTEST_DEPS): New, from deps of s-selftest.
21615 (CPP_SELFTEST_DEPS): New.
21616 (selftest): Add dependency on s-selftest-c++.
21617 (s-selftest): Rename to...
21618 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
21619 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
21620 than SELFTEST_FLAGS.
21621 (selftest-gdb): Rename to...
21622 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
21624 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
21625 (selftest-valgrind): Rename to...
21626 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
21628 (selftest-valgrind): Reintroduce as an alias for
21629 selftest-c-valgrind.
21630 (s-selftest-c++): New.
21631 (selftest-c++-gdb): New.
21632 (selftest-c++-valgrind): New.
21634 2017-07-06 Olivier Hainque <hainque@adacore.com>
21636 * gcc.c (process_command): When deciding if undefined variables
21637 should be ignored when processing specs, accept "gcc -v" as well.
21639 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21641 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
21642 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
21644 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21646 * config/arm/arm-cpus.in (armv8-r): Add new entry.
21647 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
21648 * config/arm/arm-tables.opt: Regenerate.
21649 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
21651 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
21653 2017-07-06 Carl Love <cel@us.ibm.com>
21655 * ChangeLog: Clean up from mid air collision
21657 2017-07-06 Carl Love <cel@us.ibm.com>
21659 * config/rs6000/rs6000-c.c: Add support for built-in functions
21660 vector signed int vec_subc (vector signed int, vector signed int);
21661 vector signed __int128 vec_subc (vector signed __int128,
21662 vector signed __int128);
21663 vector unsigned __int128 vec_subc (vector unsigned __int128,
21664 vector unsigned __int128);
21665 vector signed int vec_sube (vector signed int, vector signed int,
21666 vector signed int);
21667 vector unsigned int vec_sube (vector unsigned int,
21668 vector unsigned int,
21669 vector unsigned int);
21670 vector signed __int128 vec_sube (vector signed __int128,
21671 vector signed __int128,
21672 vector signed__int128);
21673 vector unsigned __int128 vec_sube (vector unsigned __int128,
21674 vector unsigned __int128,
21675 vector unsigned __int128);
21676 vector signed int vec_subec (vector signed int, vector signed int,
21677 vector signed int);
21678 vector unsigned int vec_subec (vector unsigned int,
21679 vector unsigned int,
21680 vector unsigned int);
21681 vector signed __int128 vec_subec (vector signed __int128,
21682 vector signed __int128,
21683 vector signed__int128);
21684 vector unsigned __int128 vec_subec (vector unsigned __int128,
21685 vector unsigned __int128,
21686 vector unsigned __int128);
21687 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
21688 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
21689 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
21690 BU_ALTIVEC_OVERLOAD_X definitions.
21691 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
21692 * doc/extend.texi: Update the built-in documentation file for the new
21693 built-in functions.
21695 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21698 * diagnostic-show-locus.c (layout::layout): Use start and finish
21699 spelling location for the start and finish of each range.
21700 * genmatch.c (linemap_client_expand_location_to_spelling_point):
21701 Add unused aspect param.
21702 * input.c (expand_location_1): Add "aspect" param, and use it
21703 to access the correct part of the location.
21704 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
21706 (expand_location_to_spelling_point): Likewise.
21707 (linemap_client_expand_location_to_spelling_point): Add "aspect"
21708 param, and pass it to expand_location_1.
21710 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
21712 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
21713 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
21714 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
21715 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
21716 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
21717 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
21718 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
21719 _mm_maskz_getmant_ss): New intrinsics.
21720 (__builtin_ia32_getexpss128_mask): Changed to ...
21721 __builtin_ia32_getexpss128_round ... this.
21722 (__builtin_ia32_getexpsd128_mask): Changed to ...
21723 __builtin_ia32_getexpsd128_round ... this.
21724 * config/i386/i386-builtin-types.def
21725 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
21726 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
21727 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
21728 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
21729 __builtin_ia32_getmantss_mask_round): New builtins.
21730 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
21731 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
21732 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
21733 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
21734 * config/i386/sse.md
21735 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
21736 avx512f_sgetexp<mode><mask_scalar_name>
21737 <round_saeonly_scalar_name> ... this.
21738 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
21739 %0, %1, %2<round_saeonly_op3>}): Changed to ...
21740 vgetexp<ssescalarmodesuffix>
21741 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21742 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
21743 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
21744 avx512f_vgetmant<mode><mask_scalar_name>
21745 <round_saeonly_scalar_name> ... this.
21746 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
21747 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
21748 vgetmant<ssescalarmodesuffix>
21749 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
21750 %0<mask_scalar_operand4>, %1, %2
21751 <round_saeonly_scalar_mask_op4>, %3} ... this.
21752 * config/i386/subst.md (mask_scalar_operand4,
21753 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
21754 round_saeonly_scalar_nimm_predicate): New subst attributes.
21756 2017-07-06 Julia Koval <julia.koval@intel.com>
21758 * config/i386/i386.c (ix86_erase_embedded_rounding):
21759 Remove code for old rounding pattern.
21761 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
21763 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
21765 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
21767 * doc/sourcebuild.texi (Test Directives, Variants of
21768 dg-require-support): Add documentation for dg-require-stack-check.
21770 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
21772 * config/i386/subst.md (mask_scalar, round_scalar,
21773 round_saeonly_scalar): New meta-templates.
21774 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
21775 round_scalar_mask_operand3, round_scalar_mask_op3,
21776 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
21777 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
21778 round_saeonly_scalar_constraint,
21779 round_saeonly_scalar_prefix): New subst attribute.
21780 * config/i386/sse.md
21781 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
21782 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
21783 <round_scalar_name> ... this.
21784 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
21785 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
21786 <round_scalar_name> ... this.
21787 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
21788 <sse>_vm<code><mode>3<mask_scalar_name>
21789 <round_saeonly_scalar_name> ... this.
21790 (v<plusminus_mnemonic><ssescalarmodesuffix>
21791 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21792 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21793 v<plusminus_mnemonic><ssescalarmodesuffix>
21794 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21795 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21796 (v<multdiv_mnemonic><ssescalarmodesuffix>
21797 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21798 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21799 v<multdiv_mnemonic><ssescalarmodesuffix>
21800 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21801 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21802 (v<maxmin_float><ssescalarmodesuffix>
21803 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
21804 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
21805 v<maxmin_float><ssescalarmodesuffix>
21806 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21807 %0<mask_scalar_operand3>, %1, %<iptr>2
21808 <round_saeonly_scalar_mask_op3>} ... this.
21810 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
21812 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
21813 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
21815 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
21816 Alan Hayward <alan.hayward@arm.com>
21817 David Sherwood <david.sherwood@arm.com>
21819 * combine.c (simplify_if_then_else): Remove "enum" before
21821 * compare-elim.c (can_eliminate_compare): Likewise.
21822 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
21824 (aarch64_lookup_simd_builtin_type): Likewise.
21825 (aarch64_simd_builtin_type): Likewise.
21826 (aarch64_init_simd_builtin_types): Likewise.
21827 (aarch64_simd_expand_args): Likewise.
21828 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
21830 (aarch64_reverse_mask): Likewise.
21831 (aarch64_simd_emit_reg_reg_move): Likewise.
21832 (aarch64_gen_adjusted_ldpstp): Likewise.
21833 (aarch64_ccmp_mode_to_code): Likewise.
21834 (aarch64_operands_ok_for_ldpstp): Likewise.
21835 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21836 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
21838 (aarch64_min_divisions_for_recip_mul): Likewise.
21839 (aarch64_reassociation_width): Likewise.
21840 (aarch64_get_condition_code_1): Likewise.
21841 (aarch64_simd_emit_reg_reg_move): Likewise.
21842 (aarch64_simd_attr_length_rglist): Likewise.
21843 (aarch64_reverse_mask): Likewise.
21844 (aarch64_operands_ok_for_ldpstp): Likewise.
21845 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21846 (aarch64_gen_adjusted_ldpstp): Likewise.
21847 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
21849 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
21850 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
21851 (arm_lookup_simd_builtin_type): Likewise.
21852 (arm_simd_builtin_type): Likewise.
21853 (arm_init_simd_builtin_types): Likewise.
21854 (arm_expand_builtin_args): Likewise.
21855 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
21856 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
21857 (ft32_setup_incoming_varargs): Likewise.
21858 (ft32_function_arg): Likewise.
21859 (ft32_function_arg_advance): Likewise.
21860 (ft32_pass_by_reference): Likewise.
21861 (ft32_arg_partial_bytes): Likewise.
21862 (ft32_valid_pointer_mode): Likewise.
21863 (ft32_addr_space_pointer_mode): Likewise.
21864 (ft32_addr_space_legitimate_address_p): Likewise.
21865 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
21867 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
21868 (ix86_emit_outlined_ms2sysv_restore): Likewise.
21869 (iamcu_alignment): Likewise.
21870 (canonicalize_vector_int_perm): Likewise.
21871 (ix86_noce_conversion_profitable_p): Likewise.
21872 (ix86_mpx_bound_mode): Likewise.
21873 (ix86_operands_ok_for_move_multiple): Likewise.
21874 * config/microblaze/microblaze-protos.h
21875 (microblaze_expand_conditional_branch_reg): Likewise.
21876 * config/microblaze/microblaze.c
21877 (microblaze_expand_conditional_branch_reg): Likewise.
21878 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
21880 (rs6000_reassociation_width): Likewise.
21881 (rs6000_invalid_binary_op): Likewise.
21882 (fusion_p9_p): Likewise.
21883 (emit_fusion_p9_load): Likewise.
21884 (emit_fusion_p9_store): Likewise.
21885 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
21887 (riscv_hard_regno_mode_ok_p): Likewise.
21888 (riscv_address_insns): Likewise.
21889 (riscv_split_symbol): Likewise.
21890 (riscv_legitimize_move): Likewise.
21891 (riscv_function_value): Likewise.
21892 (riscv_hard_regno_nregs): Likewise.
21893 (riscv_expand_builtin): Likewise.
21894 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
21895 (riscv_build_integer): Likewise.
21896 (riscv_split_integer): Likewise.
21897 (riscv_legitimate_constant_p): Likewise.
21898 (riscv_cannot_force_const_mem): Likewise.
21899 (riscv_regno_mode_ok_for_base_p): Likewise.
21900 (riscv_valid_base_register_p): Likewise.
21901 (riscv_valid_offset_p): Likewise.
21902 (riscv_valid_lo_sum_p): Likewise.
21903 (riscv_classify_address): Likewise.
21904 (riscv_legitimate_address_p): Likewise.
21905 (riscv_address_insns): Likewise.
21906 (riscv_load_store_insns): Likewise.
21907 (riscv_force_binary): Likewise.
21908 (riscv_split_symbol): Likewise.
21909 (riscv_force_address): Likewise.
21910 (riscv_legitimize_address): Likewise.
21911 (riscv_move_integer): Likewise.
21912 (riscv_legitimize_const_move): Likewise.
21913 (riscv_legitimize_move): Likewise.
21914 (riscv_address_cost): Likewise.
21915 (riscv_subword): Likewise.
21916 (riscv_output_move): Likewise.
21917 (riscv_canonicalize_int_order_test): Likewise.
21918 (riscv_emit_int_order_test): Likewise.
21919 (riscv_function_arg_boundary): Likewise.
21920 (riscv_pass_mode_in_fpr_p): Likewise.
21921 (riscv_pass_fpr_single): Likewise.
21922 (riscv_pass_fpr_pair): Likewise.
21923 (riscv_get_arg_info): Likewise.
21924 (riscv_function_arg): Likewise.
21925 (riscv_function_arg_advance): Likewise.
21926 (riscv_arg_partial_bytes): Likewise.
21927 (riscv_function_value): Likewise.
21928 (riscv_pass_by_reference): Likewise.
21929 (riscv_setup_incoming_varargs): Likewise.
21930 (riscv_print_operand): Likewise.
21931 (riscv_elf_select_rtx_section): Likewise.
21932 (riscv_save_restore_reg): Likewise.
21933 (riscv_for_each_saved_reg): Likewise.
21934 (riscv_register_move_cost): Likewise.
21935 (riscv_hard_regno_mode_ok_p): Likewise.
21936 (riscv_hard_regno_nregs): Likewise.
21937 (riscv_class_max_nregs): Likewise.
21938 (riscv_memory_move_cost): Likewise.
21939 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
21940 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
21941 (rl78_addr_space_address_mode): Likewise.
21942 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21944 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
21945 (rs6000_reassociation_width): Likewise.
21946 (rs6000_invalid_binary_op): Likewise.
21947 (fusion_p9_p): Likewise.
21948 (emit_fusion_p9_load): Likewise.
21949 (emit_fusion_p9_store): Likewise.
21950 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
21951 (ok_for_simple_move_operands): Likewise.
21952 (ok_for_simple_move_strict_operands): Likewise.
21953 (ok_for_simple_arith_logic_operands): Likewise.
21954 (visium_legitimize_reload_address): Likewise.
21955 (visium_select_cc_mode): Likewise.
21956 (output_cbranch): Likewise.
21957 (visium_split_double_move): Likewise.
21958 (visium_expand_copysign): Likewise.
21959 (visium_expand_int_cstore): Likewise.
21960 (visium_expand_fp_cstore): Likewise.
21961 * config/visium/visium.c (visium_pass_by_reference): Likewise.
21962 (visium_function_arg): Likewise.
21963 (visium_function_arg_advance): Likewise.
21964 (visium_libcall_value): Likewise.
21965 (visium_setup_incoming_varargs): Likewise.
21966 (visium_legitimate_constant_p): Likewise.
21967 (visium_legitimate_address_p): Likewise.
21968 (visium_legitimize_address): Likewise.
21969 (visium_secondary_reload): Likewise.
21970 (visium_register_move_cost): Likewise.
21971 (visium_memory_move_cost): Likewise.
21972 (prepare_move_operands): Likewise.
21973 (ok_for_simple_move_operands): Likewise.
21974 (ok_for_simple_move_strict_operands): Likewise.
21975 (ok_for_simple_arith_logic_operands): Likewise.
21976 (visium_function_value_1): Likewise.
21977 (rtx_ok_for_offset_p): Likewise.
21978 (visium_legitimize_reload_address): Likewise.
21979 (visium_split_double_move): Likewise.
21980 (visium_expand_copysign): Likewise.
21981 (visium_expand_int_cstore): Likewise.
21982 (visium_expand_fp_cstore): Likewise.
21983 (visium_split_cstore): Likewise.
21984 (visium_select_cc_mode): Likewise.
21985 (visium_split_cbranch): Likewise.
21986 (output_cbranch): Likewise.
21987 (visium_print_operand_address): Likewise.
21988 * expmed.c (flip_storage_order): Likewise.
21989 * expmed.h (emit_cstore): Likewise.
21990 (flip_storage_order): Likewise.
21991 * genrecog.c (validate_pattern): Likewise.
21992 * hsa-gen.c (gen_hsa_addr): Likewise.
21993 * internal-fn.c (expand_arith_overflow): Likewise.
21994 * ira-color.c (allocno_copy_cost_saving): Likewise.
21995 * lra-assigns.c (find_hard_regno_for_1): Likewise.
21996 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
21997 (process_invariant_for_inheritance): Likewise.
21998 * lra-eliminations.c (move_plus_up): Likewise.
21999 * omp-low.c (lower_oacc_reductions): Likewise.
22000 * simplify-rtx.c (simplify_subreg): Likewise.
22001 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
22002 (TARGET_CHKP_BOUND_MODE): Likewise..
22003 * targhooks.c (default_chkp_bound_mode): Likewise.
22004 (default_setup_incoming_vararg_bounds): Likewise.
22005 * targhooks.h (default_chkp_bound_mode): Likewise.
22006 (default_setup_incoming_vararg_bounds): Likewise.
22007 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
22008 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
22009 (have_whole_vector_shift): Likewise.
22010 * tree-vect-stmts.c (vectorizable_load): Likewise.
22011 * doc/tm.texi: Regenerate.
22013 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22015 Graceful degrade if Binutils PR21472 is not available.
22018 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
22019 .rodata in flash test fails.
22020 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
22021 * confgure: Regenerate.
22022 * config.in: Regenerate.
22023 * config/avr/avr.c (avr_asm_named_section)
22024 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
22025 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
22026 (avr_asm_init_sections): Same.
22028 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22030 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
22031 (fma<VH:mode>4_intrinsic): Likewise.
22032 (*fmsub<VCVTF:mode>4): Likewise.
22033 (*fmsub<VH:mode>4_intrinsic): Likewise.
22035 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22038 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
22039 Don't depend on "optimize > 0".
22040 (out_movhi_r_mr, out_movqi_mr_r): Same.
22041 (out_movhi_mr_r, out_movqi_r_mr): Same.
22042 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
22043 io_address_operand on "optimize > 0".
22045 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22047 * tree-loop-distribution.c: Add general explanantion on the pass.
22048 (generate_loops_for_partition): Mark distributed loop.
22049 (pg_add_dependence_edges): New parameter. Handle alias data
22050 dependence specially and record it in the parameter if asked.
22051 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
22052 (init_partition_graph_vertices, add_partition_graph_edge): New.
22053 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
22054 (free_partition_graph_vdata, build_partition_graph): New.
22055 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
22056 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
22057 (data_ref_segment_size, latch_dominated_by_data_ref): New.
22058 (compute_alias_check_pairs, version_loop_by_alias_check): New.
22059 (version_for_distribution_p, finalize_partitions): New.
22060 (distribute_loop): Handle alias data dependence specially. Factor
22061 out loop fusion code as functions and call these functions.
22063 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22065 * tree-loop-distribution.c (classify_partition): New parameter and
22066 better handle reduction statement.
22067 (rdg_build_partitions): Revise comment.
22068 (distribute_loop): Compute statements in all partitions and pass it
22069 to classify_partition.
22071 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22073 * tree-loop-distribution.c (enum partition_type): New.
22074 (struct partition): New field type.
22075 (partition_merge_into): Add parameter. Update partition type.
22076 (data_dep_in_cycle_p, update_type_for_merge): New functions.
22077 (build_rdg_partition_for_vertex): Compute partition type.
22078 (rdg_build_partitions): Dump partition type.
22079 (distribute_loop): Update calls to partition_merge_into.
22081 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22083 * tree-loop-distribution.c (struct ddr_hasher): New.
22084 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
22086 (classify_partition): Call get_data_dependence.
22087 (pg_add_dependence_edges): Ditto.
22088 (distribute_loop): Release data dependence hash table.
22090 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22092 * tree-loop-distribution.c (ref_base_address): Delete.
22093 (similar_memory_accesses): Rename ...
22094 (share_memory_accesses): ... to this. Check if partitions access
22095 the same memory reference.
22096 (distribute_loop): Call share_memory_accesses.
22098 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22100 * tree-loop-distribution.c (struct partition): New field recording
22101 its data reference.
22102 (partition_alloc, partition_free): Init and release data refs.
22103 (partition_merge_into): Merge data refs.
22104 (build_rdg_partition_for_vertex): Collect data refs for partition.
22105 (pg_add_dependence_edges): Change parameters from vector to bitmap.
22107 (distribute_loop): Remve data refs from vertice data of partition
22110 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22112 * tree-loop-distribution.c (params.h): Include header file.
22113 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
22114 (datarefs_vec): New global var.
22115 (create_rdg_vertices): Use datarefs_vec directly.
22116 (free_rdg): Don't free data references.
22117 (build_rdg): Update use. Don't free data references.
22118 (distribute_loop): Compute global variable for data references.
22119 Bail out if there are too many data references.
22121 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22123 * tree-loop-distribution.c (loop_nest): New global var.
22124 (build_rdg): Use loop directly, rather than loop nest.
22125 (pg_add_dependence_edges): Remove loop nest parameter. Use global
22127 (distribute_loop): Compute global variable loop nest. Update use.
22129 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22131 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
22132 (partition_merge_into): New parameter. Dump reason for fusion.
22133 (distribute_loop): Update use of partition_merge_into.
22135 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22137 * tree-loop-distribution.c (bb_top_order_index): New.
22138 (bb_top_order_index_size, bb_top_order_cmp): New.
22139 (stmts_from_loop): Use topological order.
22140 (pass_loop_distribution::execute): Compute and release topological
22141 order for basic blocks.
22143 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22145 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
22148 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22150 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
22151 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
22152 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
22153 * internal-fn.def (LOOP_DIST_ALIAS): New.
22154 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
22155 (fold_loop_internal_call): ... this.
22156 (vect_loop_dist_alias_call): New function.
22157 (set_uid_loop_bbs): Call fold_loop_internal_call.
22158 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
22161 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22164 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
22165 Require dead FLAGS_REG at the beginning of a peephole.
22167 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22170 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
22171 arguments in the call to __builtin_ia32_sbb_u32.
22172 (_subborrow_u64): Swap _X and _Y arguments in the call to
22173 __builtin_ia32_sbb_u64.
22175 2017-07-04 Jakub Jelinek <jakub@redhat.com>
22178 * tree-vrp.c (compare_assert_loc): Turn into a function template
22179 with stable template parameter. Only test if a->e is NULL,
22180 !a->e == !b->e has been verified already. Use e == NULL or
22181 e != NULL instead of e or ! e tests. If stable is true, don't use
22182 iterative_hash_expr, on the other side allow a or b or both NULL
22183 and sort the NULLs last.
22184 (process_assert_insertions): Sort using compare_assert_loc<false>
22185 instead of compare_assert_loc, later sort using
22186 compare_assert_loc<true> before calling process_assert_insertions_for
22187 in a loop. Use break instead of continue once seen NULL pointer.
22189 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22191 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
22192 Cortex-R7 and Cortex-R8 processors.
22194 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22196 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
22197 uninitialized while src is not.
22199 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
22201 * common/config/arm/arm-common.c: Adjust include path for
22203 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
22204 (arm-cpu.h): Create in build directory. Adjust dependency rules.
22205 (arm-cpu-data.h): Likewise.
22206 (arm-cpu-cdata.h): Likewise.
22207 * config/arm/arm-cpu.h: Delete.
22208 * config/arm/arm-cpu-cdata.h: Delete.
22209 * config/arm/arm-cpu-data.h: Delete.
22211 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
22213 * config/arm/arm-cpus.in (cortex-a55): New.
22214 (cortex-a75): Likewise.
22215 (cortex-a75.cortex-a55): Likewise.
22216 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
22218 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
22219 * config/arm/arm-cpu-cdata.h: Regenerate.
22220 * config/arm/arm-cpu-data.h: Regenerate.
22221 * config/arm/arm-cpu.h: Regenerate.
22222 * config/arm/arm-tables.opt: Regenerate.
22223 * config/arm/arm-tune.md: Regenerate.
22225 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22227 * haifa-sched.c (sched_create_recovery_edges): Update profile.
22229 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22231 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
22234 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
22236 PR tree-optimization/81292
22237 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
22238 full_string_p, also call adjust_related_strinfos if the adjustment
22239 is simple, otherwise invalidate related strinfos.
22241 2017-07-04 Martin Liska <mliska@suse.cz>
22244 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
22245 newly created variable as DECL_IGNORED_P.
22247 2017-07-04 Martin Liska <mliska@suse.cz>
22250 * ipa-inline.c (inline_small_functions):
22251 Use xstrdup_for_dump.
22253 2017-07-04 Tom de Vries <tom@codesourcery.com>
22255 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
22257 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22260 * config/darwin.c (darwin_function_switched_text_sections):
22263 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
22265 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
22267 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22269 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
22271 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22273 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
22274 min_profitable_iters, and th as inclusive lower bounds.
22275 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
22276 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
22277 for min_profitable_iters and min_profitable_estimate.
22278 (vect_transform_loop): Treat th as an inclusive lower bound.
22279 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22281 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22284 * config/darwin.c (darwin_function_switched_text_sections):
22285 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
22286 in two pieces, and suppress the use of buf.
22288 2017-07-03 Nathan Sidwell <nathan@acm.org>
22290 * hash-table.h (hash_table_mod1): Fix indentation.
22292 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22294 PR middle-end/81290
22295 * predict.c (force_edge_cold): Be more careful about propagation
22297 * profile-count.h (profile_probability::guessed,
22298 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
22300 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
22302 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
22304 * doc/invoke.texi (rcpc architecture extension): Document it.
22306 2017-07-03 Richard Biener <rguenther@suse.de>
22308 PR tree-optimization/60510
22309 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
22310 the scalar reduction PHI and use it.
22311 (vectorizable_reduction): Properly guard the single_defuse_cycle
22312 path for non-SLP reduction chains where we cannot use it.
22313 Rework reduc_def/index and vector type deduction. Rework
22314 vector operand gathering during reduction op code-gen.
22315 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
22316 chains dissolve the chain and leave it to non-SLP reduction
22319 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22321 * tree-data-ref.h (dr_alignment): Declare.
22322 * tree-data-ref.c (dr_alignment): New function.
22323 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
22324 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
22326 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
22328 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22330 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
22331 and base_misalignment fields.
22332 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
22333 * tree-data-ref.c: Include builtins.h.
22334 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
22335 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
22336 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
22337 * tree-vect-data-refs.c: Include tree-cfg.h.
22338 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
22339 fields instead of calculating an alignment here.
22340 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
22341 innermost_loop_behavior fields.
22343 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22345 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
22347 (DR_STEP_ALIGNMENT): New macro.
22348 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
22349 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
22350 (create_data_ref): Print it.
22351 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
22352 to tell whether the step preserves vector (mis)alignment.
22353 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22354 Move the check for an integer step and generalise to all INTEGER_CST.
22355 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
22356 Print the outer step alignment.
22358 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22360 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
22361 with offset_alignment.
22362 (DR_ALIGNED_TO): Delete.
22363 (DR_OFFSET_ALIGNMENT): New macro.
22364 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
22365 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
22366 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
22367 (create_data_ref): Likewise.
22368 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22369 (vect_analyze_data_refs): Likewise.
22370 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
22371 creating dummy innermost behavior.
22373 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22375 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
22376 with a "innermost_loop_behavior *" and refeence tree.
22377 * tree-data-ref.c (dr_analyze_innermost): Likewise.
22378 (create_data_ref): Update call accordingly.
22379 * tree-predcom.c (find_looparound_phi): Likewise.
22381 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22383 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
22384 fields with dr_wrt_vec_loop.
22385 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
22386 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
22387 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
22388 (vect_dr_behavior): New function.
22389 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22390 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
22391 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
22392 track whether the step preserves the misalignment.
22393 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22394 Use vect_dr_behavior.
22395 (vect_setup_realignment): Update call accordingly.
22396 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
22397 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
22398 call to vect_create_addr_base_for_vector_ref.
22399 (vect_create_cond_for_align_checks): Likewise.
22400 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
22401 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
22402 (vect_recog_mask_conversion_pattern): Likewise.
22403 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
22404 (new_stmt_vec_info): Remove redundant zeroing.
22406 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22408 * common/config/arm/arm-common.c (arm_be8_option): New function.
22409 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
22410 (ISA_ARMv6): Add isa_bit_be8.
22411 * config/arm/arm.h (arm_be8_option): Add prototype.
22412 (BE8_SPEC_FUNCTION): New define.
22413 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
22414 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
22415 (mlittle-endian): Similarly.
22416 (mbe8, mbe32): New options.
22417 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
22418 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
22420 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22422 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
22424 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22426 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
22427 (cleanup_tree_cfg_bb): Use it.
22428 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
22430 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
22432 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22435 * loop-doloop.c (add_test): Update profile.
22437 2017-07-03 Martin Liska <mliska@suse.cz>
22440 * sanopt.c (rewrite_usage_of_param): New function.
22441 (sanitize_rewrite_addressable_params): Likewise.
22442 (pass_sanopt::execute): Call rewrite_usage_of_param.
22444 2017-07-03 Richard Biener <rguenther@suse.de>
22446 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
22447 back to using VIEW_CONVERT_EXPR.
22449 2017-07-03 Martin Liska <mliska@suse.cz>
22452 * doc/extend.texi: Document when a resolver function is
22453 generated for target_clones.
22455 2017-07-03 Martin Liska <mliska@suse.cz>
22457 * asan.c (asan_emit_stack_protection): Unpoison just red zones
22458 and shadow memory of auto variables which are subject of
22459 use-after-scope sanitization.
22460 (asan_expand_mark_ifn): Add do set only when is_poison.
22462 2016-07-03 Richard Biener <rguenther@suse.de>
22464 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
22466 (vect_force_simple_reduction): Record reduction def -> phi mapping.
22467 (vectorizable_reduction): Perform reduction PHI creation when
22468 visiting a reduction PHI and adjust and simplify code generation
22469 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
22470 (vect_transform_loop): Visit reduction PHIs.
22471 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
22472 defs into the SLP tree.
22473 (vect_build_slp_tree): Reduction defs terminate the recursion.
22474 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
22476 (vect_get_vec_defs_for_stmt_copy): Export.
22477 (vect_get_vec_defs): Likewise.
22478 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
22480 (vect_get_vec_defs_for_stmt_copy): Declare.
22481 (vect_get_vec_defs): Likewise.
22483 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22485 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
22486 parameter with a "loop" parameter and use it instead of the
22487 loop containing DR_STMT. Don't check simple_iv when doing
22488 BB analysis. Describe the two analysis modes in the comment.
22490 2017-07-03 Tom de Vries <tom@codesourcery.com>
22492 PR tree-optimization/69468
22493 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
22494 (find_same_succ_bb): Handle ignore_edge_flags.
22496 2017-07-03 Tom de Vries <tom@codesourcery.com>
22498 PR tree-optimization/81192
22499 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
22501 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
22503 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
22505 2017-07-03 Tom de Vries <tom@codesourcery.com>
22507 PR tree-optimization/81192
22508 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
22509 BB_SAME_SUCC (bb) == NULL.
22511 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22513 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
22516 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22518 * dumpfile.c: Include profile-count.h
22519 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
22521 (insert_cond_bb): Update profile.
22522 * tree-cfg.h (insert_cond_bb): Update prototype.
22523 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
22524 * tree-dump.c: Do not include tree-cfg.
22526 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22528 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
22530 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22532 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
22535 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22537 * tree-complex.c (expand_complex_div_wide): update profile.
22539 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22540 Alan Hayward <alan.hayward@arm.com>
22541 David Sherwood <david.sherwood@arm.com>
22543 * Makefile.in (MACHMODE_H): Remove insn-modes.h
22544 (CORETYPES_H): New define.
22545 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
22546 (insn-modes-inline.h, s-modes-inline-h): New rules.
22547 (generated_files): Add insn-modes-inline.h.
22548 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
22549 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
22550 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
22551 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
22552 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
22553 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
22554 (build/gencodes.o, build/genconditions.o): Likewise.
22555 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
22556 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
22557 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
22558 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
22559 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
22560 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
22561 * coretypes.h: Include everything up to real.h for generators.
22562 Include insn-modes.h first. Include wide-int-print.h after
22563 wide-int.h. Include insn-modes-inline.h and then machmode.h.
22564 * machmode.h: Don't include insn-modes.h here.
22565 * function-tests.c: Remove includes of signop.h, machmode.h,
22566 double-int.h and wide-int.h.
22568 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
22570 * optc-save-gen.awk: Likewise.
22571 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
22572 * godump.c: Remove include of wide-int-print.h.
22573 * pretty-print.h: Likewise.
22574 * wide-int-print.cc: Likewise.
22575 * wide-int.cc: Likewise.
22576 * hash-map-tests.c: Remove include of signop.h.
22577 * hash-set-tests.c: Likewise.
22578 * rtl-tests.c: Likewise.
22579 * mkconfig.sh: Remove include of machmode.h.
22580 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
22582 (emit_insn_modes_inline_h): ...this new function. Emit the code
22583 into an insn-modes-inline.h header file, adding appropriate
22584 include guards and end comments.
22585 (emit_insn_modes_c_header): Remove include of machmode.h.
22586 (emit_min_insn_modes_c_header): Include coretypes.h rather than
22588 (main): Handle -i flag and call emit_insn_modes_inline_h when
22591 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22593 * tree-ssa-strlen.c (strinfo): Rename the length field to
22594 nonzero_chars. Add a full_string_p field.
22595 (compare_nonzero_chars, zero_length_string_p): New functions.
22596 (get_addr_stridx): Add an offset_out parameter.
22597 Use compare_nonzero_chars.
22598 (get_stridx): Update accordingly. Use compare_nonzero_chars.
22599 (new_strinfo): Update after above changes to strinfo.
22600 (set_endptr_and_length): Set full_string_p.
22601 (get_string_length): Update after above changes to strinfo.
22602 (unshare_strinfo): Update call to new_strinfo.
22603 (maybe_invalidate): Likewise.
22604 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
22605 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
22606 as a uhwi instead of an shwi. Update after above changes to
22607 strinfo and new_strinfo.
22608 (zero_length_string): Assert that chainsi contains full strings.
22609 Use zero_length_string_p. Update call to new_strinfo.
22610 (adjust_related_strinfos): Update after above changes to strinfo.
22611 Copy full_string_p from origsi.
22612 (adjust_last_stmt): Use zero_length_string_p.
22613 (handle_builtin_strlen): Update after above changes to strinfo and
22614 new_strinfo. Install the lhs as the string length if the previous
22615 entry didn't describe a full string.
22616 (handle_builtin_strchr): Update after above changes to strinfo
22618 (handle_builtin_strcpy): Likewise.
22619 (handle_builtin_strcat): Likewise.
22620 (handle_builtin_malloc): Likewise.
22621 (handle_pointer_plus): Likewise.
22622 (handle_builtin_memcpy): Likewise. Track nonzero characters
22623 that aren't necessarily followed by a nul terminator.
22624 (handle_char_store): Likewise.
22626 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22628 PR tree-optimization/80769
22629 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
22630 for malloc and calloc. Document the new invariant that all related
22631 strinfos have delayed lengths or none do.
22632 (verify_related_strinfos): Move earlier in file.
22633 (set_endptr_and_length): New function, split out from...
22634 (get_string_length): ...here. Also set the lengths of related
22636 (zero_length_string): Assert that chainsi has known (rather than
22638 (adjust_related_strinfos): Likewise.
22640 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22642 PR tree-optimization/81136
22643 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
22644 assert that two references with the same misalignment have the same
22645 compile-time misalignment if those compile-time misalignments
22648 2017-07-01 Andi Kleen <ak@linux.intel.com>
22650 * print-tree.c (print_node): Print all attributes.
22652 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22654 * cfg.c (scale_bbs_frequencies): New function.
22655 * cfg.h (scale_bbs_frequencies): Declare it.
22656 * cfgloopanal.c (single_likely_exit): Cleanup.
22657 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
22659 (scale_loop_profile): Likewise.
22660 (loop_version): Likewise.
22661 (create_empty_loop_on_edge): Update.
22662 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
22663 scale_loop_frequencies, scale_loop_profile, loopify,
22664 loop_version): Update prototypes.
22665 * modulo-sched.c (sms_schedule): Update.
22666 * predict.c (unlikely_executed_edge_p): Also check probability.
22667 (probably_never_executed_edge_p): Fix typo.
22668 * tree-if-conv.c (version_loop_for_if_conversion): Update.
22669 * tree-parloops.c (gen_parallel_loop): Update.
22670 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
22671 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22672 * tree-ssa-loop-split.c (split_loop): Update.
22673 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
22674 * tree-vect-loop-manip.c (vect_do_peeling): Update.
22675 (vect_loop_versioning): Update.
22676 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22678 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22680 * trans-mem.c (split_bb_make_tm_edge): Update profile.
22682 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22684 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
22685 to keep profile consistent.
22687 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22689 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
22690 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
22691 * profile-count.h (max_safe_multiplier): Make unsigned.
22692 (profile_count::guessed_zero): New.
22694 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22696 * bb-reorder.c (fix_up_crossing_landing_pad,
22697 fix_crossing_conditional_branches): Use make_single_succ_edge
22698 to keep profile consistent.
22700 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22702 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
22705 2017-07-01 Jakub Jelinek <jakub@redhat.com>
22708 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
22709 the right scopes, make sure cond_jump isn't preserved between multiple
22710 iterations. Search for fallthru edge whenever there are 3+ edges and
22711 use find_fallthru_edge for it.
22713 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22715 Patch by Alexander Monakov <amonakov@ispras.ru>
22716 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
22717 probabilities consistently.
22719 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22721 * pa.c (pa_expand_compare_and_swap_loop): Update call of
22722 emit_cmp_and_jump_insns.
22724 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22727 * tree-inline.c (expand_call_inline): Combine profile statuses.
22729 2017-06-30 Andrew Pinski <apinski@cavium.com>
22731 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
22732 fold_stmt returned true.
22734 2017-06-30 Nathan Sidwell <nathan@acm.org>
22736 * ggc.h (empty_string): Delete.
22737 * cfgexpand.c (expand_asm_stmt): Use plain "".
22738 * optabs.c (expand_asm_memory_barrier): Likewise.
22739 * stringpool.c (empty_string): Delete.
22740 (digit_vector, digit_string): Delete.
22741 (ggc_alloc_string): Use plain "", don't optimize single digit
22742 strings. Use ggc_alloc_atomic.
22744 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
22746 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
22747 comparison set and one other set, use the cost of the non-comparison
22750 2017-06-30 Nathan Sidwell <nathan@acm.org>
22752 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
22755 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
22757 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
22758 loops. Remove now unneeded calls to gimple_switch_set_label() that
22759 just set removed labels to NULL_TREE.
22761 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
22763 * tree-ssanames.c (set_range_info_raw): Abstract from ...
22764 (set_range_info): ...here. Only call set_range_info_raw if domain
22766 (set_nonzero_bits): Call set_range_info_raw.
22767 * tree-ssanames.h (set_range_info_raw): New.
22769 2017-06-30 Jakub Jelinek <jakub@redhat.com>
22772 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
22773 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
22774 of nonimmediate_operand and <store_mask_constraint> instead of m
22775 for the input operand. For V8FI iterator, always split if input
22776 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
22777 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
22778 <store_mask_predicate> instead of register_operand and
22779 <store_mask_constraint> instead of v for the input operand. Make
22780 sure both operands aren't MEMs for if not <mask_applied>.
22782 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
22784 * lto-wrapper.c (copy_file) Close both file descriptors before
22787 2017-06-30 Martin Liska <mliska@suse.cz>
22790 * multiple_target.c (create_dispatcher_calls): Make ifunc
22791 also for function that don't have calls or are not referenced.
22793 2017-06-30 Richard Biener <rguenther@suse.de>
22795 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
22796 analyze the first scalar stmt. Move vector type computation
22797 for the BB case here from ...
22798 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
22799 live operation processing in the SLP case properly.
22801 2017-06-30 Richard Biener <rguenther@suse.de>
22803 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
22805 2017-06-30 Martin Liska <mliska@suse.cz>
22808 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
22809 before BUILT_IN_UNWIND_RESUME when ASAN is used.
22811 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
22813 * doc/invoke.texi (AArch64): Add missing options and remove redundant
22816 2017-06-30 Richard Biener <rguenther@suse.de>
22818 PR tree-optimization/81249
22819 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
22820 condition reduction result to original scalar type.
22822 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22824 * profile-count.h (enum profile_quality): Fix typos and whitespace
22827 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22829 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
22830 type for branch probabilities.
22832 2017-06-29 Julian Brown <julian@codesourcery.com>
22833 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22835 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
22836 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
22837 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
22838 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
22840 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22842 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
22843 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
22844 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
22845 CC usage from generic code to here.
22846 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
22847 CC usage into the target macros.
22849 2017-06-29 Maya Rashish <coypu@sdf.org>
22851 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
22854 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22856 * arm/arm-builtins.c: Include profile-count.h
22857 * except.c (sjlj_emit_function_enter): Use
22858 profile_probability::unlikely.
22860 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22862 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
22863 and tocrel_offset be pointer args rather than implicitly using
22865 (legitimate_constant_pool_address_p, rs6000_emit_move,
22866 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
22867 tocrel_offset and use in toc_relative_expr_p call.
22868 (print_operand, print_operand_address): Use static tocrel_base_oac
22869 and tocrel_offset_oac.
22870 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
22873 2017-06-29 Maya Rashish <coypu@sdf.org>
22875 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
22877 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
22879 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
22880 objects, take into account only the alignment of 'op0' and 'mode1' if
22883 2017-06-29 Steve Ellcey <sellcey@cavium.com>
22885 * ccmp.c (ccmp_tree_comparison_p): New function.
22886 (ccmp_candidate_p): Update to use above function.
22887 (get_compare_parts): New function.
22888 (expand_ccmp_next): Update to use new functions.
22889 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
22891 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
22892 take mode as argument.
22893 * ccmp.h (expand_ccmp_expr): Add mode as argument.
22894 * expr.c (expand_expr_real_1): Pass mode as argument.
22896 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
22898 * combine.c (combine_instructions): Print insns to dump_file, together
22901 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22903 * asan.c (asan_emit_stack_protection): Update.
22904 (create_cond_insert_point): Update.
22905 * auto-profile.c (afdo_propagate_circuit): Update.
22906 * basic-block.h (struct edge_def): Turn probability to
22907 profile_probability.
22908 (EDGE_FREQUENCY): Update.
22909 * bb-reorder.c (find_traces_1_round): Update.
22910 (better_edge_p): Update.
22911 (sanitize_hot_paths): Update.
22912 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
22913 (make_single_succ_edge): Update.
22914 (check_bb_profile): Update.
22915 (dump_edge_info): Update.
22916 (update_bb_profile_for_threading): Update.
22917 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
22918 probabilitycount to 0.
22919 * cfgbuild.c (compute_outgoing_frequencies): Update.
22920 * cfgcleanup.c (try_forward_edges): Update.
22921 (outgoing_edges_match): Update.
22922 (try_crossjump_to_edge): Update.
22923 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
22924 (expand_gimple_tailcall): Update.
22925 (construct_init_block): Use make_single_succ_edge.
22926 (construct_exit_block): Use make_single_succ_edge.
22927 * cfghooks.c (verify_flow_info): Update.
22928 (redirect_edge_succ_nodup): Update.
22929 (split_edge): Update.
22930 (account_profile_record): Update.
22931 * cfgloopanal.c (single_likely_exit): Update.
22932 * cfgloopmanip.c (scale_loop_profile): Update.
22933 (set_zero_probability): Remove.
22934 (duplicate_loop_to_header_edge): Update.
22935 * cfgloopmanip.h (loop_version): Update prototype.
22936 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
22937 (force_nonfallthru_and_redirect): Update.
22938 (update_br_prob_note): Update.
22939 (rtl_verify_edges): Update.
22940 (purge_dead_edges): Update.
22941 (rtl_lv_add_condition_to_bb): Update.
22942 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
22943 * cgraphunit.c (init_lowered_empty_function): Update.
22944 (cgraph_node::expand_thunk): Update.
22945 * cilk-common.c: Include profile-count.h
22946 * dojump.c (inv): Remove.
22947 (jumpifnot): Update.
22948 (jumpifnot_1): Update.
22949 (do_jump_1): Update.
22951 (do_jump_by_parts_greater_rtx): Update.
22952 (do_compare_rtx_and_jump): Update.
22953 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
22954 do_jump_1. do_compare_rtx_and_jump): Update prototype.
22955 * dwarf2cfi.c: Include profile-count.h
22956 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
22957 (sjlj_emit_dispatch_table): Likewise.
22958 * explow.c: Include profile-count.h
22959 * expmed.c (emit_store_flag_force): Update.
22960 (do_cmp_and_jump): Update.
22961 * expr.c (compare_by_pieces_d::generate): Update.
22962 (compare_by_pieces_d::finish_mode): Update.
22963 (emit_block_move_via_loop): Update.
22964 (store_expr_with_bounds): Update.
22965 (store_constructor): Update.
22966 (expand_expr_real_2): Update.
22967 (expand_expr_real_1): Update.
22968 * expr.h (try_casesi, try_tablejump): Update prototypes.
22969 * gimple-pretty-print.c (dump_probability): Update.
22970 (dump_profile): New.
22971 (dump_gimple_label): Update.
22972 (dump_gimple_bb_header): Update.
22973 * graph.c (draw_cfg_node_succ_edges): Update.
22974 * hsa-gen.c (convert_switch_statements): Update.
22975 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
22976 (find_if_case_1): Update.
22977 (find_if_case_2): Update.
22978 * internal-fn.c (expand_arith_overflow_result_store): Update.
22979 (expand_addsub_overflow): Update.
22980 (expand_neg_overflow): Update.
22981 (expand_mul_overflow): Update.
22982 (expand_vector_ubsan_overflow): Update.
22983 * ipa-cp.c (good_cloning_opportunity_p): Update.
22984 * ipa-split.c (split_function): Use make_single_succ_edge.
22985 * ipa-utils.c (ipa_merge_profiles): Update.
22986 * loop-doloop.c (add_test): Update.
22987 (doloop_modify): Update.
22988 * loop-unroll.c (compare_and_jump_seq): Update.
22989 (unroll_loop_runtime_iterations): Update.
22990 * lra-constraints.c (lra_inheritance): Update.
22991 * lto-streamer-in.c (input_cfg): Update.
22992 * lto-streamer-out.c (output_cfg): Update.
22993 * mcf.c (adjust_cfg_counts): Update.
22994 * modulo-sched.c (sms_schedule): Update.
22995 * omp-expand.c (expand_omp_for_init_counts): Update.
22996 (extract_omp_for_update_vars): Update.
22997 (expand_omp_ordered_sink): Update.
22998 (expand_omp_for_ordered_loops): Update.
22999 (expand_omp_for_generic): Update.
23000 (expand_omp_for_static_nochunk): Update.
23001 (expand_omp_for_static_chunk): Update.
23002 (expand_cilk_for): Update.
23003 (expand_omp_simd): Update.
23004 (expand_omp_taskloop_for_outer): Update.
23005 (expand_omp_taskloop_for_inner): Update.
23006 * omp-simd-clone.c (simd_clone_adjust): Update.
23007 * optabs.c (expand_doubleword_shift): Update.
23008 (expand_abs): Update.
23009 (emit_cmp_and_jump_insn_1): Update.
23010 (expand_compare_and_swap_loop): Update.
23011 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
23012 * predict.c (predictable_edge_p): Update.
23013 (edge_probability_reliable_p): Update.
23014 (set_even_probabilities): Update.
23015 (combine_predictions_for_insn): Update.
23016 (combine_predictions_for_bb): Update.
23017 (propagate_freq): Update.
23018 (estimate_bb_frequencies): Update.
23019 (force_edge_cold): Update.
23020 * profile-count.c (profile_count::dump): Add missing space into dump.
23021 (profile_count::debug): Add newline.
23022 (profile_count::differs_from_p): Explicitly convert to unsigned.
23023 (profile_count::stream_in): Update.
23024 (profile_probability::dump): New member function.
23025 (profile_probability::debug): New member function.
23026 (profile_probability::differs_from_p): New member function.
23027 (profile_probability::differs_lot_from_p): New member function.
23028 (profile_probability::stream_in): New member function.
23029 (profile_probability::stream_out): New member function.
23030 * profile-count.h (profile_count_quality): Rename to ...
23031 (profile_quality): ... this one.
23032 (profile_probability): New.
23033 (profile_count): Update.
23034 * profile.c (compute_branch_probabilities): Update.
23035 * recog.c (peep2_attempt): Update.
23036 * sched-ebb.c (schedule_ebbs): Update.
23037 * sched-rgn.c (find_single_block_region): Update.
23038 (compute_dom_prob_ps): Update.
23039 (schedule_region): Update.
23040 * sel-sched-ir.c (compute_succs_info): Update.
23041 * stmt.c (struct case_node): Update.
23042 (do_jump_if_equal): Update.
23043 (get_outgoing_edge_probs): Update.
23044 (conditional_probability): Update.
23045 (emit_case_dispatch_table): Update.
23046 (expand_case): Update.
23047 (expand_sjlj_dispatch_table): Update.
23048 (emit_case_nodes): Update.
23049 * targhooks.c: Update.
23050 * tracer.c (better_p): Update.
23051 (find_best_successor): Update.
23052 * trans-mem.c (expand_transaction): Update.
23053 * tree-call-cdce.c: Update.
23054 * tree-cfg.c (gimple_split_edge): Upate.
23055 (move_sese_region_to_fn): Upate.
23056 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
23057 * tree-eh.c (lower_resx): Upate.
23058 (cleanup_empty_eh_move_lp): Upate.
23059 * tree-if-conv.c (version_loop_for_if_conversion): Update.
23060 * tree-inline.c (copy_edges_for_bb): Update.
23061 (copy_cfg_body): Update.
23062 * tree-parloops.c (gen_parallel_loop): Update.
23063 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
23064 (gimple_gen_time_profiler): Update.
23065 * tree-ssa-dce.c (remove_dead_stmt): Update.
23066 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
23067 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
23068 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
23069 (unloop_loops): Update.
23070 (try_peel_loop): Update.
23071 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23072 * tree-ssa-loop-split.c (connect_loops): Update.
23073 (split_loop): Update.
23074 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
23075 (hoist_guard): Update.
23076 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
23077 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
23078 (value_replacement): Update.
23079 * tree-ssa-reassoc.c (branch_fixup): Update.
23080 * tree-ssa-tail-merge.c (replace_block_by): Update.
23081 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
23082 (create_edge_and_update_destination_phis): Update.
23083 (compute_path_counts): Update.
23084 (recompute_probabilities): Update.
23085 (update_joiner_offpath_counts): Update.
23086 (freqs_to_counts_path): Update.
23087 (duplicate_thread_path): Update.
23088 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
23089 (struct switch_conv_info): Update.
23090 (gen_inbound_check): Update.
23091 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23092 (vect_do_peeling): Update.
23093 (vect_loop_versioning): Update.
23094 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23095 (optimize_mask_stores): Update.
23096 * ubsan.c (ubsan_expand_null_ifn): Update.
23097 * value-prof.c (gimple_divmod_fixed_value): Update.
23098 (gimple_divmod_fixed_value_transform): Update.
23099 (gimple_mod_pow2): Update.
23100 (gimple_mod_pow2_value_transform): Update.
23101 (gimple_mod_subtract): Update.
23102 (gimple_mod_subtract_transform): Update.
23103 (gimple_ic): Update.
23104 (gimple_stringop_fixed_value): Update.
23105 (gimple_stringops_transform): Update.
23106 * value-prof.h: Update.
23108 2017-06-29 Carl Love <cel@us.ibm.com>
23110 * config/rs6000/rs6000-c.c: Add support for built-in functions
23111 vector signed int vec_signed (vector float);
23112 vector signed long long vec_signed (vector double);
23113 vector signed int vec_signed2 (vector double, vector double);
23114 vector signed int vec_signede (vector double);
23115 vector signed int vec_signedo (vector double);
23116 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
23117 instruction generator.
23118 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23119 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
23120 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
23122 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
23123 vunsignede_v2df): Add define_expands.
23124 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
23125 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
23126 VEC_UNSIGNEDO): Add definitions.
23127 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23128 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
23129 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
23130 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
23131 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
23132 * config/rs6000/altivec.h (vec_signed, vec_signed2,
23133 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
23134 vec_unsignede, vec_unsignedo): Add builtin defines.
23135 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
23137 * doc/extend.texi: Update the built-in documentation file for the
23138 new built-in functions.
23140 2017-06-29 Richard Biener <rguenther@suse.de>
23142 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
23143 reduction chains to LOOP_VINFO_REDUCTIONS.
23144 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
23145 SLP reductions after processing reduction chains.
23147 2017-06-29 Nathan Sidwell <nathan@acm.org>
23149 * builtins.c (fold_builtin_FUNCTION): Use
23150 lang_hooks.decl_printable_name.
23152 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
23154 PR middle-end/81194
23155 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
23156 with only one label.
23157 * stmt.c (expand_case): Assert NCASES is greater than one.
23159 2017-06-29 Richard Biener <rguenther@suse.de>
23161 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
23163 (group_case_labels): Likewise.
23164 (find_taken_edge): Push sanity checking on val to workers...
23165 (find_taken_edge_cond_expr): ... here
23166 (find_taken_edge_switch_expr): ... and here, handle cases
23167 with just a default label.
23168 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
23169 (group_case_labels): Likewise.
23170 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
23171 group_case_labels does anything cleanup the CFG again.
23173 2017-06-29 Bin Cheng <bin.cheng@arm.com>
23175 PR tree-optimization/81196
23176 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
23177 exit condition comparing two IVs.
23179 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
23181 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
23182 profile to the dummy entry at the end of the list of architectures.
23183 * config/arm/arm-cpu-cdata.h: Regenerated.
23185 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23186 Michael Collison <michael.collison@arm.com>
23189 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
23191 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
23192 (*aarch64_reg_<mode>3_minus_mask): New pattern.
23193 (*aarch64_<optab>_reg_di3_mask2): New pattern.
23194 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
23195 of shift when the shift amount is masked with constant equal to
23196 the size of the mode.
23197 * config/aarch64/predicates.md (subreg_lowpart_operator): New
23200 2017-06-29 Martin Liska <mliska@suse.cz>
23202 * config/i386/i386.opt: Change range from [1,5] to [0,5].
23204 2017-06-29 Yury Gribov <tetra2005@gmail.com>
23207 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
23210 (edge_growth_cache_entry::edge_growth_cache_entry): New
23212 (reset_edge_growth_cache): Update to use constructor.
23214 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23216 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
23217 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23218 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23220 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
23222 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
23223 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
23225 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
23227 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
23228 (*-linux-uclibc*): Add t-uclibc tmake_file.
23229 * config/t-musl: New.
23230 * config/t-uclibc: New.
23232 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
23234 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
23236 (gen_comm_data): Emit architectural setting of arch_prof.
23237 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
23239 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
23240 (armv8-m.base, armv8-m.main): Likewise.
23241 * arm-protos.h (arm_build_target): Add profile field.
23242 (arch_option): Likewise.
23243 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
23245 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
23246 arm_active_target.profile.
23248 2017-06-28 Richard Biener <rguenther@suse.de>
23250 PR middle-end/81227
23251 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
23252 TYPE_OVERFLOW_WRAPS.
23253 * match.pd (negate_expr_p): Likewise.
23254 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
23255 fold_build2, not fold_binary.
23257 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23259 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
23260 Convert memory address to Pmode.
23261 (aarch64_print_operand): Assert MEM operands are always Pmode.
23263 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23266 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
23267 Remove redundant if.
23268 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
23269 * config/arm/aarch-common-protos.h
23270 (aarch_forward_to_shift_is_not_shifted_re): Remove.
23271 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
23273 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
23276 * multiple_target.c (create_dispatcher_calls): Set the default
23277 clone to be static, not public.
23279 2017-06-28 Richard Biener <rguenther@suse.de>
23281 * tree-vect-loop.c (vectorizable_reduction): Move special
23282 cond reduction IV var creation ...
23283 (vect_create_epilog_for_reduction): ... here. Remove induction_index
23284 parameter. Use STMT_VINFO_VECTYPE.
23285 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
23288 2017-06-28 Martin Liska <mliska@suse.cz>
23291 * ipa-visibility.c (non_local_p): Handle visibility.
23293 2017-06-28 Martin Liska <mliska@suse.cz>
23296 * common.opt: Add IntegerRange to various options.
23297 * opt-functions.awk (integer_range_info): New function.
23298 * optc-gen.awk: Add integer_range_info to cl_options struct.
23299 * opts-common.c (decode_cmdline_option): Handle
23300 CL_ERR_INT_RANGE_ARG.
23301 (cmdline_handle_error): Likewise.
23302 * opts.c (print_filtered_help): Show valid interval in
23303 when --help is provided.
23304 * opts.h (struct cl_option): Add range_min and range_max fields.
23305 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
23307 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
23309 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
23310 (x * C EQ/NE y * C): New transformation.
23312 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
23314 * genmultilib (combination_space): Accept '+' in option names.
23316 2017-06-28 Martin Liska <mliska@suse.cz>
23319 * asan.c (instrument_derefs): Bail out inner references
23320 that are hard register variables.
23322 2017-06-28 Jakub Jelinek <jakub@redhat.com>
23325 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
23326 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
23328 2017-06-28 Richard Biener <rguenther@suse.de>
23330 * tree-vectorizer.h (vect_get_vec_defs): Remove.
23331 (vect_get_slp_defs): Adjust.
23332 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
23334 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
23336 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
23337 get_initial_defs_for_reduction instead of vect_get_vec_defs.
23338 (vectorizable_reduction): Adjust.
23339 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
23341 (vect_get_slp_defs): Likewise.
23342 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
23343 (vectorizable_bswap): Adjust.
23344 (vectorizable_call): Likewise.
23345 (vectorizable_conversion): Likewise.
23346 (vectorizable_assignment): Likewise.
23347 (vectorizable_shift): Likewise.
23348 (vectorizable_operation): Likewise.
23349 (vectorizable_store): Likewise.
23350 (vectorizable_condition): Likewise.
23351 (vectorizable_comparison): Likewise.
23353 2017-06-28 Michael Collison <michael.collison@arm.com>
23356 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
23358 (arm_gen_movmemqi): Removed unused variable 'i'.
23359 Convert 'for' loop into 'while' loop.
23360 (arm_expand_prologue): Remove last unnecessary set of insn.
23361 (thumb_pop): Remove unused variable 'pushed_words'.
23362 (thumb_exit): Remove last unnecessary set of regs_to_pop.
23364 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23366 * config/s390/predicates.md: Use s390_rel_address_ok_p.
23367 * config/s390/s390-protos.h: Add prototype of
23368 s390_rel_address_ok_p.
23369 * config/s390/s390.c (s390_got_symbol): New function.
23370 (s390_rel_address_ok_p): New function.
23371 (legitimize_pic_address): Use s390_rel_address_ok_p.
23372 (s390_load_got): Use s390_got_symbol.
23373 (s390_option_override): Issue error if
23374 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
23375 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
23377 * config/s390/s390.opt: New option mpic-data-is-text-relative.
23379 2017-06-27 Andrew Pinski <apinski@cavium.com>
23381 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
23382 (X * copysign (1.0, X)): New pattern.
23383 (X * copysign (1.0, -X)): New pattern.
23384 (copysign (-1.0, CST)): New pattern.
23386 2017-06-27 Joseph Myers <joseph@codesourcery.com>
23388 * genmultilib (combination_space): Remove variable.
23389 Validate reuse rules against regular expression for any sequence
23390 of multilib options in any order.
23392 2017-06-27 Michael Collison <michael.collison@arm.com>
23394 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
23395 call aarch64_split_simd_combine.
23396 * (aarch64_combine_internal<mode>): Delete pattern.
23397 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
23398 Allow register and subreg operands.
23400 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23402 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
23403 specific need, just fallback on defaults.
23404 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
23406 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23407 Olivier Hainque <hainque@adacore.com>
23409 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
23411 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
23412 targets. Pick a default if no particular attempt applied.
23413 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
23416 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23418 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
23419 (x86_64-wrs-vxworks7): Likewise.
23421 2017-06-27 Marek Polacek <polacek@redhat.com>
23424 * ubsan.c (instrument_null): Check get_base_address's result for null.
23426 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
23428 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
23430 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
23432 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
23433 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
23434 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
23435 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
23436 New function types.
23437 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
23438 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
23439 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
23440 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
23441 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
23442 BUILT_IN_FEUPDATEENV): New builtins.
23443 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
23444 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
23445 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
23446 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
23448 (builtin_structptr_types): Adjust size.
23449 * tree.c (builtin_structptr_types): Add four entries.
23451 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23452 Olivier Hainque <hainque@adacore.com>
23454 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
23455 (TLS_SYM): New local macro, forcing reference to __tls__ on
23456 link command lines for VxWorks 7 RTPs, triggering initialization
23458 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
23459 OS features TLS support, true for RTPs on VxWorks 7.
23460 * config/vxworks.c (vxworks_override_options): Setup emutls
23463 2017-06-27 Jakub Jelinek <jakub@redhat.com>
23465 * predict.c (test_prediction_value_range): Use -1U instead of -1
23466 to avoid narrowing conversion warning.
23467 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
23468 to avoid narrowing conversion warning.
23469 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
23471 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
23473 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23475 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
23476 64bit configurations.
23477 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
23478 (SIZE_TYPE): Likewise.
23479 * config/vxworks.c (vxworks_emutls_var_fields): Use
23480 long_unsigned_type_node instead of unsigned_type_node as the offset
23481 field type, which is "pointer" mode in emutls.c.
23483 2017-06-27 Jakub Jelinek <jakub@redhat.com>
23486 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
23488 PR middle-end/81207
23489 * gimple-fold.c (replace_call_with_call_and_fold): Handle
23490 gimple_vuse copying separately from gimple_vdef copying.
23492 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23494 * value-prof.c (free_hist): Remove call to memset and the enclosing if
23497 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
23498 Olivier Hainque <hainque@adacore.com>
23500 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
23501 for all vxworks7 targets.
23502 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
23503 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
23504 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
23505 variations for VX6/VX7 and 32/64bits later on in ...
23506 (VXWORKS_LIB_SPEC): Leverage new macros.
23507 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
23508 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
23510 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
23512 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
23513 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
23515 2017-06-26 Carl Love <cel@us.ibm.com>
23517 * config/rs6000/rs6000-c.c: Add support for built-in functions
23518 vector bool char vec_reve (vector bool char);
23519 vector signed char vec_reve (vector signed char);
23520 vector unsigned char vec_reve (vector unsigned char);
23521 vector bool int vec_reve (vector bool int);
23522 vector signed int vec_reve (vector signed int);
23523 vector unsigned int vec_reve (vector unsigned int);
23524 vector bool long long vec_reve (vector bool long long);
23525 vector signed long long vec_reve (vector signed long long);
23526 vector unsigned long long vec_reve (vector unsigned long long);
23527 vector bool short vec_reve (vector bool short);
23528 vector signed short vec_reve (vector signed short);
23529 vector double vec_reve (vector double);
23530 vector float vec_reve (vector float);
23531 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
23532 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
23533 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
23534 (altivec_vreve): New pattern.
23535 * config/rs6000/altivec.h (vec_reve): New define.
23536 * doc/extend.texi (vec_rev): Update the built-in documentation file
23537 for the new built-in functions.
23539 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23541 PR tree-optimization/71815
23542 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
23544 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
23546 (slsr_process_phi): Likewise.
23547 (replace_uncond_cands_and_profitable_phis): Don't replace a
23548 multiply candidate with a stride of 1 (copy or cast).
23549 (phi_incr_cost): Call uses_consumed_by_stmt rather than
23551 (lowest_cost_path): Likewise.
23552 (total_savings): Likewise.
23554 2017-06-26 Richard Biener <rguenther@suse.de>
23557 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
23558 Use def_builtin_pure for all gather builtins.
23560 2017-06-26 Richard Biener <rguenther@suse.de>
23562 PR tree-optimization/81203
23563 * tree-tailcall.c (find_tail_calls): Do not move stmts into
23564 non-dominating BBs.
23566 2017-06-26 Marek Polacek <polacek@redhat.com>
23569 * doc/invoke.texi: Document -Wmultistatement-macros.
23571 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
23573 * doc/sourcebuild.texi (ARM-specific attributes): Document new
23574 arm_neon_ok_no_float_abi effective target.
23576 2017-06-26 Richard Biener <rguenther@suse.de>
23578 PR tree-optimization/80928
23579 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
23580 (copy_bbs): Set BB_DUPLICATED flag early.
23581 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
23583 (execute_on_shrinking_pred): Likewise.
23584 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
23585 BB_DUPLICATED blocks.
23586 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
23587 iterate over all PHIs considering removal of *gsi.
23589 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
23591 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
23594 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23596 * config/rs6000/rs6000-string.c: (expand_block_clear,
23597 do_load_for_compare, select_block_compare_mode,
23598 compute_current_alignment, expand_block_compare,
23599 expand_strncmp_align_check, expand_strn_compare,
23600 expand_block_move, rs6000_output_load_multiple)
23601 Move functions related to string/block move/compare
23602 to a separate file.
23603 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
23604 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
23605 for this function which is now used in two files.
23606 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
23607 * config.gcc: Add rs6000-string.o to extra_objs for
23608 targets powerpc*-*-* and rs6000*-*-*.
23610 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
23613 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
23614 32-bit, since indexed is not valid for DImode.
23615 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
23616 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
23617 (define_peephole2 for Altivec d-form load): Add 32-bit support.
23618 (define_peephole2 for Altivec d-form store): Likewise.
23621 * multiple_target.c (create_dispatcher_calls): Only create the
23622 dispatcher call if the function is the default clone of a
23623 versioned function.
23625 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
23627 PR middle-end/80902
23628 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
23629 a call, force the call to not be a tail call.
23631 2017-06-23 Jeff Law <law@redhat.com>
23633 * doc/contrib.texi: Add entry for Steven Pemberton's work on
23636 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
23638 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
23639 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
23640 handling for early expansion of vector shifts (sl,sr,sra,rl).
23641 (builtin_function_type): Add vector shift right instructions
23642 to the unsigned argument list.
23644 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
23646 rtl-optimizatoin/79286
23647 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
23648 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
23649 trap. PIC register plus a const unspec without offset can never trap.
23651 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
23653 * tree.h (builtin_structptr_type): New type.
23654 (builtin_structptr_types): Declare new array.
23655 * tree.c (builtin_structptr_types): New array.
23656 (free_lang_data, build_common_tree_nodes): Use it.
23658 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
23661 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
23664 2017-06-22 Matt Turner <mattst88@gmail.com>
23666 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
23667 Lake models to skylake case. Assume skylake for unknown
23668 models with clflushopt.
23670 2017-06-22 Jeff Law <law@redhat.com>
23672 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
23673 frame sizes that do not satisfy aarch64_uimm12_shift.
23675 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
23677 * profile-count.h (apply_probability,
23678 apply_scale, probability_in): Fix checks for zero.
23680 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23682 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
23683 * doc/cppdiropts.texi (-I @var{dir}): Document it.
23685 2016-06-22 Richard Biener <rguenther@suse.de>
23687 * tree-vect-loop.c (vect_model_reduction_cost): Handle
23688 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
23689 REDUC_MAX_EXPR support.
23690 (vectorizable_reduction): Likewise.
23691 (vect_create_epilog_for_reduction): Likewise.
23693 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
23695 * match.pd (A / (1 << B) -> A >> B): New.
23696 * generic-match-head.c: Include optabs-tree.h.
23697 * gimple-match-head.c: Likewise.
23698 * optabs-tree.h (target_supports_op_p): New.
23699 * optabs-tree.c (target_supports_op_p): New.
23701 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23703 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
23704 $gcc_cv_ld --help output.
23705 (gcc_cv_ld_demangle): Likewise.
23706 (gcc_cv_ld_eh_frame_hdr): Likewise.
23707 (gcc_cv_ld_pie): Likewise.
23708 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
23709 (gcc_cv_ld_buildid): Likewise.
23710 (gcc_cv_ld_sysroot): Likewise.
23711 (ld_bndplt_support): Likewise.
23712 (ld_pushpopstate_support): Likewise.
23713 * configure: Regenerate.
23714 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
23716 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23719 * config/i386/sse.md (round<mode>2): Renumber match_dup and
23720 operands indexes to avoid gap between operands and match_dups.
23722 2017-06-21 Andrew Pinski <apinski@cavium.com>
23724 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
23725 Increment Arith_shift and Arith_shift_reg by 1.
23726 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
23728 * config/aarch64/aarch64.c (thunderx_tunings): Enable
23729 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
23730 (aarch64_strip_extend): Add new argument and test for it.
23731 (aarch64_cheap_mult_shift_p): New function.
23732 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
23733 add a cost if it is true.
23734 Update calls to aarch64_strip_extend.
23735 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
23737 2017-06-21 Andrew Pinski <apinski@cavium.com>
23739 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
23741 (thunderxt88): Likewise.
23742 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
23743 (thunderx_prefetch_tune): New variable.
23744 (thunderx2t99_prefetch_tune): Update for the correct values.
23745 (thunderxt88_tunings): New variable.
23746 (thunderx_tunings): Use thunderx_prefetch_tune instead of
23747 generic_prefetch_tune.
23748 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
23750 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23752 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
23753 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
23754 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
23755 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
23756 (aarch64_atomic_cas<mode>, GPI): Likewise.
23758 2017-06-21 Martin Liska <mliska@suse.cz>
23760 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
23761 statements on cold and hot labels.
23762 * predict.c (tree_estimate_probability_bb): Remove the
23763 prediction from this place.
23765 2017-06-21 Martin Liska <mliska@suse.cz>
23767 PR tree-optimization/79489
23768 * gimplify.c (maybe_add_early_return_predict_stmt): New
23770 (gimplify_return_expr): Call the function.
23771 * predict.c (tree_estimate_probability_bb): Remove handling
23773 * predict.def: Update comment about early return predictor.
23774 * gimple-predict.h (is_gimple_predict): New function.
23775 * predict.def: Change default value of early return to 66.
23776 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
23778 * passes.def: Put pass_strip_predict_hints to the beginning of
23781 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
23783 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
23784 FUNCTION_DECL declarations.
23785 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
23787 (dwaf2out_decl): Likewise.
23788 * godump.c (go_early_global_decl): Skip call to the real debug hook
23789 for FUNCTION_DECL declarations.
23790 * passes.c (rest_of_decl_compilation): Skip call to the
23791 early_global_decl debug hook for FUNCTION_DECL declarations, unless
23792 -fdump-go-spec is passed.
23794 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23796 * config/i386/i386.c (struct builtin_isa): New field pure_p.
23797 Reorder for compactness.
23798 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
23799 (def_builtin_pure, def_builtin_pure2): New functions.
23800 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
23802 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23804 * match.pd (nop_convert): New predicate.
23805 ((A +- CST1) +- CST2): Allow some NOP conversions.
23807 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23810 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
23811 with ctors/dtors if GOVD_SHARED is set.
23813 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23815 * config/aarch64/aarch64.md (movti_aarch64):
23816 Emit mov rather than orr.
23817 (movtf_aarch64): Likewise.
23818 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
23819 Emit mov rather than orr.
23821 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23823 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
23824 Swap alternatives, make integer dup more expensive.
23826 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23828 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
23829 Return true for non-tls symbols.
23831 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
23833 * config/aarch64/aarch64-cores.def (cortex-a55): New.
23834 (cortex-a75): Likewise.
23835 (cortex-a75.cortex-a55): Likewise.
23836 * config/aarch64/aarch64-tune.md: Regenerate.
23837 * doc/invoke.texi (-mtune): Document new values for -mtune.
23839 2017-06-21 Tom de Vries <tom@codesourcery.com>
23841 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
23842 stack_size feature.
23843 (Effective-Target Keywords, Other attributes): Suggest using
23844 dg-add-options stack_size feature to get stack limit in stack_size
23845 effective target documentation.
23847 2017-06-21 Julian Brown <julian@codesourcery.com>
23848 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23850 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
23851 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
23852 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
23854 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
23855 attribute type list for neon_multiply.
23856 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
23857 attribute type list for neon_multiply.
23858 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
23859 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
23860 attribute type list for neon_multiply.
23861 * config/arm/types.md (crypto_pmull): Add.
23862 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
23863 attribute type list.
23865 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
23867 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
23870 2017-06-20 Jakub Jelinek <jakub@redhat.com>
23872 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
23873 to make sure not to dereference a NULL cost_classes_ptr pointer.
23875 2017-06-20 Carl Love <cel@us.ibm.com>
23877 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23878 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23879 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23880 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23881 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23882 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23883 VMULOSW): New enum "unspec" values.
23884 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23885 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23886 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23887 altivec_vmulosw): New patterns.
23888 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23889 VMULOSW): Add definitions.
23891 2017-06-20 Julia Koval <julia.koval@intel.com>
23893 * config/i386/i386.c: Fix rounding expand for new pattern.
23894 * config/i386/subst.md: Fix pattern (parallel -> unspec).
23896 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23898 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
23899 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
23901 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23903 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
23906 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23908 * config/aarch64/aarch64-cores.def: Rearrange to sort by
23909 architecture, then by implementer ID.
23910 * config/aarch64/aarch64-tune.md: Regenerate.
23912 2017-06-20 Richard Biener <rguenther@suse.de>
23914 PR middle-end/81097
23915 * fold-const.c (split_tree): Fold to type before negating.
23917 2017-06-20 David Malcolm <dmalcolm@redhat.com>
23919 * diagnostic-show-locus.c
23920 (selftest::test_fixit_deletion_affecting_newline): New function.
23921 (selftest::diagnostic_show_locus_c_tests): Call it.
23923 2017-06-20 Andreas Schwab <schwab@suse.de>
23926 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
23929 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23931 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
23932 __ARM_FEATURE_COPROC according to support.
23934 2017-06-20 Jakub Jelinek <jakub@redhat.com>
23936 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
23937 Rewritten to avoid overflow for > 32-bit pointers.
23940 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
23941 by removing enum keyword.
23942 (ubsan_type_descriptor): Likewise. Formatting fix.
23945 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
23946 splitter): Require TARGET_SSE2 in the condition.
23948 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
23951 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
23952 for doing vector set of SFmode on ISA 3.0.
23953 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
23954 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
23956 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
23957 SFmode value into a V4SF variable that was extracted from another
23958 V4SF variable without converting the element to double precision
23959 and back to single precision vector format.
23960 (vsx_insert_extract_v4sf_p9_2): Likewise.
23962 2017-06-19 Jakub Jelinek <jakub@redhat.com>
23964 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
23965 in UWHI to avoid undefined overflow.
23968 * ubsan.h (enum ubsan_encode_value_phase): New.
23969 (ubsan_encode_value): Change second argument to
23970 enum ubsan_encode_value_phase with default value of
23971 UBSAN_ENCODE_VALUE_GENERIC.
23972 * ubsan.c (ubsan_encode_value): Change second argument to
23973 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
23974 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
23975 create_tmp_var_raw instead of create_tmp_var and use a
23977 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
23978 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
23979 ubsan_encode_value callers.
23982 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
23983 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
23984 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
23986 2017-06-19 Richard Biener <rguenther@suse.de>
23988 PR middle-end/81118
23989 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
23990 estimates if we changed anything.
23992 2017-06-19 Richard Biener <rguenther@suse.de>
23994 PR tree-optimization/80887
23995 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
23996 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
23997 simplified lookups, then reset mprts_hook.
23998 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
24000 (try_to_simplify): Likewise.
24002 2017-06-19 Martin Liska <mliska@suse.cz>
24005 * gimplify.c (gimplify_switch_expr):
24006 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
24008 2017-06-19 Martin Liska <mliska@suse.cz>
24010 * doc/install.texi: Document that PGO runs in 4 stages.
24012 2017-06-19 Martin Liska <mliska@suse.cz>
24015 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
24016 to dispatcher function name.
24017 * multiple_target.c (replace_function_decl): New function.
24018 (create_dispatcher_calls): Redirect both edges and references.
24020 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
24022 * profile-count.c (profile_count::dump): Dump quality.
24023 (profile_count::differs_from_p): Update for unsigned val.
24024 * profile-count.h (profile_count_quality): New enum.
24025 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
24027 2017-06-19 Richard Biener <rguenther@suse.de>
24029 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
24030 struct function as arg.
24031 (estimate_numbers_of_iterations): Export overload with loop arg.
24032 (free_numbers_of_iterations_estimates_loop): Use an overload of
24033 free_numbers_of_iterations_estimates instead.
24034 * tree-cfg.c (remove_bb): Adjust.
24035 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
24036 * tree-parloops.c (gen_parallel_loop): Likewise.
24037 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
24039 (tree_unroll_loops_completely): Likewise.
24040 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
24041 Use an overload instead and export.
24042 (estimated_loop_iterations): Adjust.
24043 (max_loop_iterations): Likewise.
24044 (likely_max_loop_iterations): Likewise.
24045 (estimate_numbers_of_iterations): Take struct function as arg
24047 (loop_exits_before_overflow): Adjust.
24048 (free_numbers_of_iterations_estimates_loop): Use an overload.
24049 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
24050 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
24052 2017-06-19 Richard Biener <rguenther@suse.de>
24055 * ipa-prop.c (find_constructor_constant_at_offset): Handle
24056 RANGE_EXPR conservatively.
24058 2017-06-16 Carl Love <cel@us.ibm.com>
24060 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24061 definitions for vec_float, vec_float2, vec_floato,
24062 vec_floate built-ins.
24063 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
24064 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
24066 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
24067 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
24068 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
24069 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
24070 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
24071 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
24072 vec_floato): Add builtin defines.
24073 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
24074 Update the built-in documentation file for the new built-in
24077 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24079 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
24080 (mthumb): Mark as the negative of -marm.
24082 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24084 * doc/invoke.texi (ARM Options, -mcpu): Document supported
24086 (ARM Options, -mtune): Document that this accepts the same
24087 extension options as -mcpu.
24088 (ARM Options, -mfpu): Document addition of -mfpu=auto.
24090 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24092 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
24093 permitted extensions.
24095 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24097 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
24098 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
24099 (armv8-m.main): Add option +nodsp.
24100 * config/arm/arm-cpu-cdata.h: Regenerated.
24102 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24104 * config/arm/t-fuchsia: New file.
24105 * config.gcc (arm*-*-fuchsia*): Use it.
24107 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24109 * config/arm/t-symbian: Rewrite for new option infrastructure.
24111 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24113 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
24114 (MULTILIB_REQUIRED): Likewise.
24116 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24118 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
24119 (MULTILIB_RESUE): Likewise.
24120 (MULTILIB_MATCHES): Likewise.
24121 (MULTLIB_REQUIRED): Likewise.
24123 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24125 * config/arm/t-rtems: Rewrite for new option framework.
24127 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24129 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
24130 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
24131 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
24132 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
24133 * config/arm/t-multilib: ... here.
24134 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
24135 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
24136 armv7-a and armv8*-a when A-profile libraries have not been built.
24137 * config/arm/t-rmprofile: Rewrite.
24139 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24141 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
24142 with a backslash. Remove the backslash after substituting unescaped
24144 * doc/fragments.texi (MULTILIB_REUSE): Document it.
24146 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24148 * config.gcc: (arm*-*-*): When building a-profile libraries, force
24149 the driver to pass through the default setting of -mfloat-abi.
24150 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
24152 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
24153 (all_feat_combs): New rule.
24154 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
24156 * config/arm/t-aprofile: Rewrite.
24158 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24160 * config/arm/arm.h (FPUTYPE_AUTO): Define.
24161 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
24162 fpu is not specified by the user/command-line.
24163 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
24164 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
24165 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
24166 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
24167 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
24168 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
24170 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24172 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
24173 * config/arm/t-arm-elf: Rewritten.
24175 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24177 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
24178 have some floating-point instructions.
24179 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
24180 (TARGET_MAYBE_HARD_FLOAT): New macro.
24181 * config/arm/arm-builtins.c (arm_init_builtins): Use
24182 TARGET_MAYBE_HARD_FLOAT.
24183 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
24185 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24187 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
24188 (configargs.h): Include it.
24189 (arm_print_hint_for_fpu_option): New function.
24190 (arm_parse_fpu_option): New function.
24191 (candidate_extension): New class.
24192 (arm_canon_for_multilib): New function.
24193 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
24194 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
24195 (ARCH_CANONICAL_SPECS): New macro.
24196 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
24198 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24200 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
24201 are set after handling multilib fragments. Set target_cpu_default2
24204 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24206 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
24208 (arm*-*-*): Set target_cpu_default2 to a quoted string.
24209 * config/arm/parsecpu.awk (check_cpu): Validate any extension
24211 (check_arch): Likewise.
24212 * config/arm/arm.c (arm_configure_build_target): Handle
24213 TARGET_CPU_DEFAULT being a string constant. Scan any feature
24214 options in the default.
24216 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24218 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
24219 when an option is an alias of another.
24220 * config/arm/parsecpu.awk (optalias): New parser token.
24221 (gen_comm_data): Mark non-alias options as such. Emit entries
24222 for extension aliases.
24223 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
24224 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
24225 (armv6kz, armv6zk, armv6t2): Likewise.
24226 (armv7): Make vfpv3-d16 an alias.
24227 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
24229 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
24230 Sort in canonical order.
24231 (armv8-a): Sort in canonical order.
24232 (armv8.1-a, armv8.2-a): Likewise.
24233 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
24235 (cortex-a9): Sort in canonical order.
24236 * config/arm/arm.c (selftests.h): Include it.
24237 (arm_test_cpu_arch_data): New function.
24238 (arm_run_self_tests): New function.
24239 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
24240 (targetm): Move declaration to the end of the file.
24241 * arm-cpu-cdata.h: Regenerated.
24243 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24245 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
24246 call to target_mode_check describing the type of option passed.
24247 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
24248 (arm_target_thumb_only): Use arm_parse_arch_option_name or
24249 arm_parse_cpu_option_name to match parameters against list of
24251 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
24252 arm_arch_core_flags data structure.
24253 * config/arm/arm-cpu_cdata.h: Regenerated.
24255 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24257 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
24259 (arm_print_hint_for_cpu_option): Likewise.
24260 (arm_print_hint_for_arch_option): Likewise.
24261 (arm_parse_cpu_option_name): Likewise.
24262 (arm_parse_arch_option_name): Likewise.
24263 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
24264 of entries in the all_fpus list.
24265 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
24266 (arm_parse_cpu_option_name): Declare.
24267 (arm_parse_arch_option_name): Declare.
24268 (arm_parse_option_features): Declare.
24269 (arm_intialize_isa): Declare.
24270 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
24272 (gen_comm_data): ... here. Make definitions non-static.
24273 * config/arm/arm-cpu-data.h: Regenerated.
24274 * config/arm/arm-cpu-cdata.h: Regenerated.
24276 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24278 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
24279 (cpu_arch_extension): New structure.
24280 (cpu_arch_option, arch_option, cpu_option): New structures.
24281 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
24282 architecture types.
24283 (gen_data): Generate new format data tables.
24284 * config/arm/arm.c (cpu_tune): New structure.
24285 (cpu_option, processors): Delete.
24286 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
24287 (arm_print_hint_for_cpu_option): ... this and ...
24288 (arm_print_hint_for_arch_option): ... this.
24289 (arm_parse_arch_cpu_name): Delete. Replace with ...
24290 (arm_parse_cpu_option_name): ... this and ...
24291 (arm_parse_arch_option_name): ... this.
24292 (arm_unrecognized_feature): Change type of target parameter to
24294 (arm_parse_arch_cpu_features): Delete. Replace with ...
24295 (arm_parse_option_features): ... this.
24296 (arm_configure_build_target): Rework to use new configuration data
24298 (arm_print_tune_info): Rework for new configuration data tables.
24299 * config/arm/arm-cpu-data.h: Regenerated.
24300 * config/arm/arm-cpu.h: Regenerated.
24302 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24304 * Makefile.in (OBJS): Move sbitmap.o from here ...
24305 (OBJS-libcommon): ... to here.
24307 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24309 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
24310 (ISA_ALL_CRYPTO): New macro.
24311 (ISA_ALL_SIMD): New macro
24312 (ISA_ALL_FP): New macro.
24313 * config/arm/arm.c (fpu_bitlist): Update initializer.
24314 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
24316 (arm9e): Add fpu. Add option for nofp
24317 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
24318 (arm926ej-s, arm1026ej-s): Likewise.
24319 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
24320 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
24321 neon-fp16, neon-vfpv4, nofp and nosimd.
24322 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
24323 (cortex-a8): Add fpu. Add option for nofp.
24324 (cortex-a9): Add fpu. Add options for nosimd and nofp.
24325 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
24326 (cortex-r4f): Add fpu.
24327 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
24328 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
24330 (cortex-r8): Likewise.
24331 (cortex-m4): Add fpu. Add option for nofp.
24332 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
24333 (cortex-a17.cortex-a7): Likewise.
24334 (cortex-a32): Add fpu. Add options for crypto and nofp.
24335 (cortex-a35, cortex-a53): Likewise.
24336 (cortex-a57): Add fpu. Add option for crypto.
24337 (cortex-a72, cortex-a73): Likewise.
24338 (exynos-m1): Likewise.
24339 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
24340 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
24341 (cortex-m33): Add fpu. Add option for nofp.
24342 * config/arm/arm-cpu-cdata.h: Regenerated
24343 * config/arm/arm-cpu-data.h: Regenerated.
24345 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24347 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
24348 (armv5te, armv5tej): Likewise.
24349 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
24350 (armv7): Add options fp and vfpv3-d16.
24351 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
24352 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
24354 (armv7ve): Likewise.
24355 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
24356 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
24357 (armv8-a): Add nocrypto option.
24358 (armv8.1-a, armv8.2-a): Likewise.
24359 (armv8-m.main): add options fp, fp.dp and nofp.
24361 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24363 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
24365 (armv8-a+crc): Delete.
24366 (armv8.1-a): Add options simd, crypto and nofp.
24367 (armv8.2-a): Add options fp16, simd, crypto and nofp.
24368 (armv8.2-a+fp16): Delete.
24369 (armv8-m.main): Add option dsp.
24370 (armv8-m.main+dsp): Delete.
24371 (cortex-a8): Add fpu. Add nofp option.
24372 (cortex-a9): Add fpu. Add nofp and nosimd options.
24373 * config/arm/parsecpu.awk (gen_data): Generate option tables and
24374 link to main cpu and architecture data structures.
24375 (gen_comm_data): Only put isa attributes from the main architecture
24377 (option): New statement for architecture and CPU entries.
24378 * arm.c (struct cpu_option): New structure.
24379 (struct processors): Add entry for options.
24380 (arm_unrecognized_feature): New function.
24381 (arm_parse_arch_cpu_name): Ignore any characters after the first
24383 (arm_parse_arch_cpu_feature): New function.
24384 (arm_configure_build_target): Separate out any CPU and architecture
24385 features and parse separately. Don't error out if -mfpu=auto is
24386 used with only an architecture string.
24387 (arm_print_asm_arch_directives): New function.
24388 (arm_file_start): Call it.
24389 * config/arm/arm-cpu-cdata.h: Regenerated.
24390 * config/arm/arm-cpu-data.h: Likewise.
24391 * config/arm/arm-tables.opt: Likewise.
24393 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24395 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
24396 assembler when it is not -mfpu=auto.
24398 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24400 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
24401 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
24402 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
24403 (ASM_CPU_SPEC): Rewrite.
24404 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
24405 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
24406 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
24407 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
24408 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
24409 copied string is NUL-terminated. Also strip any characters prefixed
24411 (arm_rewrite_selected_arch): New function.
24412 (arm_rewrite_march): New function.
24414 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24416 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
24417 (x_arm_cpu_string, x_arm_tune_string): Likewise.
24418 (march, mcpu, mtune): Convert to string-based options.
24419 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
24420 (arm_parse_arch_cpu_name): New function.
24421 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
24422 identify selected architecture or CPU.
24423 (arm_option_save): New function.
24424 (TARGET_OPTION_SAVE): Redefine.
24425 (arm_option_restore): Restore string options.
24426 (arm_option_print): Print string options.
24428 2017-06-16 Martin Sebor <msebor@redhat.com>
24430 PR tree-optimization/80933
24431 PR tree-optimization/80934
24432 * builtins.c (fold_builtin_3): Do not handle bcmp here.
24433 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
24434 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
24435 (gimple_fold_builtin): Call them.
24437 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24439 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
24440 as unlikely; update profile.
24442 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24444 * predict.c (force_edge_cold): Handle declaring edges impossible
24447 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24449 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
24451 (try_unroll_loop_completely): Fix reporting.
24453 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24455 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
24457 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
24460 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
24461 if given a non-constant argument for an intrinsic which requires a
24464 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24466 * profile.c (compare_freqs): New function.
24467 (branch_prob): Sort edge list.
24468 (find_spanning_tree): Assume that the list is priority sorted.
24470 2017-06-16 Richard Biener <rguenther@suse.de>
24472 PR tree-optimization/81090
24473 * passes.def (pass_record_bounds): Remove.
24474 * tree-pass.h (make_pass_record_bounds): Likewise.
24475 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
24476 make_pass_record_bounds): Likewise.
24477 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
24478 not free niter estimates at the beginning but at the end.
24479 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
24481 2017-06-16 Richard Biener <rguenther@suse.de>
24483 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
24484 initializer to workaround ICE in host GCC 4.8.
24486 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24488 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
24490 (clone_inlined_nodes): Update.
24492 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24494 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
24495 prefetch settings, and enable prefetching by default at -O3.
24497 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24499 * config/aarch64/aarch64.c (aarch64_override_options_internal):
24500 Set flag_prefetch_loop_arrays according to tuning data.
24502 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24504 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
24505 New tune structure.
24506 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
24507 [Unrelated to main purpose of the patch] Place the pointer field last
24508 to enable type checking errors when tune structure are wrongly merged.
24509 * config/aarch64/aarch64.c (generic_prefetch_tune,)
24510 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
24511 (thunderx2t99_prefetch_tune): New tune constants.
24512 (tune_params *_tunings): Update all tunings (no functional change).
24513 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
24514 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
24515 from tunings structures.
24517 2017-06-16 Jakub Jelinek <jakub@redhat.com>
24520 * ubsan.c (instrument_null): Add T argument, use it instead
24521 of computing it based on IS_LHS.
24522 (instrument_object_size): Likewise.
24523 (pass_ubsan::execute): Adjust instrument_null and
24524 instrument_object_size callers to pass gimple_get_lhs or
24525 gimple_assign_rhs1 result to it. Use instrument_null instead of
24526 calling get_base_address and instrument_mem_ref. Handle
24527 aggregate call arguments for object-size sanitization.
24529 2017-06-16 Yury Gribov <tetra2005@gmail.com>
24531 PR tree-optimization/81089
24532 * tree-vrp.c (is_masked_range_test): Validate operands of
24535 2017-06-15 Martin Sebor <msebor@redhat.com>
24538 * dumpfile.c (dump_register): Avoid calling memset to initialize
24539 a class with a default ctor.
24540 * gcc.c (struct compiler): Remove const qualification.
24541 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
24542 * hash-table.h: Ditto.
24543 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
24545 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
24546 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
24548 * params.h (struct param_info): Make struct members non-const.
24549 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
24550 with default initialization.
24551 * vec.h (vec_copy_construct, vec_default_construct): New helper
24553 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
24554 with vec_copy_construct.
24555 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
24556 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
24557 * doc/invoke.texi (-Wclass-memaccess): Document.
24559 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24561 * emit-rtl.h (is_leaf): Update comment about local
24562 register allocator.
24564 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
24567 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
24568 for a variable to have a section before checking if the section has a
24570 Set section to.persistent if persistent attribute is set.
24571 Warn if .persistent attribute is used on an automatic variable.
24573 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
24575 PR rtl-optimization/80474
24576 * reorg.c (update_block): Do not ignore instructions in a delay slot.
24578 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
24580 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
24583 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
24585 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
24586 (casesi): Emit bounds checking as RTL.
24587 (casesi_internal_mips16_<mode>): Remove bounds checking.
24589 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
24591 * config/xtensa/xtensa.c (xtensa_option_override): Append
24592 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
24593 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
24594 xtensa_doloop_hooks): Define unconditionally.
24595 (xtensa_reorg_loops): Only call reorg_loops in the presence of
24597 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
24598 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
24599 for it in xtensa_option_override.
24600 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
24601 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
24603 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
24605 * doc/cppopts.texi: Document '-' special value to -MF.
24607 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
24609 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
24610 (cortex_a53_fconst): Likewise.
24611 (cortex_a53_fpmul): Likewise.
24612 (cortex_a53_f_load_64): Likewise.
24613 (cortex_a53_f_load_many): Likewise.
24614 (cortex_a53_advsimd_alu): Likewise.
24615 (cortex_a53_advsimd_alu_q): Likewise.
24616 (cortex_a53_advsimd_mul): Likewise.
24617 (cortex_a53_advsimd_mul_q): Likewise.
24618 (fpmac bypass): Add new bypass for fpmac-fpmac case.
24619 Add missing fmul, r2f_cvt and fconst cases.
24621 2017-06-14 Richard Biener <rguenther@suse.de>
24623 PR middle-end/81088
24624 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
24626 (fold_binary_loc): When associating do not treat pre-existing
24627 TREE_OVERFLOW on literal constants as a reason to allow
24628 TREE_OVERFLOW on associated literal constants.
24630 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24632 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
24633 (MASK_FEATURES): New macro.
24634 * config/sparc/sparc.c (sparc_option_override): Remove the special
24635 handling of -mfpu and generalize it to all MASK_FEATURES switches.
24637 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24639 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
24640 a division of 0 if non-call exceptions are enabled.
24642 2017-06-14 Andrew Pinski <apinski@cavium.com>
24643 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24646 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
24647 Improve vector initialization code gen for only variable case.
24649 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24651 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
24653 2017-06-14 Richard Biener <rguenther@suse.de>
24655 PR tree-optimization/81083
24656 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
24659 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24661 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
24662 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
24663 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
24664 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
24665 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
24666 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
24668 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24670 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
24671 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
24673 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24675 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
24677 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24679 * config/rs6000/t-rtems: Don't handle SPE.
24681 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24683 * config/rs6000/t-linux: Don't handle SPE.
24685 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24687 * config/rs6000/eabispe.h: Delete file.
24689 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24691 * config/rs6000/t-spe: Delete file.
24693 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24695 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
24696 (rs6000_legitimate_offset_address_p): Return false for anything in
24697 V2SImode or V2SFmode.
24699 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24701 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
24702 except V2SF and V2SI. Rearrange the vector modes, and add comments.
24703 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
24705 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
24706 (rs6000_legitimate_offset_address_p): Ditto.
24707 (rs6000_emit_move): Ditto.
24708 (rs6000_init_builtins): Remove V4HI_type_node.
24710 2017-06-13 Martin Liska <mliska@suse.cz>
24713 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
24714 (gate_asan): Likewise.
24715 * asan.h (asan_no_sanitize_address_p): Remove the function.
24716 (sanitize_flags_p): New function.
24717 * builtins.def: Fix coding style.
24718 * common.opt: Use renamed enum value.
24719 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
24720 * doc/extend.texi: Document no_sanitize attribute.
24721 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
24722 to SANITIZE_UNDEFINED_NONDEFAULT.
24723 * gcc.c (sanitize_spec_function): Use the renamed enum value.
24724 * gimple-fold.c (optimize_atomic_compare_exchange_p):
24725 Use sanitize_flags_p.
24726 * gimplify.c (gimplify_function_tree): Likewise.
24727 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
24728 * opts.c (parse_no_sanitize_attribute): New function.
24729 (common_handle_option): Use renamed enum value.
24730 * opts.h (parse_no_sanitize_attribute): Declare.
24731 * tree.c (sanitize_flags_p): New function.
24732 * tree.h: Declared here.
24733 * tsan.c: Use sanitize_flags_p.
24734 * ubsan.c (ubsan_expand_null_ifn): Likewise.
24735 (instrument_mem_ref): Likewise.
24736 (instrument_bool_enum_load): Likewise.
24737 (do_ubsan_in_current_function): Remove the function.
24738 (pass_ubsan::execute): Use sanitize_flags_p.
24739 * ubsan.h: Remove do_ubsan_in_current_function
24740 * tree-cfg.c (print_no_sanitize_attr_value): New function.
24741 (dump_function_to_file): Use it here.
24743 2017-06-13 Martin Jambor <mjambor@suse.cz>
24745 PR tree-optimization/80803
24746 PR tree-optimization/81063
24747 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
24748 (propagate_subaccesses_across_link): Enqueue subtree whenever
24749 necessary instead of relying on the caller.
24751 2017-06-13 Martin Jambor <mjambor@suse.cz>
24753 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
24754 that have a first_link.
24755 (sort_and_splice_var_accesses): Do not check first_link before
24757 (subtree_mark_written_and_enqueue): Likewise.
24758 (propagate_all_subaccesses): Likewise and do not stop at first
24759 parent with a first_link.
24761 2017-06-13 Martin Jambor <mjambor@suse.cz>
24763 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
24766 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24768 * match.pd: New pattern.
24770 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24772 * tree-vrp.c (is_masked_range_test): New function.
24773 (register_edge_assert_for): Determine ranges for
24776 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24778 PR tree-optimization/67328
24779 * fold-const.c (maskable_range_p): New function.
24780 (build_range_check): Generate bittests if possible.
24782 2017-06-13 Martin Liska <mliska@suse.cz>
24784 * gimple-pretty-print.c (dump_probability): Add new argument.
24785 (dump_edge_probability): Dump both probability and count.
24786 (dump_gimple_label): Likewise.
24787 (dump_gimple_bb_header): Likewise.
24789 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
24792 * config/avr/avr-devices.c: Fix indentation.
24793 * config/avr/gen-avr-mmcu-specs.c: Dito.
24795 2017-06-13 Richard Biener <rguenther@suse.de>
24797 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
24798 instead get vector type from stmt_info.
24799 (vectorizable_reduction): Adjust. Remove dead code.
24801 2017-06-13 Richard Biener <rguenther@suse.de>
24803 PR middle-end/81065
24804 * fold-const.c (extract_muldiv_1): Remove bogus distribution
24805 case of C * (x * C2 + C3).
24806 (fold_addr_of_array_ref_difference): Properly fold index difference.
24808 2017-06-12 David S. Miller <davem@davemloft.net>
24811 * config/sparc/sparc.md (return expander): Emit frame blockage if
24812 function uses alloca.
24814 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
24816 * combine.c (make_field_assignment): Check len rather than the mode
24817 precision when calling force_to_mode.
24819 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
24821 Support multilibs and devices that see flash in RAM address range.
24824 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
24825 (avr_mcu_t) <flash_pm_offset>: New field.
24826 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
24827 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
24828 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
24829 (AVR_TINY_PM_OFFSET): Remove macro.
24830 * config/avr/avr.opt (-mshort-calls): New option.
24831 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24832 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
24833 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
24834 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
24835 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
24836 instead of avr_arch->have_jmp_call.
24837 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
24838 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
24839 avr_arch->flash_pm_offset to define.
24840 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
24841 new field flash_pm_offset. Add entry for avrxmega3.
24842 (avr_texinfo): Add entry for avrxmega3.
24843 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
24844 attiny212, attiny214,
24845 attiny412, attiny414, attiny416, attiny417,
24846 attiny814, attiny816, attiny817,
24847 attiny1614, attiny1616, attiny1617,
24848 attiny3214, attiny3216, attiny3217.
24849 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
24850 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
24851 (avr_print_operand_address) [AVR_TINY]: Same.
24852 (avr_asm_init_sections) <readonly_data_section>: Only patch
24853 callback if avr_arch->flash_pm_offset = 0.
24854 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
24855 for rodata if avr_arch->flash_pm_offset != 0.
24856 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
24857 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
24858 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
24859 (m_options): Append opt_rcall.
24860 (m_dirnames): Append dir_rcall.
24861 * config/avr/t-multilib: Regenerate.
24863 * configure.ac [target=avr]: Check whether avrxmega3 default
24864 linker description file works as needed.
24865 * configure: Regenerate.
24866 * doc/avr-mmcu.texi: Regenerate.
24867 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
24868 <__AVR_ARCH__>: Document avrxmega3 and 103.
24869 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
24870 <__AVR_SHORT_CALLS__>: Document it.
24871 <__AVR_PM_BASE_ADDRESS__>: Document it.
24872 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
24873 (AVR Variable Attributes) <progmem>: Document this is
24874 not needed for avrxmega3.
24875 (AVR Named Address Spaces) <__flash>: Dito.
24877 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
24879 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
24881 2017-06-12 Doug Rupp <rupp@adacore.com>
24883 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
24884 Append vxworks-stdint.h to the tm_file list.
24885 * config/vxworks-stdint.h: New file.
24887 2017-06-12 Martin Liska <mliska@suse.cz>
24889 PR tree-optimization/81041
24890 * tree-profile.c (gimple_gen_ic_func_profiler):
24891 Create an extra BB in profile-generate
24892 (gimple_gen_time_profiler): Likewise.
24894 2017-06-12 Jakub Jelinek <jakub@redhat.com>
24896 PR tree-optimization/81003
24897 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
24898 (update_range_test): Use it instead of force_gimple_operand_gsi.
24900 2017-06-12 Richard Biener <rguenther@suse.de>
24902 PR tree-optimization/81053
24903 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
24904 with backedge value not defined in loop. Simplify def stmt
24907 2017-06-11 Tom de Vries <tom@codesourcery.com>
24910 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
24912 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
24913 nvptx_cannot_force_const_mem.
24915 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24917 * opts.c (finish_options): Move test for flag_split_stack after
24918 it has been initialized.
24920 2017-06-11 Jason Merrill <jason@redhat.com>
24922 * tree.h (id_equal): New.
24923 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
24924 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
24925 instead of strcmp of IDENTIFIER_POINTER.
24927 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24929 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
24930 (mark_all_inlined_calls_cdtor): Fix formating.
24931 (inline_transform): Rescale profile before inlining.
24933 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24935 * cgraph.h (cgraph_edge::clone): Update prototype.
24936 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
24937 (cgraph_node::create_clone): Update.
24938 (cgraph_node::create_version_clone): Update.
24939 * tree-inline.c (copy_bb): Update.
24940 (expand_call_inline): Update.
24942 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
24944 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
24945 factored out from ...
24946 (rs6000_emit_prologue): ... here.
24948 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
24950 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
24951 factored out from ...
24952 (rs6000_emit_prologue): ... here.
24954 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24956 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
24958 (handle_missing_profiles): Fix computation of tp_first_run.
24959 (counts_to_freqs): Do not touch freqs when count is 0.
24961 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24963 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
24966 2017-06-10 Tom de Vries <tom@codesourcery.com>
24968 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
24969 attributes): Document signal effective target.
24971 2017-06-10 Tom de Vries <tom@codesourcery.com>
24973 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
24974 Document effective target stack_size.
24976 2017-06-09 David Malcolm <dmalcolm@redhat.com>
24978 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
24979 to the edit_context if they can be auto-applied.
24981 2017-06-9 Ian Lance Taylor <iant@golang.org>
24983 * opts.c (finish_options): If -fsplit-stack, disable implicit
24984 -forder-blocks-and-partition.
24985 * doc/invoke.texi (Optimize Options): Document that when using
24986 -fsplit-stack -forder-blocks-and-partition is not implicitly
24989 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24991 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
24992 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
24993 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
24994 * builtins.def (abort, trap, unreachable): Declare cold.
24995 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
24996 * tree-core.h (ECF_COLD): New.
24997 * tree.c (set_call_expr_flags): Handle ECF_COLD.
24998 (build_common_builtin_nodes): Mark unreachable and abort as cold.
25000 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25002 * predict.c (unlikely_executed_stmt_p): Cleanup.
25004 2017-06-09 Richard Biener <rguenther@suse.de>
25006 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
25007 model if the ref is always written to.
25009 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25011 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
25013 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25015 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
25018 2017-06-09 Tom de Vries <tom@codesourcery.com>
25021 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
25022 "target cannot support label values" when encountering LABEL_REF.
25024 2017-06-09 Martin Liska <mliska@suse.cz>
25026 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
25027 (gimple_gen_ic_func_profiler): Emit direct comparison
25028 of __gcov_indirect_call_callee with NULL.
25029 (gimple_gen_time_profiler): Change probability from
25030 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
25032 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25034 * profile.c (edge_gcov_counts): Turn to pointer.
25035 (compute_branch_probabilities, compute_branch_probabilities): Update.
25036 (branch_prob): Do not clear edge_gcov_count.
25037 * profile.h (edge_gcov_counts): Turn to pointer.
25038 (edge_gcov_count): Update.
25040 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25042 * gimple.h (gimple_check_failed): Mark cold.
25044 2017-06-09 Richard Biener <rguenther@suse.de>
25046 PR tree-optimization/66623
25047 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
25048 refactor check_reduction into two parts, properly computing
25049 whether we have to check reduction validity for outer loop
25052 2017-06-09 Richard Biener <rguenther@suse.de>
25054 PR tree-optimization/79483
25055 * graphite-scop-detection.c (order): New global.
25056 (get_order): Compute bb to order mapping that satisfies code
25057 generation constraints.
25058 (cmp_pbbs): New helper.
25059 (build_scops): Start domwalk at entry block, sort generated
25062 2017-06-09 Richard Biener <rguenther@suse.de>
25064 PR middle-end/81007
25065 * ipa-polymorphic-call.c
25066 (ipa_polymorphic_call_context::restrict_to_inner_class):
25067 Skip FIELD_DECLs with error_mark_node type.
25068 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
25071 2017-06-09 Martin Liska <mliska@suse.cz>
25073 * predict.c (struct branch_predictor): New struct.
25074 (test_prediction_value_range): New test.
25075 (predict_c_tests): New function.
25076 * selftest-run-tests.c (selftest::run_tests): Run the function.
25077 * selftest.h: Declare new tests.
25079 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
25082 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
25083 gen_add3_insn did not fail.
25084 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
25085 r0, construct that number in a temporary reg and add that reg to r0.
25086 If asked to put the result in r0 as well, fail.
25088 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
25090 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25091 for early expansion of vec_eqv.
25093 2017-06-08 Jakub Jelinek <jakub@redhat.com>
25095 PR middle-end/81005
25096 * ubsan.c (instrument_null): Avoid pointless code temporary.
25097 (pass_ubsan::execute): Instrument aggregate arguments of calls.
25099 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
25103 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
25106 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
25109 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25111 * predict.c (unlikely_executed_edge_p): Move ahead.
25112 (probably_never_executed_edge_p): Use it.
25114 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25116 PR middle-end/79988
25117 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
25118 gimple_call_builtin_p call.
25120 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25122 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
25123 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
25124 rtl_check_failed_type2, rtl_check_failed_code1,
25125 rtl_check_failed_code2, rtl_check_failed_code_mode,
25126 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
25127 rtvec_check_failed_bounds, rtl_check_failed_flag,
25128 _fatal_insn_not_found, _fatal_insn): Likewise.
25129 * tree.h (tree_contains_struct_check_failed,
25130 tree_check_failed, tree_not_check_failed,
25131 tree_class_check_failed, tree_range_check_failed,
25132 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
25133 tree_vec_elt_check_failed, phi_node_elt_check_failed,
25134 tree_operand_check_failed, omp_clause_check_failed,
25135 omp_clause_operand_check_failed, omp_clause_range_check_failed):
25138 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25140 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
25141 flag_branch_probabilities.
25142 * ipa-inline.c (edge_badness): Likewise.
25143 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
25144 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
25145 * predict.c (maybe_hot_frequency_p): Likewise.
25146 (probably_never_executed): Likewise.
25147 * sched-ebb.c (schedule_ebbs): Likewise.
25148 * sched-rgn.c (find_single_block_region): Likewise.
25149 * tracer.c (tail_duplicate): Likewise.
25151 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25153 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
25154 longer requires x_flag_profile_use.
25156 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25158 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
25159 instead of flag_reorder_blocks_and_partition.
25160 * dbxout.c (dbxout_function_end): Likewise.
25161 * dwarf2out.c (gen_subprogram_die): Likewise.
25162 * haifa-sched.c (sched_create_recovery_edges): Likewise.
25163 * hw-doloop.c (reorg_loops): Likewise.
25164 * varasm.c (assemble_start_function,
25165 assemble_end_function): Likewise.
25166 (decide_function_section): Do not check for
25167 flag_reorder_blocks_and_partition.
25169 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25171 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
25173 (chkp_get_hard_register_fake_addr_expr): Ditto.
25174 (chkp_build_addr_expr): Add check for hard reg case.
25175 (chkp_parse_array_and_component_ref): Ditto.
25176 (chkp_find_bounds_1): Ditto.
25177 (chkp_process_stmt): Don't generate bounds store for
25180 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25182 * predict.c (maybe_hot_bb_p): Do not check profile status.
25183 (maybe_hot_edge_p): Likewise.
25184 (probably_never_executed): Check for zero counts even if profile
25186 (unlikely_executed_edge_p): New function.
25187 (unlikely_executed_stmt_p): New function.
25188 (unlikely_executed_bb_p): New function.
25189 (set_even_probabilities): Use unlikely predicates.
25190 (combine_predictions_for_bb): Likewise.
25191 (predict_paths_for_bb): Likewise.
25192 (predict_paths_leading_to_edge): Likewise.
25193 (determine_unlikely_bbs): New function.
25194 (estimate_bb_frequencies): Use it.
25195 (compute_function_frequency): Use zero counts even if profile is
25197 * profile-count.h: Fix typo.
25199 2017-08-08 Julia Koval <julia.koval@intel.com>
25201 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
25202 _mm512_mask_cvtsepi16_storeu_epi8,
25203 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
25204 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
25205 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
25206 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
25207 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
25208 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
25209 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
25210 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
25211 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
25212 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
25213 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
25214 __builtin_ia32_pmovuswb256mem_mask,
25215 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
25216 __builtin_ia32_pmovwb512mem_mask): New builtins.
25218 2017-08-08 Julia Koval <julia.koval@intel.com>
25220 PR target/73350,80862
25221 * config/i386/subst.md (round): Fix round pattern.
25222 * config/i386/i386.c (ix86_erase_embedded_rounding):
25223 Fix erasing rounding for the fixed pattern.
25225 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25227 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
25229 2017-06-08 Martin Liska <mliska@suse.cz>
25231 PR gcov-profile/80911
25232 * gcov.c (block_info::block_info): New constructor.
25234 2017-06-07 Carl Love <cel@us.ibm.com>
25236 * config/rs6000/rs6000-c: The return type of the following
25237 built-in functions was implemented as int not long long. Fix sign
25238 of return value for the unsigned version of vec_mulo and vec_mule.
25239 vector unsigned long long vec_bperm (vector unsigned long long,
25240 vector unsigned char)
25241 vector signed long long vec_mule (vector signed int,
25243 vector unsigned long long vec_mule (vector unsigned int,
25244 vector unsigned int)
25245 vector signed long long vec_mulo (vector signed int,
25247 vector unsigned long long vec_mulo (vector unsigned int,
25248 vector unsigned int)
25249 * doc/extend.texi: Fix the documentation for the built-in
25252 2017-06-07 Carl Love <cel@us.ibm.com>
25255 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
25258 2017-06-07 Carl Love <cel@us.ibm.com>
25260 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
25261 support, Generate doublehv for signed int/float for BE case only.
25263 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
25265 * doc/invoke.texi (mcx16): Rewrite.
25267 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25269 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
25270 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
25271 *mov<mode>_softfloat, and an anonymous splitter): Use
25272 nonimmediate_operand instead of rs6000_nonimmediate_operand.
25274 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25276 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
25278 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25279 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
25280 (rs6000_debug_reg_global): Adjust.
25281 (rs6000_init_hard_regno_mode_ok): Adjust.
25282 (rs6000_dbx_register_number): Adjust.
25283 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
25284 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25285 Remove SPE_ACC and SPEFSCR.
25286 (REG_ALLOC_ORDER): Ditto.
25287 (FRAME_POINTER_REGNUM): Change to 111.
25288 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
25289 (REG_CLASS_NAMES): Ditto.
25290 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
25291 (REGISTER_NAMES): Ditto.
25292 (ADDITIONAL_REG_NAMES): Ditto.
25293 (rs6000_reg_names): Ditto.
25294 * config/rs6000/rs6000.md: Renumber some register number
25297 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25299 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
25301 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25302 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
25304 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
25305 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25306 Delete the SPE high registers.
25307 (REG_ALLOC_ORDER): Ditto.
25308 (enum reg_class): Remove SPE_HIGH_REGS.
25309 (REG_CLASS_NAMES): Ditto.
25310 (REG_CLASS_CONTENTS): Delete the SPE high registers.
25311 (REGISTER_NAMES): Ditto.
25312 (rs6000_reg_names): Ditto.
25313 * doc/tm.texi.in: Remove SPE as example.
25314 * doc/tm.texi: Regenerate.
25316 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25318 * config/rs6000/8540.md (ppc8540_brinc): Delete.
25319 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
25320 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
25321 * config/rs6000/rs6000.md (type): Remove "brinc".
25323 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25325 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
25326 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
25327 * config/rs6000/linuxspe.h: Delete file.
25328 * config/rs6000/rs6000.md: Don't include spe.md.
25329 * config/rs6000/spe.h: Delete file.
25330 * config/rs6000/spe.md: Delete file.
25331 * config/rs6000/t-rs6000: Remove spe.md.
25333 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25335 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
25336 (reg_or_none500mem_operand): Delete.
25337 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
25338 instead of reg_or_none500mem_operand.
25340 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25342 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
25343 handling of SPE flags.
25344 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
25346 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25348 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
25350 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
25351 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
25352 paired_divv2sf3): Similar.
25353 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
25354 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
25355 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
25357 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
25358 Rename the paired_* instruction patterns.
25359 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
25361 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
25362 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
25363 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
25364 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
25365 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
25366 PAIRED_VECTOR_MODE.
25367 (struct machine_function): Delete field spe_insn_chain_scanned_p.
25368 (spe_func_has_64bit_regs_p): Delete.
25369 (spe_expand_predicate_builtin): Delete.
25370 (spe_expand_evsel_builtin): Delete.
25371 (TARGET_DWARF_REGISTER_SPAN): Do not define.
25372 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
25373 (invalid_e500_subreg): Delete.
25374 (rs6000_legitimize_address): Always force_reg op2 as well, for
25375 paired single memory accesses.
25376 (rs6000_member_type_forces_blk): Delete.
25377 (rs6000_spe_function_arg): Delete.
25378 (rs6000_expand_unop_builtin): Delete SPE handling.
25379 (rs6000_expand_binop_builtin): Ditto.
25380 (spe_expand_stv_builtin): Delete.
25381 (bdesc_2arg_spe): Delete.
25382 (spe_expand_builtin): Delete.
25383 (spe_expand_predicate_builtin): Delete.
25384 (spe_expand_evsel_builtin): Delete.
25385 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
25386 (spe_init_builtins): Delete.
25387 (spe_func_has_64bit_regs_p): Delete.
25388 (savres_routine_name): Delete "info" parameter. Adjust callers.
25389 (rs6000_emit_stack_reset): Ditto.
25390 (rs6000_dwarf_register_span): Delete.
25391 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
25392 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
25393 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
25395 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
25397 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
25398 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
25399 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
25400 mulv2sf3, divv2sf3): Delete expanders.
25402 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25404 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
25406 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25408 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
25409 * config/rs6000/rs6000.c: Ditto.
25411 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25413 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
25414 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
25415 comparison_operator.
25417 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25419 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
25420 * config/rs6000/rs6000.opt: Ditto.
25421 * config/rs6000/t-rtems: Ditto.
25423 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25425 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
25426 TARGET_E500_SINGLE by 0, simplify.
25427 * config/rs6000/rs6000.c: Ditto.
25428 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
25429 (spe_build_register_parallel): Delete.
25430 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
25431 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
25432 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
25433 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
25434 (E500_CONVERT): Delete.
25435 * config/rs6000/spe.md: Remove many patterns and all define_constants.
25437 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25439 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
25440 * config/rs6000/dfp.md: Ditto.
25441 (negdd2, *negdd2_fpr): Merge.
25442 (absdd2, *absdd2_fpr): Merge.
25443 (negtd2, *negtd2_fpr): Merge.
25444 (abstd2, *abstd2_fpr): Merge.
25445 * config/rs6000/e500.h: Delete file.
25446 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
25447 TARGET_FPRS by 1 and simplify.
25448 * config/rs6000/rs6000-c.c: Ditto.
25449 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
25450 TARGET_DF_SPE by 0.
25451 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
25453 * config/rs6000/rs6000.md: Ditto.
25454 (floatdidf2, *floatdidf2_fpr): Merge.
25455 (move_from_CR_gt_bit): Delete.
25456 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
25457 (E500_CR_IOR_COMPARE): Delete.
25458 (All patterns that require !TARGET_FPRS): Delete.
25459 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
25461 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25463 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
25465 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25467 * graphds.c (add_edge): Intitialize edge's attached data.
25468 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
25469 pointer parameter. Call pointed function on each edge during
25470 graph traversing. Skip traversing the edge when the function
25472 (graphds_dfs, graphds_scc): Ditto.
25473 (for_each_edge): New parameter. Pass the new parameter to callback
25475 * graphds.h (skip_edge_callback): New function pointer type.
25476 (graphds_dfs, graphds_scc): New function pointer parameter.
25477 (graphds_edge_callback, for_each_edge): New parameter.
25479 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25481 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
25482 out code checking if runtime alias check is possible to below ...
25483 Call the new function.
25484 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
25485 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
25487 2017-06-07 Marek Polacek <polacek@redhat.com>
25490 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
25491 TYPE_OVERFLOW_WRAPS check.
25493 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25495 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
25496 if versioning is required.
25497 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
25498 peeling with the check for versioning.
25500 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25502 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
25503 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
25504 Set true to new parameter if new ssa variable is defined.
25505 (vect_gen_vector_loop_niters): Refactor. Set range information
25506 for the new vector loop bound variable.
25507 (vect_do_peeling): Ditto.
25509 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25511 * tree-affine.c (ssa.h): Include header file.
25512 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
25513 has wrapping overflow behavior.
25515 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25517 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
25519 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25521 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
25522 (tree_to_aff_combination): ... here.
25524 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25526 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
25527 reg_pressure model function.
25528 (ivopts_global_cost_for_size): Delete.
25529 (determine_set_costs, iv_ca_recount_cost): Call new model function
25530 ivopts_estimate_reg_pressure.
25532 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25534 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
25535 expensive than udiv. Remove floating point cases from mod.
25537 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25539 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
25540 Increase idiv cost.
25542 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25544 * config/aarch64/aarch64.md
25545 (copysignsf3): Fix mask generation.
25547 2017-06-07 Jakub Jelinek <jakub@redhat.com>
25549 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
25551 (class dump_manager): Add register_dumps method.
25552 * dumpfile.c: Include langhooks.h.
25553 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
25554 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
25555 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
25556 (dump_manager::dump_register): Start with 512 entries instead of 32.
25557 (dump_manager::register_dumps): New method.
25558 * toplev.c (general_init): Instead of invoking register_dumps
25559 langhook, invoke register_dumps method on the dump manager.
25560 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
25563 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
25565 * doc/md.texi: Clarify the restrictions on a define_insn condition.
25566 Say that # requires an associated define_split to exist, and that
25567 the define_split must be suitable for use after register allocation.
25569 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25571 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
25572 (compute_outgoing_frequencies): Also initialize zero counts.
25573 (find_many_sub_basic_blocks): Do not produce uninitialized profile
25574 around loops; preserve more of profile when nothing changes.
25576 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
25578 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
25580 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
25581 * config/arm/arm-cpu-cdata.h: Regenerate.
25582 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
25583 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
25584 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
25585 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
25586 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
25588 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
25590 * config/arm/t-rmprofile: Likewise.
25591 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
25593 2017-06-06 David S. Miller <davem@davemloft.net>
25596 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
25597 blockage if function uses alloca.
25599 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25601 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
25602 New "uid" fields to hold pretty-print IDs of group and ref.
25603 Memory references are now identified as <group_id>:<ref_id>
25604 instead of using [random] addresses.
25605 (dump_mem_details): Simplify, no functional change.
25606 (dump_mem_ref): Simplify and make output more concise.
25607 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
25608 (find_or_create_group): Initialize group uid.
25609 (record_ref): Initialize ref uid. Improve debug output.
25610 (prune_group_by_reuse, should_issue_prefetch_p,)
25611 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
25612 (mark_nontemporal_store, determine_loop_nest_reuse):
25613 Improve debug output.
25615 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25617 * dbgcnt.def (prefetch): New debug counter.
25618 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
25619 (schedule_prefetches): Stop issueing prefetches if debug counter
25622 2017-06-06 Tom de Vries <tom@codesourcery.com>
25624 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
25625 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
25627 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25629 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
25630 Use aarch64_reg_or_zero predicate for operand 4.
25631 (aarch64_compare_and_swap<mode> define_insn_and_split):
25632 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
25633 (aarch64_store_exclusive<mode>): Likewise for operand 2.
25635 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25637 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
25638 (arm_compute_save_core_reg_mask): This.
25639 (thumb1_compute_save_reg_mask): Rename into ...
25640 (thumb1_compute_save_core_reg_mask): This.
25641 (arm_compute_save_reg0_reg12_mask): Adapt comment.
25642 (arm_compute_frame_layout): Likewise.
25644 2017-06-06 Richard Biener <rguenther@suse.de>
25646 PR tree-optimization/80974
25647 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
25648 keep or clear leaders SSA info.
25650 2017-06-06 Tom de Vries <tom@codesourcery.com>
25652 * config/nvptx/nvptx.c (split_mode_p): New function.
25653 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
25655 2017-06-06 Tom de Vries <tom@codesourcery.com>
25657 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
25659 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25662 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
25665 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25667 * shrink-wrap.c (handle_simple_exit): Update profile.
25668 (try_shrink_wrapping): Upate profile.
25670 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25672 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
25673 (tree_guess_outgoing_edge_probabilities): New.
25674 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
25675 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
25677 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25679 * ipa-split.c (split_function): Initialize return bb profile.
25681 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25683 * profile.c (compute_branch_probabilities): Also initialize
25684 EXIT_BLOCK profile.
25686 2017-06-06 Richard Biener <rguenther@suse.de>
25688 PR tree-optimization/80928
25689 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
25690 (vect_analyze_loop_operations): Properly guard analysis for
25692 (vect_transform_loop): Likewise.
25693 (vect_analyze_loop_2): Also reset SLP type on PHIs.
25694 (vect_model_induction_cost): Do not cost for pure SLP.
25695 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
25696 of induction in inner loop vectorization.
25697 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
25698 (vect_get_and_check_slp_defs): Handle vect_induction_def.
25699 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
25701 (vect_analyze_slp_cost_1): Cost induction.
25702 (vect_detect_hybrid_slp_stmts): Handle PHIs.
25703 (vect_get_slp_vect_defs): Likewise.
25704 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
25705 (vect_transform_stmt): Handle SLP reductions.
25706 * tree-vectorizer.h (vectorizable_induction): Adjust.
25708 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25710 * config/rs6000/rs6000.c (make_resolver_func): Update
25711 init_lowered_empty_function call.
25713 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
25715 * doc/invoke.texi: Document the -fprofile-abs-path option.
25716 * common.opt (fprofile-abs-path): New option.
25717 * gcov-io.h (gcov_write_filename): Declare.
25718 * gcov-io.c (gcov_write_filename): New function.
25719 * coverage.c (coverage_begin_function): Use gcov_write_filename.
25720 * profile.c (output_location): Likewise.
25722 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25724 * shring-wrap.c: Revert accidental commit.
25726 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
25728 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
25730 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25732 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
25734 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
25735 profile in callgraph edge.
25736 * profile-count.h (apply_probability): If THIS is 0, then result is 0
25737 (apply_scale): Likewise.
25738 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
25739 Also scale profile when inlining function with zero profile.
25740 (initialize_cfun): Update exit block profile even when it is zero.
25741 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
25742 when profile is read.
25744 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25746 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
25747 (CLONE_*): New constants to define the processors we can generate
25748 code for with the target_clone attribute.
25749 (rs6000_clone_map): New array to identify which clone processors
25750 the current program is running on.
25751 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
25752 target_clone attribute.
25753 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
25754 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
25755 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
25756 (cpu_expand_builtin): Add support for target_clone attribute.
25757 (rs6000_valid_attribute_p): Allow "default" attribute.
25758 (get_decl_name): New debug function to simplify printing the
25759 current function name in debugging statements.
25760 (rs6000_clone_priority): New functions to support the target_clone
25761 attribute, and be able to generate code to switch between ISA 2.05
25762 through ISA 3.0 (power6 through power9).
25763 (rs6000_compare_version_priority): Likewise.
25764 (rs6000_get_function_versions_dispatcher): Likewise.
25765 (make_resolver_func): Likewise.
25766 (add_condition_to_bb): Likewise.
25767 (dispatch_function_versions): Likewise.
25768 (rs6000_generate_version_dispatcher_body): Likewise.
25769 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
25770 (fusion_gpr_load_p): Fix a spacing issue.
25771 * doc/extend.texi (Common Function Attributes): Document that the
25772 PowerPC supports the target_clone attribute.
25774 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25776 * config/arm/arm.h: explain F symbol found in description of ARM
25777 register allocation in its legend.
25779 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25781 * config/mips/frame-header-opt.c: Include profile-count.h.
25782 * config/riscv/riscv.c: Include profile-count.h
25784 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25786 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
25788 (sm_set_flag_if_changed): Add bbs field.
25789 (execute_sm_if_changed_flag_set): Pass BBS.
25790 (execute_sm): Update.
25792 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25794 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
25797 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25799 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
25800 (peephole2): New peephole2 to emit the above.
25801 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
25803 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25805 * config/aarch64/aarch64.c (define_peephole2 above
25806 *sub_<shift>_<mode>): New peephole.
25808 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25810 * config/i386/i386.c (make_resolver_func): Update.
25811 * Makefile.in: Add profile-count.h and profile-count.o
25812 * auto-profile.c (afdo_indirect_call): Update to new API.
25813 (afdo_set_bb_count): Update.
25814 (afdo_propagate_edge): Update.
25815 (afdo_propagate_circuit): Update.
25816 (afdo_calculate_branch_prob): Update.
25817 (afdo_annotate_cfg): Update.
25818 * basic-block.h: Include profile-count.h
25819 (struct edge_def): Turn count to profile_count.
25820 (struct basic_block_def): Likewie.
25821 (REG_BR_PROB_BASE): Move to profile-count.h
25822 (RDIV): Move to profile-count.h
25823 * bb-reorder.c (max_entry_count): Turn to profile_count.
25824 (find_traces): Update.
25825 (rotate_loop):Update.
25826 (connect_traces):Update.
25827 (sanitize_hot_paths):Update.
25828 * bt-load.c (migrate_btr_defs): Update.
25829 * cfg.c (RDIV): Remove.
25830 (init_flow): Use alloc_block.
25831 (alloc_block): Uninitialize count.
25832 (unchecked_make_edge): Uninitialize count.
25833 (check_bb_profile): Update.
25834 (dump_edge_info): Update.
25835 (dump_bb_info): Update.
25836 (update_bb_profile_for_threading): Update.
25837 (scale_bbs_frequencies_int): Update.
25838 (scale_bbs_frequencies_gcov_type): Update.
25839 (scale_bbs_frequencies_profile_count): New.
25840 * cfg.h (update_bb_profile_for_threading): Update.
25841 (scale_bbs_frequencies_profile_count): Declare.
25842 * cfgbuild.c (compute_outgoing_frequencies): Update.
25843 (find_many_sub_basic_blocks): Update.
25844 * cfgcleanup.c (try_forward_edges): Update.
25845 (try_crossjump_to_edge): Update.
25846 * cfgexpand.c (expand_gimple_tailcall): Update.
25847 (construct_exit_block): Update.
25848 * cfghooks.c (verify_flow_info): Update.
25849 (dump_bb_for_graph): Update.
25850 (split_edge): Update.
25851 (make_forwarder_block): Update.
25852 (duplicate_block): Update.
25853 (account_profile_record): Update.
25854 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
25855 (get_estimated_loop_iterations): Update.
25856 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
25857 (single_likely_exit): Update.
25858 * cfgloopmanip.c (scale_loop_profile): Update.
25860 (set_zero_probability): Update.
25861 (lv_adjust_loop_entry_edge): Update.
25862 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
25863 (purge_dead_edges): Update.
25864 (rtl_account_profile_record): Update.
25865 * cgraph.c (cgraph_node::create): Uninitialize count.
25866 (symbol_table::create_edge): Uninitialize count.
25867 (cgraph_update_edges_for_call_stmt_node): Update.
25868 (cgraph_edge::dump_edge_flags): Update.
25869 (cgraph_node::dump): Update.
25870 (cgraph_edge::maybe_hot_p): Update.
25871 * cgraph.h: Include profile-count.h
25872 (create_clone), create_edge, create_indirect_edge): Update.
25873 (cgraph_node): Turn count to profile_count.
25874 (cgraph_edge0: Likewise.
25875 (make_speculative, clone): Update.
25876 (create_edge): Update.
25877 (init_lowered_empty_function): Update.
25878 * cgraphclones.c (cgraph_edge::clone): Update.
25879 (duplicate_thunk_for_node): Update.
25880 (cgraph_node::create_clone): Update.
25881 * cgraphunit.c (cgraph_node::analyze): Update.
25882 (cgraph_node::expand_thunk): Update.
25883 * final.c (dump_basic_block_info): Update.
25884 * gimple-streamer-in.c (input_bb): Update.
25885 * gimple-streamer-out.c (output_bb): Update.
25886 * graphite.c (print_global_statistics): Update.
25887 (print_graphite_scop_statistics): Update.
25888 * hsa-brig.c: Include basic-block.h.
25889 * hsa-dump.c: Include basic-block.h.
25890 * hsa-gen.c (T sum_slice): Update.
25891 (convert_switch_statements):Update.
25892 * hsa-regalloc.c: Include basic-block.h.
25893 * ipa-chkp.c (chkp_produce_thunks): Update.
25894 * ipa-cp.c (struct caller_statistics): Update.
25895 (init_caller_stats): Update.
25896 (gather_caller_stats): Update.
25897 (ipcp_cloning_candidate_p): Update.
25898 (good_cloning_opportunity_p): Update.
25899 (get_info_about_necessary_edges): Update.
25900 (dump_profile_updates): Update.
25901 (update_profiling_info): Update.
25902 (update_specialized_profile): Update.
25903 (perhaps_add_new_callers): Update.
25904 (decide_about_value): Update.
25905 (ipa_cp_c_finalize): Update.
25906 * ipa-devirt.c (struct odr_type_warn_count): Update.
25907 (struct decl_warn_count): Update.
25908 (struct final_warning_record): Update.
25909 (possible_polymorphic_call_targets): Update.
25910 (ipa_devirt): Update.
25911 * ipa-fnsummary.c (redirect_to_unreachable): Update.
25912 * ipa-icf.c (sem_function::merge): Update.
25913 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
25914 * ipa-inline.c (compute_uninlined_call_time): Update.
25915 (compute_inlined_call_time): Update.
25916 (want_inline_small_function_p): Update.
25917 (want_inline_self_recursive_call_p): Update.
25918 (edge_badness): Update.
25919 (lookup_recursive_calls): Update.
25920 (recursive_inlining): Update.
25921 (inline_small_functions): Update.
25922 (dump_overall_stats): Update.
25923 (dump_inline_stats): Update.
25924 * ipa-profile.c (ipa_profile_generate_summary): Update.
25925 (ipa_propagate_frequency): Update.
25926 (ipa_profile): Update.
25927 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
25928 * ipa-utils.c (ipa_merge_profiles): Update.
25929 * loop-doloop.c (doloop_modify): Update.
25930 * loop-unroll.c (report_unroll): Update.
25931 (unroll_loop_runtime_iterations): Update.
25932 * lto-cgraph.c (lto_output_edge): Update.
25933 (lto_output_node): Update.
25934 (input_node): Update.
25935 (input_edge): Update.
25936 (merge_profile_summaries): Update.
25937 * lto-streamer-in.c (input_cfg): Update.
25938 * lto-streamer-out.c (output_cfg): Update.
25939 * mcf.c (create_fixup_graph): Update.
25940 (adjust_cfg_counts): Update.
25941 (sum_edge_counts): Update.
25942 * modulo-sched.c (sms_schedule): Update.
25943 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
25944 * predict.c (maybe_hot_count_p): Update.
25945 (probably_never_executed): Update.
25946 (dump_prediction): Update.
25947 (combine_predictions_for_bb): Update.
25948 (propagate_freq): Update.
25949 (handle_missing_profiles): Update.
25950 (counts_to_freqs): Update.
25951 (rebuild_frequencies): Update.
25952 (force_edge_cold): Update.
25953 * predict.h: Include profile-count.h
25954 (maybe_hot_count_p, counts_to_freqs): UPdate.
25955 * print-rtl-function.c: Do not include cfg.h
25956 * print-rtl.c: Include basic-block.h
25957 * profile-count.c: New file.
25958 * profile-count.h: New file.
25959 * profile.c (is_edge_inconsistent): Update.
25960 (correct_negative_edge_counts): Update.
25961 (is_inconsistent): Update.
25962 (set_bb_counts): Update.
25963 (read_profile_edge_counts): Update.
25964 (compute_frequency_overlap): Update.
25965 (compute_branch_probabilities): Update; Initialize and deinitialize
25967 (branch_prob): Update.
25968 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
25969 (edge_gcov_count): New.
25970 (bb_gcov_count): New.
25971 * shrink-wrap.c (try_shrink_wrapping): Update.
25972 * tracer.c (better_p): Update.
25973 * trans-mem.c (expand_transaction): Update.
25974 (ipa_tm_insert_irr_call): Update.
25975 (ipa_tm_insert_gettmclone_call): Update.
25976 * tree-call-cdce.c: Update.
25977 * tree-cfg.c (gimple_duplicate_sese_region): Update.
25978 (gimple_duplicate_sese_tail): Update.
25979 (gimple_account_profile_record): Update.
25980 (execute_fixup_cfg): Update.
25981 * tree-inline.c (copy_bb): Update.
25982 (copy_edges_for_bb): Update.
25983 (initialize_cfun): Update.
25984 (freqs_to_counts): Update.
25985 (copy_cfg_body): Update.
25986 (expand_call_inline): Update.
25987 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25988 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
25989 (try_unroll_loop_completely): Update.
25990 (try_peel_loop): Update.
25991 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25992 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
25993 * tree-ssa-loop-split.c (connect_loops): Update.
25994 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
25995 * tree-ssa-reassoc.c (branch_fixup): Update.
25996 * tree-ssa-tail-merge.c (replace_block_by): Update.
25997 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
25998 (compute_path_counts): Update.
25999 (update_profile): Update.
26000 (recompute_probabilities): Update.
26001 (update_joiner_offpath_counts): Update.
26002 (estimated_freqs_path): Update.
26003 (freqs_to_counts_path): Update.
26004 (clear_counts_path): Update.
26005 (ssa_fix_duplicate_block_edges): Update.
26006 (duplicate_thread_path): Update.
26007 * tree-switch-conversion.c (case_bit_test_cmp): Update.
26008 (struct switch_conv_info): Update.
26009 * tree-tailcall.c (decrease_profile): Update.
26010 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26011 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26012 * value-prof.c (check_counter): Update.
26013 (gimple_divmod_fixed_value): Update.
26014 (gimple_mod_pow2): Update.
26015 (gimple_mod_subtract): Update.
26016 (gimple_ic_transform): Update.
26017 (gimple_stringop_fixed_value): Update.
26018 * value-prof.h (gimple_ic): Update.
26020 2017-06-02 Carl Love <cel@us.ibm.com>
26022 * config/rs6000/rs6000-c: Add support for built-in functions
26023 vector double vec_doublee (vector signed int);
26024 vector double vec_doublee (vector unsigned int);
26025 vector double vec_doublee (vector float);
26026 vector double vec_doubleh (vector signed int);
26027 vector double vec_doubleh (vector unsigned int);
26028 vector double vec_doubleh (vector float);
26029 vector double vec_doublel (vector signed int);
26030 vector double vec_doublel (vector unsigned int);
26031 vector double vec_doublel (vector float);
26032 vector double vec_doubleo (vector signed int);
26033 vector double vec_doubleo (vector unsigned int);
26034 vector double vec_doubleo (vector float);.
26035 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
26036 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
26038 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
26039 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
26040 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
26042 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
26043 vec_doublel, vec_doubleh.
26044 * doc/extend.texi: Update the built-in documentation file for the
26045 new built-in functions.
26047 2017-06-02 David Malcolm <dmalcolm@redhat.com>
26050 * ipa-inline-analysis.c (free_growth_caches): Set
26051 edge_removal_hook_holder to NULL after removing it.
26053 2017-06-02 Sudakshina Das <sudi.das@arm.com>
26055 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
26056 comparision with zero.
26058 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26059 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26060 for early expansion of vec_min and vec_max builtins.
26061 (builtin_function_type): Add min/max unsigned variants to those
26062 identified as having unsigned arguments.
26064 2017-06-02 Olivier Hainque <hainque@adacore.com>
26066 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
26068 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26070 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
26071 Use VALL_F16 iterator rather than VALL.
26073 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26075 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
26076 Emit CBNZ inside loop when doing a strong exchange and comparing
26077 against zero. Generate the CC flags after the loop.
26079 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
26081 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
26082 (dl_section_ref): New.
26083 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
26084 On AIX, append an expression to subtract the size of the
26085 section length to dl_section_ref.
26087 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26089 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26090 for early expansion of vector absolute builtins.
26092 2017-06-02 Richard Biener <rguenther@suse.de>
26094 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
26095 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
26097 2017-06-02 Richard Biener <rguenther@suse.de>
26099 PR tree-optimization/80948
26100 * tree-tailcall.c (find_tail_calls): Track stmts to move in
26101 stmt order as well.
26103 2017-06-02 Richard Biener <rguenther@suse.de>
26105 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
26107 * tree-vect-stmts.c (process_use): Do not mark backedge defs
26108 for inductions as relevant.
26110 2017-06-02 Richard Biener <rguenther@suse.de>
26112 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
26113 (vectorizable_induction): ... this. Remove dead code.
26115 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
26117 * builtins. (expand_builtin_alloca): Remove second parameter and
26118 infer its value from the first parameter instead.
26119 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
26121 2017-06-02 Jakub Jelinek <jakub@redhat.com>
26123 PR rtl-optimization/80903
26124 * loop-doloop.c (add_test): Unshare sequence.
26126 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
26128 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
26130 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
26132 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
26134 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
26135 xlogue_layout::get_instance, logue_layout::xlogue_layout,
26136 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
26137 (xlogue_layout::get_stub_rtx): Make static.
26138 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
26139 (xlogue_layout::compute_stub_managed_regs): Rename to...
26140 (xlogue_layout::count_stub_managed_regs): ...this.
26141 (xlogue_layout::is_stub_managed_reg): New function.
26142 (xlogue_layout::m_stub_names): Rename to...
26143 (xlogue_layout::s_stub_names): ...this, make static.
26144 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
26145 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
26146 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
26147 xlogue_layout::s_stub_names): Instantiate statics.
26148 (stub_managed_regs): Remove.
26149 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
26150 (disable_call_ms2sysv_xlogues): Rename to...
26151 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
26152 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
26154 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
26155 change after reload_completed.
26156 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
26158 (ix86_expand_prologue): Likewise.
26159 (ix86_expand_epilogue): Likewise.
26160 (ix86_expand_split_stack_prologue): Likewise.
26161 (ix86_compute_frame_layout): Remove frame parameter ...
26162 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
26163 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
26165 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
26166 (ix86_frame): Move from here ...
26167 * config/i386/i386.h (ix86_frame): ... to here.
26168 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
26169 complete ix86_frame data structure instead. Remove some_ld_name.
26171 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26173 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
26174 symbols that hold a DECL_VALUE_EXPR.
26176 2017-06-01 Martin Jambor <mjambor@suse.cz>
26178 PR tree-optimization/80898
26179 * tree-sra.c (process_subtree_disqualification): Removed.
26180 (disqualify_candidate): Do not acll
26181 process_subtree_disqualification.
26182 (subtree_mark_written_and_enqueue): New function.
26183 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
26184 RHS has been disqualified and re-queue LHS if necessary. Apart
26185 from that, ignore disqualified RHS.
26187 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26189 * config/s390/s390.c (s390_emit_epilogue): Disable early return
26190 address fetch for z10 or later.
26192 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26194 * config/arc/arc.md (tst_movb): Add guard when splitting.
26196 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26198 * config/arc/arc.c (arc_can_eliminate): Test against
26199 arc_frame_pointer_needed.
26201 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26203 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
26204 to prevent store reordering.
26205 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
26206 (type): Add block type.
26207 (stack_tie): Define special instruction to be used in
26210 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26212 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
26213 constraint. It is not valid for the pattern.
26214 (noncommutative_binary_comparison): Likewise.
26216 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26218 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
26221 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26223 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
26224 be used by the reg-alloc.
26226 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26228 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
26229 reg-alloc when having mul64 or mul32x16 instructions.
26230 (mulsidi3): Likewise.
26231 (umulsidi3): Likewise.
26232 (mulsi32x16): New pattern.
26233 (mulsi64): Likewise.
26234 (mulsidi64): Likewise.
26235 (umulsidi64): Likewise.
26236 (MUL32x16_REG): Define.
26237 (mul64_600): Use MUL32x16_REG.
26238 (mac64_600): Likewise.
26239 (umul64_600): Likewise.
26240 (umac64_600): Likewise.
26242 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26244 * config/arc/arc.md (mulsi3_700): Make it commutative.
26246 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
26248 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
26250 (*sign_extendsidi2_insn): Likewise for movstosw.
26252 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26254 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
26255 the type of the input discriminant value. Convert the
26256 discriminant value of signedness vary.
26258 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
26260 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
26261 Add to -Wall section.
26263 2017-06-01 Richard Biener <rguenther@suse.de>
26265 PR middle-end/66313
26266 * fold-const.c (fold_plusminus_mult_expr): If the factored
26267 factor may be zero use a wrapping type for the inner operation.
26268 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
26269 and handle moved defs.
26270 (process_assignment): Properly guard the unary op case. Return a
26271 tri-state indicating that moving the stmt before the call may allow
26272 to continue. Pass through to_move.
26273 (find_tail_calls): Handle moving unrelated defs before
26276 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
26279 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
26280 splitter result in the canonical way.
26282 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
26284 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
26285 also for 32bit target. Update insn attributes.
26286 (zero-extendsidi2 splitter): Allow all registers for operand 1.
26288 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
26290 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
26291 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
26292 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
26293 (_mm_maskz_min_ss): New intrinsics.
26295 2017-05-31 Martin Liska <mliska@suse.cz>
26297 * tree-vect-loop.c (vect_create_epilog_for_reduction):
26298 Change comment style to one we normally use.
26299 (vectorizable_reduction): Likewise.
26300 (vectorizable_induction): Likewise.
26301 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
26302 (vectorizable_call): Likewise.
26303 (vectorizable_simd_clone_call): Likewise.
26304 (vectorizable_conversion): Likewise.
26305 (vectorizable_assignment): Likewise.
26306 (vectorizable_shift): Likewise.
26307 (vectorizable_operation): Likewise.
26308 (vectorizable_store): Likewise.
26309 (vectorizable_load): Likewise.
26310 * tree-vectorizer.h: Likewise.
26312 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
26314 * passes.c (emergency_dump_function): New.
26315 * tree-pass.h (emergency_dump_function): Declare.
26316 * plugin.c (plugins_internal_error_function): Remove.
26317 * plugin.h (plugins_internal_error_function): Remove declaration.
26318 * toplev.c (internal_error_function): New static function. Use it...
26319 (general_init): ...here.
26321 2017-05-31 Graham Markall <graham.markall@embecosm.com>
26323 * config/arc/arc.c (arc_print_operand): Handle constant operands.
26324 (arc_rtx_costs): Add costs for new patterns.
26325 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
26326 * config/arc/predicates.md: Add _1_2_3_operand predicate.
26328 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26330 * tree-ssa-strlen.c (get_next_strinfo): New function.
26331 (get_stridx_plus_constant): Use it.
26332 (zero_length_string): Likewise.
26333 (adjust_related_strinfos): Likewise.
26334 (adjust_last_stmt): Likewise.
26336 2017-05-31 Richard Biener <rguenther@suse.de>
26339 * config/i386/i386.c (ix86_expand_builtin): Remove assert
26340 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
26342 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26344 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
26345 loop_vinfo argument and use of dependence distance vectors.
26346 Check instead whether the two references differ only in their
26347 initial value and assume that they have the same alignment if the
26348 difference is a multiple of the vector alignment.
26349 (vect_analyze_data_refs_alignment): Update call accordingly.
26351 2017-05-31 Martin Liska <mliska@suse.cz>
26354 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
26356 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26358 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
26359 (create_intersect_range_checks): Move from ...
26360 * tree-data-ref.c (create_intersect_range_checks_index)
26361 (create_intersect_range_checks): ... to here.
26362 (create_runtime_alias_checks): New function factored from ...
26363 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
26364 here. Call above function.
26365 * tree-data-ref.h (create_runtime_alias_checks): New function.
26367 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26369 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
26370 segment length for dr_b and compute it in wide_int.
26372 2017-05-31 Richard Biener <rguenther@suse.de>
26374 PR tree-optimization/80906
26375 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
26376 and pass through iv_map.
26377 (copy_bb_and_scalar_dependences): Adjust.
26378 (translate_pending_phi_nodes): Likewise.
26379 (copy_loop_close_phi_args): Handle code-generating IVs instead
26382 2017-05-30 David Malcolm <dmalcolm@redhat.com>
26384 * diagnostic-color.c (color_dict): Add "type-diff".
26385 (parse_gcc_colors): Update comment.
26386 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
26387 -fdiagnostics-show-template-tree and -fno-elide-type.
26388 (GCC_COLORS): Add type-diff to example.
26390 (-fdiagnostics-show-template-tree): New.
26391 (-fno-elide-type): New.
26392 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
26393 the pp_format_decoder callback. Call any m_format_postprocessor's
26395 (pretty_printer::pretty_printer): Initialize
26396 m_format_postprocessor.
26397 (pretty_printer::~pretty_printer): Delete any
26398 m_format_postprocessor.
26399 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
26400 (class format_postprocessor): New class.
26401 (struct pretty_printer::format_decoder): Document the new parameters.
26402 (struct pretty_printer::m_format_postprocessor): New field.
26403 * tree-diagnostic.c (default_tree_printer): Update for new
26404 bool and const char ** params.
26405 * tree-diagnostic.h (default_tree_printer): Likewise.
26407 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
26409 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
26410 (lwa_operand): Delete rs6000_gen_cell_microcode test.
26411 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26412 rs6000_gen_cell_microcode code.
26413 (rs6000_final_prescan_insn): Delete.
26414 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
26415 "warn-cell-microcode" entries.
26416 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
26417 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
26418 throughout. Change cc_reg_not_micro_cr0_operand to
26419 cc_reg_not_cr0_operand throughout.
26420 (*extendhi<mode>2_noload): Delete.
26421 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
26422 (mwarn-cell-microcode): Delete.
26423 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
26424 -mgen-cell-microcode and -mwarn-cell-microcode.
26426 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
26429 * config/i386/constraints.md (Yd): New constraint.
26431 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
26432 and (?Yd, r) alternatives. Update insn attributes.
26433 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
26434 and (?*Yd, r) alternatives. Update insn attributes.
26435 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
26437 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
26439 * gimplify.c (gimplify_modify_expr): Don't create a
26440 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
26443 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
26445 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
26447 2017-05-30 Richard Biener <rguenther@suse.de>
26449 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
26450 and reduc_def fields.
26451 (STMT_VINFO_REDUC_TYPE): New define.
26452 (STMT_VINFO_REDUC_DEF): Likewise.
26453 (vect_force_simple_reduction): Adjust prototype.
26454 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
26455 (vect_is_simple_reduction): Remove check_reduction argument.
26456 (vect_force_simple_reduction): Adjust and set
26457 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
26458 (vectorizable_reduction): Do not re-do reduction analysis
26459 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
26460 * tree-parloops.c (gather_scalar_reductions): Adjust.
26462 2017-05-30 Richard Biener <rguenther@suse.de>
26464 PR middle-end/80901
26465 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
26468 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
26470 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
26471 Introduce unknown_misalignment parameter and remove vf.
26472 (vect_peeling_hash_get_lowest_cost):
26473 Pass unknown_misalignment parameter.
26474 (vect_enhance_data_refs_alignment):
26475 Fix unsupportable data ref treatment.
26477 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26479 * tree-vect-data-refs.c (vect_get_data_access_cost):
26480 Workaround for SLP handling.
26481 (vect_enhance_data_refs_alignment):
26482 Compute costs for doing no peeling at all, compare to the best
26483 peeling costs so far and avoid peeling if cheaper.
26485 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26487 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
26488 Return peeling info and set costs to zero for unlimited cost
26490 (vect_enhance_data_refs_alignment): Also inspect all datarefs
26491 with unknown misalignment. Compute and costs for unknown
26492 misalignment, compare them to the costs for known misalignment
26493 and choose the cheapest for peeling.
26495 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26497 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
26498 (vect_get_peeling_costs_all_drs): Create function.
26499 (vect_peeling_hash_get_lowest_cost):
26500 Use vect_get_peeling_costs_all_drs.
26501 (vect_peeling_supportable): Create function.
26502 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
26504 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26506 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
26507 DR_HAS_NEGATIVE_STEP.
26508 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
26509 (vect_enhance_data_refs_alignment): Use.
26510 (vect_duplicate_ssa_name_ptr_info): Use.
26511 * tree-vectorizer.h (dr_misalignment): Use.
26512 (known_alignment_for_access_p): Use.
26514 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
26517 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
26519 (has_section_name): New function.
26521 2017-05-30 Martin Liska <mliska@suse.cz>
26524 * auto-profile.c (get_function_decl_from_block): Fix
26527 2017-05-30 Richard Biener <rguenther@suse.de>
26529 PR middle-end/80876
26530 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
26532 2017-05-30 Martin Liska <mliska@suse.cz>
26534 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
26535 * dumpfile.h (struct dump_file_info): Remove ctors.
26537 2017-05-30 Martin Liska <mliska@suse.cz>
26539 * predict.def: Fix GNU coding style.
26541 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
26543 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
26544 Mark 'to' argument with ATTRIBUTE_UNUSED.
26546 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
26548 * config/xtensa/xtensa.c (xtensa_emit_call): Use
26549 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
26550 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
26553 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
26555 * doc/install.texi (Options specification): Restore entry of
26556 --enable-sjlj-exceptions.
26558 2017-05-27 Michael Eager <eager@eagercon.com>
26561 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
26563 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
26565 * config/microblaze/microblaze.h
26566 (FIXED_REGISTERS): Update in macro.
26567 (CALL_USED_REGISTERS): Update in macro.
26569 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
26571 * doc/install.texi: Add links to macOS binary distributions.
26573 2017-05-27 Jakub Jelinek <jakub@redhat.com>
26577 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26579 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
26581 2017-05-26 Martin Liska <mliska@suse.cz>
26583 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
26585 2017-05-26 Martin Liska <mliska@suse.cz>
26587 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
26588 always leading ';; '.
26589 (dump_bb_info): Likewise.
26590 (brief_dump_cfg): Likewise.
26591 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
26592 * dumpfile.c: Remove usage of TDF_VERBOSE.
26593 * dumpfile.h (enum dump_kind): Likewise.
26594 (dump_gimple_bb_header): Do not use TDF_COMMENT.
26595 * print-tree.c (debug_verbose): Remove.
26596 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
26597 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
26598 * tree-diagnostic.c (default_tree_printer): Replace
26599 TDF_DIAGNOSTIC with TDF_SLIM.
26601 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26603 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
26604 in parameter loop, rather than loop_vinfo.
26605 (create_intersect_range_checks): Ditto.
26606 (vect_create_cond_for_alias_checks): Update call to above functions.
26608 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26610 PR tree-optimization/80815
26611 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
26612 for merging runtime alias checks. Handle negative DR_STEPs.
26614 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26616 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
26618 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
26619 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
26620 out code pruning runtime alias checks.
26621 * tree-data-ref.c (prune_runtime_alias_test_list): New function
26622 factored out from above.
26623 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
26625 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
26627 (prune_runtime_alias_test_list): New decalaration.
26629 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26631 * tree-vect-data-refs.c (compare_tree): Rename and move ...
26632 * tree-data-ref.c (data_ref_compare_tree): ... to here.
26633 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
26634 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
26635 (operator==, comp_dr_with_seg_len_pair): Ditto.
26636 (vect_prune_runtime_alias_test_list): Ditto.
26638 2017-05-26 Martin Liska <mliska@suse.cz>
26641 * params.def: Bound partial-inlining-entry-probability param.
26643 2017-05-26 Marek Polacek <polacek@redhat.com>
26646 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
26649 2017-05-26 Richard Biener <rguenther@suse.de>
26651 PR tree-optimization/80842
26652 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
26655 2017-05-26 Richard Biener <rguenther@suse.de>
26657 PR tree-optimization/80844
26658 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
26660 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
26662 * doc/md.texi (Machine Constraints): Update x86 family
26663 machine constraints section to match 'config/i386/constraints.md'.
26665 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
26667 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
26669 2017-05-25 Nathan Sidwell <nathan@acm.org>
26671 * doc/invoke.texi (--enable-languages): Update documentation.
26673 2017-05-25 Martin Liska <mliska@suse.cz>
26675 * dumpfile.c: Add TDF_FOLDING.
26676 * dumpfile.h (enum dump_kind): Likewise.
26677 * genmatch.c (dt_simplify::gen_1): Use it.
26679 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26681 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
26683 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26685 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
26686 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
26688 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26690 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
26691 * match.pd (X == C): Rewrite it here.
26692 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
26693 with_certain_nonzero_bits2): New predicates.
26694 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
26696 2017-05-24 Nathan Sidwell <nathan@acm.org>
26698 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
26701 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
26704 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26706 * config/powerpcspe: New port. Files are copied from the rs6000
26707 port, with "rs6000" in filenames replaced by "powerpcspe".
26709 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
26711 PR rtl-optimization/80754
26712 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
26714 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
26716 * config/sparc/sparc.md (length): Return the correct value for -mflat
26717 sibcalls to match output_sibcall.
26719 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26723 * config/rs6000/rs6000.c (struct machine_function): Add new field
26725 (rs6000_get_separate_components): Init that field, use it.
26726 (rs6000_components_for_bb): Use the field.
26728 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
26730 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
26732 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
26734 PR middle-end/80823
26735 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
26737 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26740 * config/s390/s390.c (s390_check_qrst_address): Check incoming
26741 address against address_operand predicate.
26742 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
26744 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
26746 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
26747 parameters passed indirectly.
26749 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
26751 * config/i386/i386.md (*movdi_internal): Remove SSE4
26752 alternative 18 (?r, *v). Update insn attributes.
26753 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
26754 Update insn attributes.
26755 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
26756 Update insn attributes.
26757 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
26758 alternative 1 (r, v). Remove isa attribute.
26759 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
26760 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
26761 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
26763 2017-05-23 Tom de Vries <tom@codesourcery.com>
26765 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
26768 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26770 * cgraphunit.c (symbol_table::process_new_functions): Update.
26771 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
26772 (inline_generate_summary): Rename to ...
26773 (ipa_fn_summary_generate): ... this one.
26774 (inline_read_summary): Rename to ...
26775 (ipa_fn_summary_read): ... this one.
26776 (inline_write_summary): Rename to ...
26777 (ipa_fn_summary_write): ... this one.
26778 (inline_free_summary): Rename to ...
26779 (ipa_free_fn_summary): ... this one.
26780 (pass_data_local_fn_summary, pass_local_fn_summary,
26781 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
26782 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
26783 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
26784 make_pass_ipa_fn_summary): New.
26785 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
26786 inline_write_summary, inline_free_summary): Remove.
26787 (ipa_free_fn_summary) : New.
26788 * ipa-inline.c (ipa_inline): Update.
26789 (pass_ipa_inline): Do not generate summaries.
26790 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
26792 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
26793 and add pass_ipa_fn_summary.
26794 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
26796 (make_pass_inline_parameters): Remove.
26798 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
26800 * omp-low.c (struct omp_context): Remove "default_kind" member.
26803 * omp-offload.c (execute_oacc_device_lower): Remove the
26804 parallelism dimensions function attributes for unparallelized
26805 OpenACC kernels constructs.
26807 2017-05-23 Martin Liska <mliska@suse.cz>
26809 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
26811 (cgraph_edge::make_speculative): Likewise.
26812 (cgraph_edge::resolve_speculation): Likewise.
26813 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
26814 (cgraph_node::dump): Likewise.
26815 * cgraph.h: Likewise.
26816 * cgraphunit.c (analyze_functions): Likewise.
26817 (symbol_table::compile): Likewise.
26818 * ipa-cp.c (print_all_lattices): Likewise.
26819 (determine_versionability): Likewise.
26820 (initialize_node_lattices): Likewise.
26821 (ipcp_verify_propagated_values): Likewise.
26822 (estimate_local_effects): Likewise.
26823 (update_profiling_info): Likewise.
26824 (create_specialized_node): Likewise.
26825 (perhaps_add_new_callers): Likewise.
26826 (decide_about_value): Likewise.
26827 (decide_whether_version_node): Likewise.
26828 (identify_dead_nodes): Likewise.
26829 (ipcp_store_bits_results): Likewise.
26830 * ipa-devirt.c (dump_targets): Likewise.
26831 (ipa_devirt): Likewise.
26832 * ipa-icf.c (sem_item::dump): Likewise.
26833 (sem_function::equals): Likewise.
26834 (sem_variable::equals): Likewise.
26835 (sem_item_optimizer::read_section): Likewise.
26836 (sem_item_optimizer::execute): Likewise.
26837 (congruence_class::dump): Likewise.
26838 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
26839 (dump_inline_summary): Likewise.
26840 (estimate_node_size_and_time): Likewise.
26841 (inline_analyze_function): Likewise.
26842 * ipa-inline-transform.c (inline_call): Likewise.
26843 * ipa-inline.c (report_inline_failed_reason): Likewise.
26844 (want_early_inline_function_p): Likewise.
26845 (edge_badness): Likewise.
26846 (update_edge_key): Likewise.
26847 (inline_small_functions): Likewise.
26848 * ipa-profile.c (ipa_profile): Likewise.
26849 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
26850 (ipa_make_edge_direct_to_target): Likewise.
26851 (remove_described_reference): Likewise.
26852 (ipa_impossible_devirt_target): Likewise.
26853 (propagate_controlled_uses): Likewise.
26854 (ipa_print_node_params): Likewise.
26855 (ipcp_transform_function): Likewise.
26856 * ipa-pure-const.c (pure_const_read_summary): Likewise.
26857 (propagate_pure_const): Likewise.
26858 * ipa-reference.c (generate_summary): Likewise.
26859 (read_write_all_from_decl): Likewise.
26860 (propagate): Likewise.
26861 (ipa_reference_read_optimization_summary): Likewise.
26862 * ipa-utils.c (ipa_merge_profiles): Likewise.
26863 * ipa.c (walk_polymorphic_call_targets): Likewise.
26864 (symbol_table::remove_unreachable_nodes): Likewise.
26865 (ipa_single_use): Likewise.
26866 * passes.c (execute_todo): Likewise.
26867 * predict.c (drop_profile): Likewise.
26868 * symtab.c (symtab_node::get_dump_name): New function.
26869 (symtab_node::dump_name): Likewise.
26870 (symtab_node::dump_asm_name): Likewise.
26871 (symtab_node::dump_references): Likewise.
26872 (symtab_node::dump_referring): Likewise.
26873 (symtab_node::dump_base): Likewise.
26874 (symtab_node::debug_symtab): Likewise.
26875 * tree-sra.c (convert_callers_for_node): Likewise.
26876 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
26877 * value-prof.c (init_node_map): Likewise.
26879 2017-05-23 Martin Liska <mliska@suse.cz>
26881 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
26882 and symtab_node::debug_symtab to symbol_table::debug.
26883 * cgraphunit.c (analyze_functions): Use the renamed function.
26884 (symbol_table::compile): Likewise.
26885 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
26886 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
26887 * passes.c (execute_todo): Likewise.
26888 * symtab.c (symbol_table::dump): New function.
26889 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
26891 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26893 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
26894 that nonconst implies exec.
26896 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26898 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
26899 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
26900 (inline_edge_summary_vec): Turn into ...
26901 (ipa_call_summaries): ... this one.
26902 (redirect_to_unreachable, edge_set_predicate,
26903 evaluate_properties_for_edge, inline_summary_alloc,
26904 reset_ipa_call_summary, reset_inline_summary,
26905 inline_summary_t::duplicate): Update.
26906 (inline_edge_duplication_hook): Turn to ...
26907 (ipa_call_summary_t::duplicate): ... this one.
26908 (inline_edge_removal_hook): Turn to ...
26909 (ipa_call_summary_t::remove): ... this one.
26910 (dump_inline_edge_summary): Turn to ...
26911 (dump_ipa_call_summary): ... this one.
26912 (estimate_function_body_sizes): Update.
26913 (inline_update_callee_summaries): Update.
26914 (remap_edge_change_prob): Update.
26915 (remap_edge_summaries): Update.
26916 (inline_merge_summary): Update.
26917 (do_estimate_edge_time): Update.
26918 (inline_generate_summary): Update.
26919 (inline_read_section): Update.
26920 (inline_read_summary): Update.
26921 (inline_free_summary): Update.
26922 * ipa-inline.c (can_inline_edge_p): Update.
26923 (compute_inlined_call_time): Update.
26924 (want_inline_small_function_p): Update.
26925 (edge_badness): Update.
26926 (early_inliner): Update.
26927 * ipa-inline.h (inline_edge_summary): Turn to ...
26928 (ipa_call_summary): ... this one.
26929 (ipa_call_summary_t): New class.
26930 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
26931 (ipa_call_summaries): New.
26932 (inline_edge_summary): Remove.
26933 (estimate_edge_growth): Update.
26934 * ipa-profile.c (ipa_propagate_frequency_1): Update.
26935 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26936 * ipa-split.c (execute_split_functions): Update.
26937 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
26939 2017-05-23 Tom de Vries <tom@codesourcery.com>
26941 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
26942 attributes): Document rdrand effective target.
26944 2017-05-23 Tom de Vries <tom@codesourcery.com>
26946 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
26947 attributes): Sort alphabetically.
26949 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
26951 * config/avr/genmultilib.awk: Use gsub instead of gensub.
26953 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
26956 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
26957 V2DF/V2DI splat into two separate patterns, one that handles
26958 registers, and the other that only handles memory. Drop support
26959 for splatting from a GPR on ISA 2.07 and then splitting the
26960 splat into direct move and splat.
26961 (vsx_splat_<mode>_reg): Likewise.
26962 (vsx_splat_<mode>_mem): Likewise.
26964 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
26966 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
26968 2017-05-22 Jakub Jelinek <jakub@redhat.com>
26970 PR middle-end/80809
26971 * omp-low.c (finish_taskreg_remap): New function.
26972 (finish_taskreg_scan): If unit size of ctx->record_type
26973 is non-constant, unshare the size expression and replace
26974 decls in it with possible outer var refs.
26976 PR middle-end/80809
26977 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
26978 GOVD_SHARED rather than GOVD_PRIVATE with it.
26979 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
26980 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
26982 PR middle-end/80853
26983 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
26984 as last argument to build_outer_var_ref for pointer bases of array
26985 section reductions.
26987 2017-05-19 Martin Sebor <msebor@redhat.com>
26989 * print-tree.c (print_node): Print DECL_READ_P flag.
26991 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26993 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
26994 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
26995 * cgraph.c: Likewise.
26996 * cgraphunit.c: Likewise.
26997 * gengtype.c: Likewise.
26998 * ipa-cp.c: Likewise.
26999 * ipa-devirt.c: Likewise.
27000 * ipa-icf.c: Likewise.
27001 * ipa-predicate.c: Likewise.
27002 * ipa-profile.c: Likewise.
27003 * ipa-prop.c: Likewise.
27004 * ipa-split.c: Likewise.
27006 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
27007 edge_predicate_pool, dump_inline_hints,
27008 inline_summary::account_size_time, redirect_to_unreachable,
27009 edge_set_predicate, set_hint_predicate,
27010 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
27011 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
27012 inline_summary_t::remove, remap_hint_predicate_after_duplication,
27013 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
27014 ipa_call_summary_t::remove, initialize_growth_caches,
27015 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
27016 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
27017 mark_modified, unmodified_parm_1, unmodified_parm,
27018 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
27019 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
27020 compute_bb_predicates, will_be_nonconstant_expr_predicate,
27021 will_be_nonconstant_predicate, record_modified_bb_info,
27022 get_minimal_bb, record_modified, param_change_prob,
27023 phi_result_unknown_predicate, predicate_for_phi_result,
27024 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
27025 estimate_function_body_sizes, compute_inline_parameters,
27026 compute_inline_parameters_for_curren, pass_data_inline_parameters,
27027 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
27028 inline_update_callee_summaries, remap_edge_change_prob,
27029 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
27030 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
27031 inline_analyze_function, inline_summary_t::insert,
27032 inline_generate_summary, read_ipa_call_summary, inline_read_section,
27033 inline_read_summary, write_ipa_call_summary, inline_write_summary,
27034 inline_free_summary): Move to ipa-fnsummary.h
27035 (predicate_t): Remove.
27036 * ipa-fnsummary.c: New file.
27037 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
27038 (enum inline_hints_vals, inline_hints, agg_position_info,
27039 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
27040 inline_summaries, ipa_call_summary, ipa_call_summary_t,
27041 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
27042 dump_inline_summary, dump_inline_hints, inline_generate_summary,
27043 inline_read_summary, inline_write_summary, inline_free_summary,
27044 inline_analyze_function, initialize_inline_failed,
27045 inline_merge_summary, inline_update_overall_summary,
27046 compute_inline_parameters): Move to ipa-fnsummary.h
27047 * ipa-fnsummary.h: New file.
27048 * ipa-inline-transform.h: Include ipa-inline.h.
27049 * ipa-inline.c: LIkewise.
27051 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27053 * ipa-inline.c (edge_badness): Use inlined_time instead of
27054 inline_summaries->get.
27056 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27058 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
27060 2017-05-22 Nathan Sidwell <nathan@acm.org>
27062 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
27063 (fdump-lang): Document 'raw' option.
27064 * dumpfile.h (TDI_tu): Delete.
27065 * dumpfile.c (dump_files): Remove translation-unit.
27066 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
27068 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
27070 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
27071 command option from $(AWK) call.
27072 * config/avr/genmultilib.awk: Simplify and rewrite so that it
27073 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
27074 [FORMAT]: Remove handling of variable.
27075 * config/avr/t-multilib: Regenerate.
27077 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27079 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
27081 (dump_inline_summary): Do not print self_time.
27082 (estimate_function_body_sizes): Do not set self_time.
27083 (compute_inline_parameters): Likewise.
27084 (inline_read_section, inline_write_summary): Do not stream self_time.
27085 * ipa-inline.h (inline_summary): Drop self_time.
27087 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27089 * ipa-inline-analysis.c (account_size_time): Rename to ...
27090 (inline_summary::account_size_time): ... this one.
27091 (reset_ipa_call_summary): Turn to ...
27092 (ipa_call_summary::reset): ... this one.
27093 (reset_inline_summary): Turn to ...
27094 (inline_summary::reset): ... this one.
27095 (inline_summary_t::remove): Update.
27096 (inline_summary_t::duplicate): Update.
27097 (ipa_call_summary_t::remove): Update.
27098 (dump_inline_summary): Update.
27099 (estimate_function_body_sizes): Update.
27100 (compute_inline_parameters): Update.
27101 (estimate_node_size_and_time): Update.
27102 (inline_merge_summary): Update.
27103 (inline_update_overall_summary): Update.
27104 (inline_read_section): Update.
27105 (inline_write_summary): Update.
27106 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
27107 add account_size_time and reset member functions.
27108 (ipa_call_summary): Add reset function.
27109 * ipa-predicate.h (predicate::operator &): Constify.
27111 2017-05-22 Richard Biener <rguenther@suse.de>
27113 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
27115 2017-05-19 Jason Merrill <jason@redhat.com>
27117 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
27119 2017-05-19 Marek Polacek <polacek@redhat.com>
27122 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
27123 TYPE_OVERFLOW_WRAPS checks.
27125 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
27127 * tree-core.h (enum omp_clause_default_kind): Add
27128 "OMP_CLAUSE_DEFAULT_PRESENT".
27129 * tree-pretty-print.c (dump_omp_clause): Handle it.
27130 * gimplify.c (enum gimplify_omp_var_data): Add
27131 "GOVD_MAP_FORCE_PRESENT".
27132 (gimplify_adjust_omp_clauses_1): Map it to
27133 "GOMP_MAP_FORCE_PRESENT".
27134 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
27136 * gimplify.c (oacc_default_clause): Clarify.
27138 2017-05-19 Nathan Sidwell <nathan@acm.org>
27140 LANG_HOOK_REGISTER_DUMPS
27141 * toplev.c (general_init): Call register dump lang hook.
27142 * doc/invoke.texi: Document -fdump-lang option family.
27143 * dumpfile.c (dump_files): Remove class dump here.
27144 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
27145 * dumpfile.h (tree_dump_index): Remove TDI_class.
27146 * langhooks-def.h (lhd_register_dumps): Declare.
27147 (LANG_HOOKS_REGISTER_DUMPS): Define.
27148 (LANG_HOOKS_INITIALIZER): Add it.
27149 * langhooks.c (lhd_register_dumps): Define.
27150 * langhooks.h (struct lang_hooks): Add register_dumps.
27152 2017-05-19 Nathan Sidwell <nathan@acm.org>
27154 * context.h (context::set_passes): New.
27155 * context.c (context::context): Do not create pass manager.
27156 * toplev.c (general_init): Create pass manager here.
27158 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
27160 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
27161 use this splitter if two add or or instructions would also work for
27162 the constant we want to generate.
27164 2017-05-19 Richard Biener <rguenther@suse.de>
27167 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
27168 predicate evaluation.
27170 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27172 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
27174 * ipa-inline.c (want_inline_small_function_p): Do not cast to
27177 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27179 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
27180 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27181 (inline_edge_summary_vec): Turn into ...
27182 (ipa_call_summaries): ... this one.
27183 (redirect_to_unreachable, edge_set_predicate,
27184 evaluate_properties_for_edge, inline_summary_alloc,
27185 reset_ipa_call_summary, reset_inline_summary,
27186 inline_summary_t::duplicate): Update.
27187 (inline_edge_duplication_hook): Turn to ...
27188 (ipa_call_summary_t::duplicate): ... this one.
27189 (inline_edge_removal_hook): Turn to ...
27190 (ipa_call_summary_t::remove): ... this one.
27191 (dump_inline_edge_summary): Turn to ...
27192 (dump_ipa_call_summary): ... this one.
27193 (estimate_function_body_sizes): Update.
27194 (inline_update_callee_summaries): Update.
27195 (remap_edge_change_prob): Update.
27196 (remap_edge_summaries): Update.
27197 (inline_merge_summary): Update.
27198 (do_estimate_edge_time): Update.
27199 (inline_generate_summary): Update.
27200 (inline_read_section): Update.
27201 (inline_read_summary): Update.
27202 (inline_free_summary): Update.
27203 * ipa-inline.c (can_inline_edge_p): Update.
27204 (compute_inlined_call_time): Update.
27205 (want_inline_small_function_p): Update.
27206 (edge_badness): Update.
27207 (early_inliner): Update.
27208 * ipa-inline.h (inline_edge_summary): Turn to ...
27209 (ipa_call_summary): ... this one.
27210 (ipa_call_summary_t): New class.
27211 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27212 (ipa_call_summaries): New.
27213 (inline_edge_summary): Remove.
27214 (estimate_edge_growth): Update.
27215 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27216 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27217 * ipa-split.c (execute_split_functions): Update.
27218 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27220 2017-05-19 Richard Biener <rguenther@suse.de>
27222 PR middle-end/80764
27223 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
27225 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
27227 * config/rs6000/rs6000.c (struct machine_function): Add field
27228 fpr_is_wrapped_separately.
27229 (rs6000_get_separate_components): Use 64 components. Handle the
27230 new FPR components.
27231 (rs6000_components_for_bb): Handle the FPR components.
27232 (rs6000_emit_prologue_components): Handle the FPR components.
27233 (rs6000_emit_epilogue_components): Handle the FPR components.
27234 (rs6000_set_handled_components): Handle the FPR components.
27235 (rs6000_emit_prologue): Don't output prologue code for those FPRs
27236 that are already separately shrink-wrapped.
27237 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
27238 that are already separately shrink-wrapped.
27240 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27243 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
27246 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
27247 (define_peephole2 for Altivec d-form load): Add peepholes to catch
27248 cases where the register allocator uses a move and an offsettable
27249 memory operation to/from a FPR register on ISA 2.06/2.07.
27250 (define_peephole2 for Altivec d-form store): Likewise.
27252 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
27255 * config/i386/mmx.md (*mov<mode>_internal): Enable
27256 alternatives 11, 12, 13 and 14 also for 32bit targets.
27257 Remove alternatives 15, 16, 17 and 18.
27258 * config/i386/sse.md (vec_concatv2di): Change
27259 alternative (!x, *y) to (x, ?!*Yn).
27261 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
27263 * dumpfile.h (enum dump_kind): Remove stray comma.
27265 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27267 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
27268 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
27269 predicate::num_conditions
27270 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
27271 (CHANGED): turn into predicate::changed.
27272 (agg_position_info): Move to ipa-predicate.h
27273 (add_condition, predicate::add_clause, predicate::operator &=,
27274 predicate::or_with, predicate::evaluate, predicate::probability,
27275 dump_condition, dump_clause, predicate::dump,
27276 predicate::remap_after_duplication, predicate::remap_after_inlining,
27277 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
27278 (evaluate_conditions_for_known_args): Update.
27279 (set_cond_stmt_execution_predicate): Update.
27280 * ipa-inline.h: Include ipa-predicate.h
27281 (condition, inline_param_summary, conditions, agg_position_info,
27282 predicate): Move to ipa-predicate.h
27283 * ipa-predicate.c: New file.
27284 * ipa-predicate.h: New file.
27286 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
27288 * final.c (leaf_function_p): Check we are not in a sequence.
27290 2017-05-18 Martin Liska <mliska@suse.cz>
27292 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
27293 * dumpfile.c (dump_register): Use new enum dump_kind.
27294 (get_dump_file_name): Likewise.
27295 (dump_enable_all): Likewise.
27296 (dump_switch_p_1): Likewise.
27297 (enable_rtl_dump_file): Remove usage of TDF_RTL.
27298 * dumpfile.h (enum dump_kind): New enum type.
27299 (struct dump_file_info): Create constructor and
27300 format fields and comments.
27301 * passes.c (pass_manager::register_one_dump_file):
27303 * statistics.c (statistics_early_init): Likewise.
27304 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
27305 TDF_TREE with TDF_SLIM.
27306 (gather_memory_references_ref): Likewise.
27308 2017-05-18 Martin Liska <mliska@suse.cz>
27310 * vec.h (struct vnull): Use it.
27312 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27314 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
27315 (true_predicate, false_predicate, true_predicate_p,
27316 false_predicate_p): Remove.
27317 (single_cond_predicate, not_inlined_predicate): Turn to member function
27319 (add_condition): Update.
27320 (add_clause): Turn to...
27321 (predicate::add_clause): ... this one; update; allow passing NULL
27323 (and_predicates): Turn to ...
27324 (predicate::operator &=): ... this one.
27325 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
27326 (or_predicates): Turn to ...
27327 (predicate::or_with): ... this one.
27328 (evaluate_predicate): Turn to ...
27329 (predicate::evaluate): ... this one.
27330 (predicate_probability): Turn to ...
27331 (predicate::probability): ... this one.
27332 (dump_condition): Update.
27333 (dump_predicate): Turn to ...
27334 (predicate::dump): ... this one.
27335 (account_size_time): Update.
27336 (edge_set_predicate): Update.
27337 (set_hint_predicate): UPdate.
27338 (evaluate_conditions_for_known_args): Update.
27339 (evaluate_properties_for_edge): Update.
27340 (remap_predicate_after_duplication): Turn to...
27341 (predicate::remap_after_duplication): ... this one.
27342 (remap_hint_predicate_after_duplication): Update.
27343 (inline_summary_t::duplicate): UPdate.
27344 (dump_inline_edge_summary): Update.
27345 (dump_inline_summary): Update.
27346 (set_cond_stmt_execution_predicate): Update.
27347 (set_switch_stmt_execution_predicate): Update.
27348 (compute_bb_predicates): Update.
27349 (will_be_nonconstant_expr_predicate): Update.
27350 (will_be_nonconstant_predicate): Update.
27351 (phi_result_unknown_predicate): Update.
27352 (predicate_for_phi_result): Update.
27353 (array_index_predicate): Update.
27354 (estimate_function_body_sizes): Update.
27355 (estimate_node_size_and_time): Update.
27356 (estimate_ipcp_clone_size_and_time): Update.
27357 (remap_predicate): Rename to ...
27358 (predicate::remap_after_inlining): ... this one.
27359 (remap_hint_predicate): Update.
27360 (inline_merge_summary): Update.
27361 (inline_update_overall_summary): Update.
27362 (estimate_size_after_inlining): Update.
27363 (read_predicate): Rename to ...
27364 (predicate::stream_in): ... this one.
27365 (read_inline_edge_summary): Update.
27366 (write_predicate): Rename to ...
27367 (predicate::stream_out): ... this one.
27368 (write_inline_edge_summary): Update.
27369 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
27370 (clause_t): Turn to uint32_t
27371 (predicate): Turn to class; implement constructor and operators
27373 (size_time_entry): Update.
27374 (inline_summary): Update.
27375 (inline_edge_summary): Update.
27377 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
27379 * fold-const.c (fold_binary_loc): Move transformation...
27380 * match.pd (C - X CMP X): ... here.
27382 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
27384 * config/sparc/sparc.c (sparc_option_override): Set function
27385 alignment for -mcpu=niagara7 to 64 to match the I$ line.
27386 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
27388 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
27390 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
27392 2017-05-18 Marek Polacek <polacek@redhat.com>
27395 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
27396 (pass_ubsan::execute): Call gimple_assign_single_p instead of
27397 gimple_assign_load_p.
27399 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
27401 PR middle-end/80692
27402 * real.c (do_compare): Give decimal_do_compare preference over
27403 comparing just the signs.
27405 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
27407 * doc/md.texi (Canonicalization of Instructions): Describe the
27408 canonical form of instructions that inherently set a condition
27411 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
27413 PR middle-end/80775
27414 * tree-cfg.c: Move deletion of unreachable case statements to after
27415 the merging of consecutive case labels.
27417 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
27419 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
27420 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
27421 restoring of callee-saved registers.
27423 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
27425 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
27426 * config/visium/visium.c (single_set_and_flags): Likewise.
27427 * config/visium/visium.md (Substitutions): Likewise.
27429 2017-05-17 Martin Liska <mliska@suse.cz>
27431 * cfg.c: Introduce dump_flags_t type and
27432 use it instead of int type.
27434 * cfghooks.c: Likewise.
27435 * cfghooks.h (struct cfg_hooks): Likewise.
27436 * cfgrtl.c: Likewise.
27437 * cfgrtl.h: Likewise.
27438 * cgraph.c (cgraph_node::get_body): Likewise.
27439 * coretypes.h: Likewise.
27440 * domwalk.c: Likewise.
27441 * domwalk.h: Likewise.
27442 * dumpfile.c (struct dump_option_value_info): Likewise.
27443 (dump_enable_all): Likewise.
27444 (dump_switch_p_1): Likewise.
27445 (opt_info_switch_p): Likewise.
27446 * dumpfile.h (enum tree_dump_index): Likewise.
27447 (struct dump_file_info): Likewise.
27448 * genemit.c: Likewise.
27449 * generic-match-head.c: Likewise.
27450 * gengtype.c (open_base_files): Likewise.
27451 * gimple-pretty-print.c: Likewise.
27452 * gimple-pretty-print.h: Likewise.
27453 * graph.c (print_graph_cfg): Likewise.
27454 * graphite-scop-detection.c (dot_all_sese): Likewise.
27455 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
27456 * loop-unroll.c (report_unroll): Likewise.
27457 * passes.c (pass_manager::register_one_dump_file): Likewise.
27458 * print-tree.c: Likewise.
27459 * statistics.c: Likewise.
27460 * tree-cfg.c: Likewise.
27461 * tree-cfg.h: Likewise.
27462 * tree-dfa.c: Likewise.
27463 * tree-dfa.h: Likewise.
27464 * tree-dump.c (dump_function): Likewise.
27465 * tree-dump.h (struct dump_info): Likewise.
27466 * tree-pretty-print.c: Likewise.
27467 * tree-pretty-print.h: Likewise.
27468 * tree-ssa-live.c: Likewise.
27469 * tree-ssa-live.h: Likewise.
27470 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
27471 * tree-vect-loop.c: Likewise.
27472 * tree-vect-slp.c: Likewise.
27474 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
27475 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27477 PR tree-optimization/80457
27478 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
27479 of all arguments to a statement as scalar_to_vec operations.
27480 (vectorizable_call): Adjust call to vect_model_simple_cost for
27482 (vectorizable_conversion): Likewise.
27483 (vectorizable_assignment): Likewise.
27484 (vectorizable_shift): Likewise.
27485 (vectorizable_operation): Likewise.
27486 (vectorizable_comparison): Likewise.
27487 (vect_is_simple_cond): Record the def types for operands.
27488 (vectorizable_condition): Likewise, call vect_model_simple_cost.
27489 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
27490 for statement argument count.
27492 2017-05-16 Carl Love <cel@us.ibm.com>
27494 * config/rs6000/rs6000-c: Add support for built-in functions
27495 vector unsigned long long vec_bperm (vector unsigned long long,
27496 vector unsigned char)
27497 vector signed long long vec_mule (vector signed int,
27499 vector unsigned long long vec_mule (vector unsigned int,
27500 vector unsigned int)
27501 vector signed long long vec_mulo (vector signed int,
27503 vector unsigned long long vec_mulo (vector unsigned int,
27504 vector unsigned int)
27505 vector signed char vec_sldw (vector signed char,
27506 vector signed char,
27508 vector unsigned char vec_sldw (vector unsigned char,
27509 vector unsigned char,
27511 vector signed short vec_sldw (vector signed short,
27512 vector signed short,
27514 vector unsigned short vec_sldw (vector unsigned short,
27515 vector unsigned short,
27517 vector signed int vec_sldw (vector signed int,
27520 vector unsigned int vec_sldw (vector unsigned int,
27521 vector unsigned int,
27523 vector signed long long vec_sldw (vector signed long long,
27524 vector signed long long,
27526 vector unsigned long long vec_sldw (vector unsigned long long,
27527 vector unsigned long long,
27529 * config/rs6000/rs6000-c: Add support for built-in functions
27530 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
27531 * config/rs6000/altivec.h: Add defintion for vec_sldw.
27532 * doc/extend.texi: Update the built-in documentation for the
27533 new built-in functions.
27535 2017-05-16 Marek Polacek <polacek@redhat.com>
27539 * tree.c (save_expr): Don't fold the expression.
27541 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
27543 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
27544 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
27545 and (?*y,m). Update insn attributes.
27547 2017-05-16 Martin Liska <mliska@suse.cz>
27549 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
27550 flags argument of print_gimple_stmt, print_gimple_expr,
27551 print_generic_stmt and print_generic_expr.
27552 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
27553 * coretypes.h: Likewise.
27554 * except.c (dump_eh_tree): Likewise.
27555 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
27556 * gimple-pretty-print.h: Likewise.
27557 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
27558 (backprop::push_to_worklist): Likewise.
27559 (backprop::pop_from_worklist): Likewise.
27560 (backprop::process_use): Likewise.
27561 (backprop::intersect_uses): Likewise.
27562 (note_replacement): Likewise.
27563 * gimple-ssa-store-merging.c
27564 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
27565 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
27566 (pass_store_merging::execute): Likewise.
27567 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
27568 (ssa_base_cand_dump_callback): Likewise.
27569 (dump_incr_vec): Likewise.
27570 (replace_refs): Likewise.
27571 (replace_mult_candidate): Likewise.
27572 (create_add_on_incoming_edge): Likewise.
27573 (create_phi_basis): Likewise.
27574 (insert_initializers): Likewise.
27575 (all_phi_incrs_profitable): Likewise.
27576 (introduce_cast_before_cand): Likewise.
27577 (replace_one_candidate): Likewise.
27578 * gimplify.c (gimplify_expr): Likewise.
27579 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
27580 (set_rename): Likewise.
27581 (rename_uses): Likewise.
27582 (copy_loop_phi_nodes): Likewise.
27583 (add_close_phis_to_merge_points): Likewise.
27584 (copy_loop_close_phi_args): Likewise.
27585 (copy_cond_phi_args): Likewise.
27586 (graphite_copy_stmts_from_block): Likewise.
27587 (translate_pending_phi_nodes): Likewise.
27588 * graphite-poly.c (print_pdr): Likewise.
27589 (dump_gbb_cases): Likewise.
27590 (dump_gbb_conditions): Likewise.
27591 (print_scop_params): Likewise.
27592 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
27593 (build_cross_bb_scalars_use): Likewise.
27594 (gather_bbs::before_dom_children): Likewise.
27595 * hsa-dump.c (dump_hsa_immed): Likewise.
27596 * ipa-cp.c (print_ipcp_constant_value): Likewise.
27597 (get_replacement_map): Likewise.
27598 * ipa-inline-analysis.c (dump_condition): Likewise.
27599 (estimate_function_body_sizes): Likewise.
27600 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
27601 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
27602 * ipa-prop.c (ipa_dump_param): Likewise.
27603 (ipa_print_node_jump_functions_for_edge): Likewise.
27604 (ipa_modify_call_arguments): Likewise.
27605 (ipa_modify_expr): Likewise.
27606 (ipa_dump_param_adjustments): Likewise.
27607 (ipa_dump_agg_replacement_values): Likewise.
27608 (ipcp_modif_dom_walker::before_dom_children): Likewise.
27609 * ipa-pure-const.c (check_stmt): Likewise.
27610 (pass_nothrow::execute): Likewise.
27611 * ipa-split.c (execute_split_functions): Likewise.
27612 * omp-offload.c (dump_oacc_loop_part): Likewise.
27613 (dump_oacc_loop): Likewise.
27614 * trans-mem.c (tm_log_emit): Likewise.
27615 (tm_memopt_accumulate_memops): Likewise.
27616 (dump_tm_memopt_set): Likewise.
27617 (dump_tm_memopt_transform): Likewise.
27618 * tree-cfg.c (gimple_verify_flow_info): Likewise.
27619 (print_loop): Likewise.
27620 * tree-chkp-opt.c (chkp_print_addr): Likewise.
27621 (chkp_gather_checks_info): Likewise.
27622 (chkp_get_check_result): Likewise.
27623 (chkp_remove_check_if_pass): Likewise.
27624 (chkp_use_outer_bounds_if_possible): Likewise.
27625 (chkp_reduce_bounds_lifetime): Likewise.
27626 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
27627 (chkp_mark_completed_bounds): Likewise.
27628 (chkp_register_incomplete_bounds): Likewise.
27629 (chkp_mark_invalid_bounds): Likewise.
27630 (chkp_maybe_copy_and_register_bounds): Likewise.
27631 (chkp_build_returned_bound): Likewise.
27632 (chkp_get_bound_for_parm): Likewise.
27633 (chkp_build_bndldx): Likewise.
27634 (chkp_get_bounds_by_definition): Likewise.
27635 (chkp_generate_extern_var_bounds): Likewise.
27636 (chkp_get_bounds_for_decl_addr): Likewise.
27637 * tree-chrec.c (chrec_apply): Likewise.
27638 * tree-data-ref.c (dump_data_reference): Likewise.
27639 (dump_subscript): Likewise.
27640 (dump_data_dependence_relation): Likewise.
27641 (analyze_overlapping_iterations): Likewise.
27642 * tree-inline.c (expand_call_inline): Likewise.
27643 (tree_function_versioning): Likewise.
27644 * tree-into-ssa.c (dump_defs_stack): Likewise.
27645 (dump_currdefs): Likewise.
27646 (dump_names_replaced_by): Likewise.
27647 (dump_update_ssa): Likewise.
27648 (update_ssa): Likewise.
27649 * tree-object-size.c (pass_object_sizes::execute): Likewise.
27650 * tree-parloops.c (build_new_reduction): Likewise.
27651 (try_create_reduction_list): Likewise.
27652 (ref_conflicts_with_region): Likewise.
27653 (oacc_entry_exit_ok_1): Likewise.
27654 (oacc_entry_exit_single_gang): Likewise.
27655 * tree-pretty-print.h: Likewise.
27656 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
27657 (get_scalar_evolution): Likewise.
27658 (add_to_evolution): Likewise.
27659 (get_loop_exit_condition): Likewise.
27660 (analyze_evolution_in_loop): Likewise.
27661 (analyze_initial_condition): Likewise.
27662 (analyze_scalar_evolution): Likewise.
27663 (instantiate_scev): Likewise.
27664 (number_of_latch_executions): Likewise.
27665 (gather_chrec_stats): Likewise.
27666 (final_value_replacement_loop): Likewise.
27667 (scev_const_prop): Likewise.
27668 * tree-sra.c (dump_access): Likewise.
27669 (disqualify_candidate): Likewise.
27670 (create_access): Likewise.
27671 (reject): Likewise.
27672 (maybe_add_sra_candidate): Likewise.
27673 (create_access_replacement): Likewise.
27674 (analyze_access_subtree): Likewise.
27675 (analyze_all_variable_accesses): Likewise.
27676 (sra_modify_assign): Likewise.
27677 (initialize_constant_pool_replacements): Likewise.
27678 (find_param_candidates): Likewise.
27679 (decide_one_param_reduction): Likewise.
27680 (replace_removed_params_ssa_names): Likewise.
27681 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
27682 * tree-ssa-copy.c (dump_copy_of): Likewise.
27683 (copy_prop_visit_cond_stmt): Likewise.
27684 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
27685 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
27686 (record_equivalences_from_stmt): Likewise.
27687 * tree-ssa-dse.c (compute_trims): Likewise.
27688 (delete_dead_call): Likewise.
27689 (delete_dead_assignment): Likewise.
27690 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
27691 (forward_propagate_into_cond): Likewise.
27692 (pass_forwprop::execute): Likewise.
27693 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
27694 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
27696 (move_computations_worker): Likewise.
27697 (execute_sm): Likewise.
27698 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
27699 (remove_exits_and_undefined_stmts): Likewise.
27700 (remove_redundant_iv_tests): Likewise.
27701 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
27702 (adjust_iv_update_pos): Likewise.
27703 * tree-ssa-math-opts.c (bswap_replace): Likewise.
27704 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
27705 (value_replacement): Likewise.
27706 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
27707 * tree-ssa-pre.c (print_pre_expr): Likewise.
27708 (get_representative_for): Likewise.
27709 (create_expression_by_pieces): Likewise.
27710 (insert_into_preds_of_block): Likewise.
27711 (eliminate_insert): Likewise.
27712 (eliminate_dom_walker::before_dom_children): Likewise.
27713 (eliminate): Likewise.
27714 (remove_dead_inserted_code): Likewise.
27715 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
27716 * tree-ssa-reassoc.c (get_rank): Likewise.
27717 (eliminate_duplicate_pair): Likewise.
27718 (eliminate_plus_minus_pair): Likewise.
27719 (eliminate_not_pairs): Likewise.
27720 (undistribute_ops_list): Likewise.
27721 (eliminate_redundant_comparison): Likewise.
27722 (update_range_test): Likewise.
27723 (optimize_range_tests_var_bound): Likewise.
27724 (optimize_vec_cond_expr): Likewise.
27725 (rewrite_expr_tree): Likewise.
27726 (rewrite_expr_tree_parallel): Likewise.
27727 (linearize_expr): Likewise.
27728 (break_up_subtract): Likewise.
27729 (linearize_expr_tree): Likewise.
27730 (attempt_builtin_powi): Likewise.
27731 (attempt_builtin_copysign): Likewise.
27732 (transform_stmt_to_copy): Likewise.
27733 (transform_stmt_to_multiply): Likewise.
27734 (dump_ops_vector): Likewise.
27735 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
27736 (print_scc): Likewise.
27737 (set_ssa_val_to): Likewise.
27738 (visit_reference_op_store): Likewise.
27739 (visit_use): Likewise.
27740 (sccvn_dom_walker::before_dom_children): Likewise.
27741 (run_scc_vn): Likewise.
27742 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
27744 (expr_hash_elt::print): Likewise.
27745 (const_and_copies::pop_to_marker): Likewise.
27746 (const_and_copies::record_const_or_copy_raw): Likewise.
27747 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
27748 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
27749 (dump_predicates): Likewise.
27750 (find_uninit_use): Likewise.
27751 (warn_uninitialized_phi): Likewise.
27752 (pass_late_warn_uninitialized::execute): Likewise.
27753 * tree-ssa.c (verify_vssa): Likewise.
27754 (verify_ssa): Likewise.
27755 (maybe_optimize_var): Likewise.
27756 * tree-vrp.c (dump_value_range): Likewise.
27757 (dump_all_value_ranges): Likewise.
27758 (dump_asserts_for): Likewise.
27759 (register_edge_assert_for_2): Likewise.
27760 (vrp_visit_cond_stmt): Likewise.
27761 (vrp_visit_switch_stmt): Likewise.
27762 (vrp_visit_stmt): Likewise.
27763 (vrp_visit_phi_node): Likewise.
27764 (simplify_cond_using_ranges_1): Likewise.
27765 (fold_predicate_in): Likewise.
27766 (evrp_dom_walker::before_dom_children): Likewise.
27767 (evrp_dom_walker::push_value_range): Likewise.
27768 (evrp_dom_walker::pop_value_range): Likewise.
27769 (execute_early_vrp): Likewise.
27771 2017-05-16 Richard Biener <rguenther@suse.de>
27773 * dwarf2out.c (loc_list_from_tree_1): Do not create
27774 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
27776 2017-05-16 Richard Biener <rguenther@suse.de>
27778 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
27780 (note_variable_value_in_expr): If we resolved the decl ref
27781 do not push to the stack.
27783 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
27785 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
27786 operations in fast-math mode.
27787 (vaddq_f16): Likewise.
27788 (vmul_f16): Likewise.
27789 (vmulq_f16): Likewise.
27790 (vsub_f16): Likewise.
27791 (vsubq_f16): Likewise.
27792 * config/arm/neon.md (add<mode>3): New.
27794 (fma:<VH:mode>3): New. Also remove outdated comment.
27797 2017-05-16 Martin Liska <mliska@suse.cz>
27801 * ipa-devirt.c (warn_types_mismatch): Fix typo.
27802 (odr_types_equivalent_p): Likewise.
27804 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
27806 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
27808 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
27811 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
27812 non-interunit SSE move alternatives with '?'.
27813 (zero-extendsidi peephole2): New peephole to skip intermediate
27814 general register in SSE zero-extend sequence.
27816 2017-05-15 Jeff Law <law@redhat.com>
27818 * reorg.c (relax_delay_slots): Create a new variable to hold
27819 the temporary target rather than clobbering TARGET_LABEL.
27821 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
27822 missing argument to extract_bit_field call.
27823 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
27825 2017-05-15 Martin Liska <mliska@suse.cz>
27828 * gcc.c (process_command): Do not allow empty argument of -o option.
27830 2017-05-15 Renlin Li <renlin.li@arm.com>
27832 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
27833 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
27834 * config/aarch64/constraints.md (Usf): Add long call check.
27835 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
27836 (call_value): Likewise.
27837 (sibcall): Likewise.
27838 (sibcall_value): Likewise.
27840 (call_value_insn): New.
27841 (sibcall_insn): Update rtx pattern.
27842 (sibcall_value_insn): Likewise.
27843 (call_internal): Remove.
27844 (call_value_internal): Likewise.
27845 (sibcall_internal): Likewise.
27846 (sibcall_value_internal): Likewise.
27847 (call_reg): Likewise.
27848 (call_symbol): Likewise.
27849 (call_value_reg): Likewise.
27850 (call_value_symbol): Likewise.
27852 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
27855 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
27857 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
27859 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
27860 compatible with CCGOCmode and with CCZmode.
27862 2017-05-14 Martin Sebor <msebor@redhat.com>
27864 PR middle-end/77671
27865 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
27866 (gimple_fold_builtin_snprintf): Same.
27867 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
27868 (gimple_fold_builtin_snprintf): Same.
27869 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
27870 of character types.
27871 (is_call_safe): New function.
27872 (try_substitute_return_value): Call it.
27873 (try_simplify_call): New function.
27874 (pass_sprintf_length::handle_gimple_call): Call it.
27876 2017-05-14 Martin Sebor <msebor@redhat.com>
27878 PR middle-end/80669
27879 * builtins.c (expand_builtin_stpncpy): Simplify.
27881 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
27883 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
27884 * config/i386/i386.h
27885 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
27886 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
27887 (struct machine_function): Add new members call_ms2sysv,
27888 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
27889 (struct machine_frame_state): New fields sp_realigned and
27890 sp_realigned_offset.
27891 * config/i386/i386.c
27892 (enum xlogue_stub): New enum.
27893 (enum xlogue_stub_sets): New enum.
27894 (class xlogue_layout): New class.
27895 (struct ix86_frame): New fields stack_realign_allocate_offset,
27896 stack_realign_offset and outlined_save_offset. Modify comments to
27897 detail stack layout when using out-of-line stubs.
27898 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
27899 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
27900 -mcall-ms2sysv-xlogues.
27901 (stub_managed_regs): New static variable.
27902 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
27903 registers managed by out-of-line stub.
27904 (disable_call_ms2sysv_xlogues): New function.
27905 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
27906 m->call_ms2sysv when appropriate and compute frame layout for
27908 (sp_valid_at, fp_valid_at): New inline functions.
27909 (choose_basereg): New function.
27910 (choose_baseaddr): Add align parameter, use choose_basereg and modify
27912 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
27913 Use align parameter of choose_baseaddr to generated aligned SSE movs
27915 (pro_epilogue_adjust_stack): Modify to track
27916 machine_frame_state::sp_realigned.
27917 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
27918 (ix86_nsaved_sseregs): Likewise.
27919 (ix86_emit_save_regs): Likewise.
27920 (ix86_emit_save_regs_using_mov): Likewise.
27921 (ix86_emit_save_sse_regs_using_mov): Likewise.
27922 (get_scratch_register_on_entry): Likewise.
27923 (gen_frame_set): New function.
27924 (gen_frame_load): Likewise.
27925 (gen_frame_store): Likewise.
27926 (emit_outlined_ms2sysv_save): Likewise.
27927 (emit_outlined_ms2sysv_restore): Likewise.
27928 (ix86_expand_prologue): Modify stack re-alignment code and call
27929 emit_outlined_ms2sysv_save when appropriate.
27930 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
27931 parameter rtx_insn *insn, which allows the function to be used to only
27932 generate the notes.
27933 (ix86_expand_epilogue): Modify validity checks of frame and stack
27934 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
27935 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
27936 * config/i386/predicates.md
27937 (save_multiple): New predicate.
27938 (restore_multiple): Likewise.
27939 * config/i386/sse.md
27940 (save_multiple<mode>): New pattern.
27941 (save_multiple_realign<mode>): Likewise.
27942 (restore_multiple<mode>): Likewise.
27943 (restore_multiple_and_return<mode>): Likewise.
27944 (restore_multiple_leave_return<mode>): Likewise.
27945 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
27947 2017-05-14 Julia Koval <julia.koval@intel.com>
27949 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
27950 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
27951 (__builtin_ia32_xsetbv): New builtins.
27952 * config/i386/i386.c (ix86_expand_special_args_builtin):
27954 (ix86_expand_builtin): Special expand for new intrinsics.
27955 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
27956 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
27957 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
27959 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27961 * cfganal.c (inverted_post_order_compute): Change argument type
27963 * cfganal.h (inverted_post_order_compute): Adjust prototype.
27964 * df-core.c (rest_of_handle_df_initialize): Adjust.
27965 (rest_of_handle_df_finish): Likewise.
27966 (df_analyze_1): Likewise.
27967 (df_analyze): Likewise.
27968 (loop_inverted_post_order_compute): Change argument to be a vec *.
27969 (df_analyze_loop): Adjust.
27970 (df_get_n_blocks): Likewise.
27971 (df_get_postorder): Likewise.
27972 * df.h (struct df_d): Change field to be a vec.
27973 * lcm.c (compute_laterin): Adjust.
27974 (compute_available): Likewise.
27975 * lra-lives.c (lra_create_live_ranges_1): Likewise.
27976 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
27977 * tree-ssa-pre.c (compute_antic): Likewise.
27979 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27981 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
27982 (depth_first_search::depth_first_search): Change structure init
27983 function to this constructor.
27984 (depth_first_search::add_bb): Rename function to this member.
27985 (depth_first_search::execute): Likewise.
27986 (flow_dfs_compute_reverse_finish): Adjust.
27988 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27990 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
27991 (longest_simple_path): Likewise.
27992 * shrink-wrap.c (spread_components): Likewise.
27993 (disqualify_problematic_components): Likewise.
27994 (emit_common_heads_for_components): Likewise.
27995 (emit_common_tails_for_components): Likewise.
27996 (insert_prologue_epilogue_for_components): Likewise.
27998 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28000 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
28003 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28005 * df-core.c (df_set_blocks): Start using auto_bitmap.
28006 (df_compact_blocks): Likewise.
28007 * df-problems.c (df_rd_confluence_n): Likewise.
28008 * df-scan.c (df_insn_rescan_all): Likewise.
28009 (df_process_deferred_rescans): Likewise.
28010 (df_update_entry_block_defs): Likewise.
28011 (df_update_exit_block_uses): Likewise.
28012 (df_entry_block_bitmap_verify): Likewise.
28013 (df_exit_block_bitmap_verify): Likewise.
28014 (df_scan_verify): Likewise.
28015 * lra-constraints.c (lra_constraints): Likewise.
28016 (undo_optional_reloads): Likewise.
28017 (lra_undo_inheritance): Likewise.
28018 * lra-remat.c (calculate_gen_cands): Likewise.
28019 (do_remat): Likewise.
28020 * lra-spills.c (assign_spill_hard_regs): Likewise.
28021 (spill_pseudos): Likewise.
28022 * tree-ssa-pre.c (bitmap_set_and): Likewise.
28023 (bitmap_set_subtract_values): Likewise.
28025 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28027 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
28028 management with auto_bitmap.
28029 (fix_inter_tick): Likewise.
28030 (fix_recovery_deps): Likewise.
28031 * ira.c (add_store_equivs): Likewise.
28032 (find_moveable_pseudos): Likewise.
28033 (split_live_ranges_for_shrink_wrap): Likewise.
28034 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
28035 (rtx_reuse_manager::seen_def_p): Likewise.
28036 (rtx_reuse_manager::set_seen_def): Likewise.
28037 * print-rtl.h (class rtx_reuse_manager): Likewise.
28039 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28041 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
28043 (migrate_btr_def): Likewise.
28044 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
28045 * df-core.c (loop_post_order_compute): Likewise.
28046 (loop_inverted_post_order_compute): Likewise.
28047 * hsa-common.h: Likewise.
28048 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
28049 * init-regs.c (initialize_uninitialized_regs): Likewise.
28050 * ipa-inline.c (resolve_noninline_speculation): Likewise.
28051 (inline_small_functions): Likewise.
28052 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
28053 * ira.c (combine_and_move_insns): Likewise.
28054 (build_insn_chain): Likewise.
28055 * loop-invariant.c (find_invariants): Likewise.
28056 * lower-subreg.c (propagate_pseudo_copies): Likewise.
28057 * predict.c (tree_predict_by_opcode): Likewise.
28058 (predict_paths_leading_to): Likewise.
28059 (predict_paths_leading_to_edge): Likewise.
28060 (estimate_loops_at_level): Likewise.
28061 (estimate_loops): Likewise.
28062 * shrink-wrap.c (try_shrink_wrapping): Likewise.
28063 (spread_components): Likewise.
28064 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
28065 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
28066 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
28067 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
28068 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
28069 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
28070 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
28071 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
28072 (mark_threaded_blocks): Likewise.
28073 (thread_through_all_blocks): Likewise.
28074 * tree-ssa.c (verify_ssa): Likewise.
28075 (execute_update_addresses_taken): Likewise.
28076 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
28078 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28080 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
28082 (post_order_compute): Likewise.
28083 (inverted_post_order_compute): Likewise.
28084 (pre_and_rev_post_order_compute_fn): Likewise.
28086 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28088 * genrecog.c (int_set::int_set): Explicitly construct our
28089 auto_vec base class.
28090 * vec.h (auto_vec::auto_vec): New constructor.
28092 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28094 * bitmap.h (class auto_bitmap): New constructor taking
28095 bitmap_obstack * argument.
28097 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28099 * bitmap.h (class auto_bitmap): Change type of m_bits to
28100 bitmap_head, and adjust ctor / dtor and member operators.
28102 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
28104 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
28105 when returned register mode doesn't match original mode.
28107 2017-05-12 Jeff Law <law@redhat.com>
28108 Jakub Jelinek <jakub@redhat.com>
28110 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
28111 we look for cc setter after the compare-elim changes.
28112 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
28113 within the vector to match what compare-elim now expects.
28114 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
28115 (xorsi3_flags, one_cmplsi2_flags): Likewise.
28117 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
28118 after the compare-elim changes.
28119 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
28120 the vector to match what compare-elim now expects.
28121 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
28122 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
28123 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
28124 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
28125 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
28127 * config/visium/visium.c (single_set_and_flags): Fix where
28128 we look for cc setter after the compare-elim changes.
28129 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
28130 with the vector to match what compare-elim now expects.
28131 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
28132 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
28133 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
28134 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
28135 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
28136 (neg<mode>2_insn_set_overflow): Likewise.
28138 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
28140 PR middle-end/79794
28141 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
28142 maybe_expand_insn call, set ops[0].target. If still set after call,
28143 set alt_rtl. Add extra arg to recursive calls.
28144 (extract_bit_field): Add alt_rtl argument. Pass to
28146 * expmed.h (extract_bit_field): Fix prototype.
28147 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
28148 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
28149 to extract_bit_field_calls.
28150 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
28151 Pass alt_rtl to extract_bit_field calls.
28152 * calls.c (store_unaligned_arguments_into_psuedos)
28153 load_register_parameters): Pass extra NULL to extract_bit_field calls.
28154 * optabs.c (maybe_legitimize_operand): Clear op->target when call
28156 * optabs.h (struct expand_operand): Add target bitfield.
28158 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28160 * compare-elim.c (try_eliminate_compare): Canonicalize
28161 operation with embedded compare to
28162 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
28163 (set (reg) (operation)].
28165 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
28167 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28170 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
28171 cost of adding a carry flag for ADC instruction.
28172 [case MINUS]: Ignore the cost of subtracting a carry flag
28173 for SBB instruction.
28175 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
28177 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
28179 * config/rs6000/bmiintrin.h: New file.
28180 * config/rs6000/bmi2intrin.h: New file.
28181 * config/rs6000/x86intrin.h: New file.
28183 2017-05-12 Jeff Law <law@redhat.com>
28185 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
28188 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
28190 PR middle-end/80707
28191 * tree-cfg.c: Remove cfg edges of unreachable case statements.
28193 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28195 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28196 early expansion of vector divide builtins.
28197 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
28198 builtins identified as having unsigned arguments.
28200 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28202 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
28203 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
28204 expansion of vector logical operations (and, andc, or, xor,
28207 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28209 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
28210 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
28212 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28214 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28215 early GIMPLE expansion of vector multiplies.
28217 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28219 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
28220 TARGET_HAVE_MOVT conditional.
28221 (movt splitter): Likewise.
28223 2017-05-12 Richard Biener <rguenther@suse.de>
28225 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
28226 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28227 Fold all stmts not inplace.
28229 2017-05-12 Richard Biener <rguenther@suse.de>
28231 PR tree-optimization/80713
28232 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
28233 inserted_exprs bit for not removed stmts.
28235 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
28237 PR middle-end/69921
28238 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
28239 parallelized" attribute for parallelized OpenACC kernels.
28240 * omp-offload.c (execute_oacc_device_lower): Use it.
28242 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
28243 Set "oacc kernels" attribute.
28244 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
28245 parameter. Adjust all users.
28246 (oacc_fn_attrib_kernels_p): Remove function.
28247 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
28248 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
28249 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
28250 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
28251 assert "oacc kernels" attribute is set.
28253 2017-05-11 Carl Love <cel@us.ibm.com>
28255 * config/rs6000/rs6000-c: Add support for built-in functions
28256 vector unsigned char vec_popcnt (vector signed char)
28257 vector unsigned char vec_popcnt (vector unsigned char)
28258 vector unsigned short vec_popcnt (vector signed short)
28259 vector unsigned short vec_popcnt (vector unsigned short)
28260 vector unsigned int vec_popcnt (vector signed int)
28261 vector unsigned int vec_popcnt (vector unsigned int)
28262 vector unsigned long long vec_popcnt (vector signed long long)
28263 vector unsigned long long vec_popcnt (vector unsigned long long)
28264 vector signed long long vec_slo (vector signed long long,
28265 vector signed char)
28266 vector signed long long vec_slo (vector signed long long,
28267 vector unsigned char)
28268 vector unsigned long long vec_slo (vector unsigned long long,
28269 vector signed char)
28270 vector unsigned long long vec_slo (vector unsigned long long,
28271 vector unsigned char)
28272 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
28273 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
28274 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
28275 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
28276 * doc/extend.texi: Update the built-in documentation file for the
28277 new built-in functions.
28279 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
28281 * attribs.h (sorted_attr_string): Move machine independent
28282 functions for target clone support from the i386 port to common
28283 code. Rename ix86_function_versions to common_function_versions.
28284 Rename make_name to make_unique_name.
28285 (common_function_versions): Likewise.
28286 (make_unique_name): Likewise.
28287 (make_dispatcher_decl): Likewise.
28288 (is_function_default_version): Likewise.
28289 * attribs.c (attr_strcmp): Likewise.
28290 (sorted_attr_string): Likewise.
28291 (common_function_versions): Likewise.
28292 (make_unique_name): Likewise.
28293 (make_dispatcher_decl): Likewise.
28294 (is_function_default_version): Likewise.
28295 * config/i386/i386.c (attr_strcmp): Likewise.
28296 (sorted_attr_string): Likewise.
28297 (ix86_function_versions): Likewise.
28298 (make_name): Likewise.
28299 (make_dispatcher_decl): Likewise.
28300 (is_function_default_version): Likewise.
28301 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
28303 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28306 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
28307 Account for direct move costs for vec_construct of integer
28310 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
28313 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
28314 (UNSPEC_STX_ATOMIC): Ditto.
28315 (loaddi_via_sse): New insn.
28316 (storedi_via_sse): Ditto.
28317 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
28318 Update corresponding peephole2 patterns.
28319 (atomic_storedi_fpu): Ditto.
28321 2017-05-11 Julia Koval <julia.koval@intel.com>
28323 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
28324 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
28326 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
28327 (__builtin_ia32_rsqrt14ss_mask): New builtins.
28328 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
28330 2017-05-11 Nathan Sidwell <nathan@acm.org>
28332 * graphite-poly.c: Include dumpfile.h.
28334 * dumpfle.h (dump_function): Declare here ...
28335 * tree-dump.h (dump_function): ... not here.
28336 * dumpfile.c: #include tree-cfg.h.
28337 (dump_function): Move here from ...
28338 * tree-dump.c (dump_function): ... here.
28339 * gimplify.c: #include splay-tree.h, not tree-dump.h.
28340 * graphite-poly.c: Don't include tree-dump.h.
28341 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
28342 * print-tree.c: Likewise.
28343 * stor-layout.c: Likewise.
28344 * tree-nested.c: Likewise.
28346 * dumpfile.c (dump_start): Use TDF_FLAGS.
28347 (dump_enable_all): Fix TDF_KIND check thinko.
28349 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
28351 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
28352 array entries to represent two legal parameterizations of the
28353 overloaded __builtin_cmpb function, as represented by the
28354 P6_OV_BUILTIN_CMPB constant.
28355 (altivec_resolve_overloaded_builtin): Add special case handling
28356 for the __builtin_cmpb function, as represented by the
28357 P6_OV_BUILTIN_CMPB constant.
28358 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
28359 (BU_P6_64BIT_2): New macro.
28360 (BU_P6_OVERLOAD_2): New macro
28361 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
28362 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
28363 (CMPB): Add overload support to represent both 32-bit and 64-bit
28364 compare-bytes function.
28365 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
28366 support for TARGET_CMPB.
28367 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
28368 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28369 documentation of the __builtin_cmpb overloaded built-in function.
28371 2017-05-11 Richard Biener <rguenther@suse.de>
28373 PR tree-optimization/80705
28374 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
28375 bases are not vectorizable.
28377 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28379 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
28380 when counting register pressure.
28382 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28384 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
28385 (iv_ca_more_deps): Renamed to ...
28386 (iv_ca_compare_deps): ... this.
28387 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
28389 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28391 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
28393 (determine_group_iv_costs): ... here.
28394 (find_inv_vars_cb): Record inv var if it's not recorded before.
28396 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28398 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
28399 (get_shiftadd_cost): Ditto.
28401 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28403 * tree-ssa-address.c: Include header file.
28404 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
28406 (add_to_parts): Refactor.
28407 (addr_to_parts): New parameter. Update use of move_hint_to_base.
28408 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
28411 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28413 PR tree-optimization/53090
28414 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
28416 (extract_cond_operands): Detect condition with IV on both sides
28417 and return COMP_IV_EXPR_2.
28418 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
28419 (rewrite_use_compare): Simplify by removing call to function
28420 extract_cond_operands.
28422 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28424 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
28425 (extract_cond_operands): Detect condition comparing against non-
28426 invariant bound and return appropriate enum value.
28427 (find_interesting_uses_cond): Update use of extract_cond_operands.
28428 Handle its return value accordingly.
28429 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
28431 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28433 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
28434 nonlinear iv_use computation in loop invariant sensitive way.
28436 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28438 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
28439 (find_iv_candidates): Call relate_compare_use_with_all_cands.
28441 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28443 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
28444 (dump_cand): Support iv_cand.inv_exprs.
28445 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
28447 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
28450 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28452 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
28454 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
28455 as local function. Include necessary header files.
28456 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
28458 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28460 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
28462 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28464 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
28465 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
28466 RSHIFT_EXPR and BIT_NOT_EXPR.
28468 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28470 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
28471 (adjust_setup_cost): New parameter supporting round up adjustment.
28472 (struct address_cost_data): Delete.
28473 (force_expr_to_var_cost): Don't bound cost with spill_cost.
28474 (split_address_cost, ptr_difference_cost): Delete.
28475 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
28476 (struct ainc_cost_data): New struct.
28477 (get_address_cost_ainc): New function.
28478 (get_address_cost, get_computation_cost): Reimplement.
28479 (determine_group_iv_cost_address): Record inv_expr for all uses of
28481 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
28482 (iv_ca_has_deps): Reimplemented to ...
28483 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
28485 (iv_ca_extend): Call iv_ca_more_deps.
28487 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28489 * tree-ssa-address.c (struct mem_address): Move to header file.
28490 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
28491 * tree-ssa-address.h (struct mem_address): Move from C file.
28492 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
28494 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28496 * tree-affine.h (aff_combination_type): New interface.
28497 (aff_combination_zero_p): Remove static.
28498 (aff_combination_const_p): New interface.
28499 (aff_combination_singleton_var_p): New interfaces.
28501 2017-05-11 Richard Biener <rguenther@suse.de>
28503 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28504 Skip unreachable blocks and destinations.
28505 (eliminate): Move stmt removal and fixup ...
28506 (fini_eliminate): ... here. Skip inserted exprs.
28507 (pass_pre::execute): Move fini_pre after fini_eliminate.
28508 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
28509 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
28510 PRE to get rid of dead code that has invalid SSA form and
28511 split critical edges again.
28513 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28515 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28517 2017-05-11 Richard Biener <rguenther@suse.de>
28519 * passes.c (execute_function_todo): Verify loops if they are
28520 said to be up-to-date.
28521 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
28522 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
28524 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
28527 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
28528 handle calling assemble_external ourself.
28531 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
28532 modes with zero size. Enhance comment.
28534 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28536 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
28537 built-ins for vec_xl and vec_xst with short and char pointer
28540 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
28542 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
28543 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
28544 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
28545 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
28546 (_mm_maskz_min_round_ss): New intrinsics.
28547 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
28548 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
28549 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
28550 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
28551 (__builtin_ia32_minss_mask_round): New builtins.
28552 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28553 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28554 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
28556 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
28557 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
28559 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
28562 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
28564 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
28565 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
28566 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
28567 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
28568 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
28569 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
28570 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
28571 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28572 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
28573 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
28574 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
28575 (__builtin_ia32_mulss_mask_round): New builtins.
28576 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28577 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28578 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
28580 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
28581 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
28583 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
28586 2017-05-10 Julia Koval <julia.koval@intel.com>
28588 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
28589 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
28590 (_mm256_setr_m128i): New intrinsics.
28592 2017-05-10 Julia Koval <julia.koval@intel.com>
28594 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
28595 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
28596 (_mm_maskz_rcp14_ss): New intrinsics.
28597 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
28598 (__builtin_ia32_rcp14ss_mask): New builtins.
28599 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
28601 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
28603 PR tree-optimization/51513
28604 * tree-cfg.c (gimple_seq_unreachable_p): New function.
28605 (assert_unreachable_fallthru_edge_p): Use it.
28606 (group_case_labels_stmt): Likewise.
28607 * tree-cfg.h: Prototype it.
28608 * stmt.c: Include cfghooks.h and tree-cfg.h.
28609 (emit_case_dispatch_table) <gap_label>: New local variable.
28610 Use it to fill dispatch table gaps.
28611 Test for default_label before updating probabilities.
28612 (expand_case) <default_label>: Remove unneeded initialization.
28613 Test for unreachable default case statement and remove its edge.
28614 Set default_label accordingly.
28615 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
28617 2017-05-10 Carl Love <cel@us.ibm.com>
28619 * config/rs6000/rs6000-c: Add support for built-in functions
28620 vector signed char vec_neg (vector signed char)
28621 vector signed short int vec_neg (vector short int)
28622 vector signed int vec_neg (vector signed int)
28623 vector signed long long vec_neg (vector signed long long)
28624 vector float vec_neg (vector float)
28625 vector double vec_neg (vector double)
28626 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
28628 * config/rs6000/altivec.h: Add define for vec_neg
28629 * doc/extend.texi: Update the built-in documentation for the
28630 new built-in functions.
28632 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28634 PR tree-optimization/77644
28635 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
28637 2017-05-10 Nathan Sidwell <nathan@acm.org>
28639 * dumpfile.h (TDI_lang_all): New.
28640 (TDF_KIND): New. Renumber others
28641 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
28643 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
28645 (get_dump_file_name): Adjust suffix generation.
28646 (dump_enable_all): Use TDF_KIND.
28647 * doc/invoke.texi (-fdump-lang-all): Document.
28649 * dumpfile.h: Tabify.
28651 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
28654 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
28655 Move member access before delete.
28657 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
28659 * tree-inline.c (expand_call_inline): Split block at stmt
28662 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
28665 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
28666 are now unused after splitting mov{sf,sd}_hardfloat.
28667 (f32_lr2): Likewise.
28668 (f32_lm): Likewise.
28669 (f32_lm2): Likewise.
28670 (f32_li): Likewise.
28671 (f32_li2): Likewise.
28672 (f32_lv): Likewise.
28673 (f32_sr): Likewise.
28674 (f32_sr2): Likewise.
28675 (f32_sm): Likewise.
28676 (f32_sm2): Likewise.
28677 (f32_si): Likewise.
28678 (f32_si2): Likewise.
28679 (f32_sv): Likewise.
28680 (f32_dm): Likewise.
28681 (f32_vsx): Likewise.
28682 (f32_av): Likewise.
28683 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
28684 For movsf, order stores so the VSX stores occur before the GPR
28685 store which encourages the register allocator to use a traditional
28686 FPR instead of a GPR. For movsd, order the stores so that the GPR
28687 store comes before the VSX stores to allow the power6 to work.
28688 This is due to the power6 not having a 32-bit integer store
28689 instruction from a FPR.
28690 (movsf_hardfloat): Likewise.
28691 (movsd_hardfloat): Likewise.
28693 2017-05-09 Martin Sebor <msebor@redhat.com>
28695 PR translation/80280
28696 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
28699 PR translation/80280
28700 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
28701 data member added in r247778.
28702 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
28704 2017-05-09 Nathan Sidwell <nathan@acm.org>
28706 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
28708 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
28711 2017-05-09 Marek Polacek <polacek@redhat.com>
28713 * doc/invoke.texi: Fix typo.
28715 2017-05-09 Richard Biener <rguenther@suse.de>
28717 * tree-vrp.c (vrp_val_is_max): Adjust comment.
28718 (vrp_val_is_min): Likewise.
28719 (set_value_range_to_value): Likewise.
28720 (set_value_range_to_nonnegative): Likewise.
28721 (gimple_assign_nonzero_p): Likewise.
28722 (gimple_stmt_nonzero_p): Likewise.
28723 (vrp_int_const_binop): Likewise. Remove unreachable case.
28724 (adjust_range_with_scev): Adjust comments.
28725 (compare_range_with_value): Likewise.
28726 (extract_range_from_phi_node): Likewise.
28727 (test_for_singularity): Likewise.
28729 2017-05-09 Richard Biener <rguenther@suse.de>
28731 * tree-vrp.c (get_single_symbol): Add assert that we don't
28732 get overflowed constants as invariant part.
28733 (compare_values_warnv): Add comment before the TREE_NO_WARNING
28734 checks. Use wi::cmp instead of recursing for integer constants.
28735 (compare_values): Just ignore whether we assumed undefined
28736 overflow instead of failing the compare.
28737 (extract_range_for_var_from_comparison_expr): Add comment before the
28738 TREE_NO_WARNING sets.
28739 (test_for_singularity): Likewise.
28740 (extract_range_from_comparison): Do not disable optimization
28741 when we assumed undefined overflow.
28742 (extract_range_basic): Remove init of unused var.
28744 2017-05-09 Richard Biener <rguenther@suse.de>
28746 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
28747 (extract_range_from_multiplicative_op_1): Adjust.
28748 (extract_range_from_binary_expr_1): Use int_const_binop.
28750 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
28753 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
28754 rs6000_store_data_bypass_p in seven define_bypass directives and
28755 in several comments.
28756 * config/rs6000/rs6000-protos.h: Add prototype for
28757 rs6000_store_data_bypass_p function.
28758 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
28759 function implements slightly different (rs6000-specific) semantics
28760 than store_data_bypass_p, returning false rather than aborting
28761 with assertion error when arguments do not satisfy the
28762 requirements of store data bypass.
28763 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
28764 rs6000_store_data_bypass_p.
28766 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
28768 * config/xtensa/xtensa-protos.h
28769 (xtensa_initial_elimination_offset): New declaration.
28770 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28771 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
28772 macro definition, add case for FRAME_POINTER_REGNUM when
28773 FRAME_GROWS_DOWNWARD.
28774 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
28775 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
28776 xtensa_initial_elimination_offset.
28778 2017-05-08 Nathan Sidwell <nathan@acm.org>
28780 * doc/invoke.texi: Alphabetize -fdump options.
28782 2017-05-08 Martin Sebor <msebor@redhat.com>
28784 PR translation/80280
28785 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
28787 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
28789 * target.def (compute_frame_layout): New optional target hook.
28790 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
28791 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
28792 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
28794 * reload1.c (verify_initial_elim_offsets): Likewise.
28795 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
28796 (use_simple_return_p): Call arm_compute_frame_layout if needed.
28797 (arm_get_frame_offsets): Split up into this ...
28798 (arm_compute_frame_layout): ... and this function.
28800 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
28802 * config/aarch64/constraints.md (Usa): New constraint.
28803 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
28805 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
28807 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
28808 with_multilib_list after it has been checked.
28810 2017-05-08 Richard Biener <rguenther@suse.de>
28812 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
28813 (bitmap_set_subtract_values): Likewise.
28815 2017-05-08 Richard Biener <rguenther@suse.de>
28817 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
28818 (gimple_assign_nonzero): ... this and remove strict_overflow_p
28820 (gimple_stmt_nonzero_warnv_p): Rename to ...
28821 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
28823 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
28824 (extract_range_basic): Adjust, do not disable propagation on
28825 strict overflow sensitive simplification.
28826 (vrp_visit_cond_stmt): Likewise.
28828 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
28830 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
28831 body size unconditionally.
28833 2017-05-07 Jeff Law <law@redhat.com>
28836 2017-05-06 Jeff Law <law@redhat.com>
28837 PR tree-optimization/78496
28838 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28841 PR tree-optimization/78496
28842 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28843 (simplify_stmt_using_ranges): Call it.
28844 (vrp_dom_walker::before_dom_children): Extract equivalences
28845 from an ASSERT_EXPR with an equality comparison against a
28848 2017-05-06 Jeff Law <law@redhat.com>
28850 PR tree-optimization/78496
28851 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28854 PR tree-optimization/78496
28855 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28856 (simplify_stmt_using_ranges): Call it.
28857 (vrp_dom_walker::before_dom_children): Extract equivalences
28858 from an ASSERT_EXPR with an equality comparison against a
28861 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
28863 * lra-constraints.c (lra_copy_reg_equiv): New function.
28864 (split_reg): Use it to copy equivalence information from the
28865 original register to the spill register.
28867 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
28869 PR rtl-optimization/75964
28870 * simplify-rtx.c (simplify_const_relational_operation): Remove
28871 invalid handling of comparisons of integer ABS.
28873 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
28875 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
28876 initialize to zero.
28877 (init_regs): Remove declaration.
28878 (function_arg_advance_32): Initialize error_p as boolean variable.
28880 2017-05-05 Nathan Sidwell <nathan@acm.org>
28882 * store-motion.c (remove_reachable_equiv_notes): Reformat long
28883 lines. Use for (;;).
28885 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28887 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
28888 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
28889 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
28890 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
28891 VF=2 that require versioning.
28893 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28895 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
28898 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28900 * diagnostic.h (diagnostic_override_option_index): Convert from
28901 macro to inline function.
28903 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28905 * diagnostic.c (last_module_changed_p): New function.
28906 (set_last_module): New function.
28907 (diagnostic_report_current_module): Convert macro usage to
28908 the above functions.
28909 * diagnostic.h (diagnostic_context::last_module): Strengthen
28910 from const line_map * to const line_map_ordinary *.
28911 (diagnostic_last_module_changed): Delete macro.
28912 (diagnostic_set_last_module): Delete macro.
28914 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28916 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
28917 with diagnostic_report_diagnostic.
28918 (diagnostic_n_impl_richloc): Likewise.
28919 * diagnostic.h (report_diagnostic): Delete macro.
28920 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
28921 with diagnostic_report_diagnostic.
28922 * substring-locations.c (format_warning_va): Likewise.
28924 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28926 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
28927 save/restor of format_spec. Move option-printing code to...
28928 (print_option_information): ...this new function, and
28929 reimplement by simply printing to the pretty_printer,
28930 rather than appending to the format string.
28932 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28934 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
28935 handling logic into...
28936 (update_effective_level_from_pragmas): ...this new function.
28938 2017-05-04 Andrew Waterman <andrew@sifive.com>
28940 * config/riscv/riscv.opt (mstrict-align): New option.
28941 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
28942 (SLOW_UNALIGNED_ACCESS): Define.
28943 (riscv_slow_unaligned_access): Declare.
28944 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
28946 (riscv_slow_unaligned_access): New variable.
28947 (rocket_tune_info): Set slow_unaligned_access to true.
28948 (optimize_size_tune_info): Set slow_unaligned_access to false.
28949 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
28950 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
28951 (riscv_option_override): Set riscv_slow_unaligned_access.
28952 * doc/invoke.texi: Add -mstrict-align to RISC-V.
28954 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
28956 * config/riscv/riscv.md: Unify indentation.
28958 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28963 * config/rs6000/rs6000.md (u code attribute): Add FIX and
28965 (extendsi<mode>2): Add support for doing sign extension via
28966 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
28967 don't have ISA 3.0 instructions.
28968 (extendsi<mode>2 splitter): Likewise.
28969 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
28970 generate the normal insns since SImode can now go in vector
28971 registers. Disallow the special UNSPECs needed for previous
28972 machines to hide SImode being used. Add new insns
28973 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
28974 (fix_trunc<mode>si2_stfiwx): Likewise.
28975 (fix_trunc<mode>si2_internal): Likewise.
28976 (fixuns_trunc<mode>si2): Likewise.
28977 (fixuns_trunc<mode>si2_stfiwx): Likewise.
28978 (fctiw<u>z_<mode>_smallint): Likewise.
28979 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
28980 of floating point to 32-bit integer from doing a direct move to
28981 the GPR registers to do a store.
28982 (fctiwz_<mode>): Break long line.
28984 2017-05-05 Bin Cheng <bin.cheng@arm.com>
28986 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
28987 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
28988 (addr_list, addr_offset_valid_p): New.
28989 (split_address_groups): Check offset validity with above function.
28990 (gt-tree-ssa-loop-ivopts.h): Include header file.
28992 2017-05-05 Nathan Sidwell <nathan@acm.org>
28994 * config.gcc (arm*-*-*): Add missing 'fi'.
28996 2017-05-05 Steve Ellcey <sellcey@cavium.com>
28998 * doc/invoke.texi (-fopt-info): Explicitly say order of options
28999 included in -fopt-info does not matter.
29000 * doc/optinfo.texi (-fopt-info): Fix description of default
29001 behavour. Explicitly say order of options included in -fopt-info
29004 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29006 * config.gcc: Allow combinations of aprofile and rmprofile values for
29007 --with-multilib-list.
29008 * config/arm/t-multilib: New file.
29009 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
29010 variables. Remove setting of ISA and floating-point ABI in
29011 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
29012 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
29013 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
29014 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
29016 * config/arm/t-rmprofile: Likewise except for the matches changes.
29017 * doc/install.texi (--with-multilib-list): Document the combination of
29018 aprofile and rmprofile values and warn about pitfalls in doing that.
29020 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29022 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
29023 (movdi_aarch64): Likewise.
29025 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29027 PR tree-optimization/80632
29028 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
29030 (build_arrays): Initialize it for virtual phis.
29031 (fix_phi_nodes): Use it for virtual phis.
29033 PR tree-optimization/80558
29034 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
29035 [x, y] op z into [x op, y op z] for op & or | if conditions
29038 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
29039 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29042 * config/arm/arm.md (use_literal_pool): Remove.
29043 (64-bit immediate split): No longer takes cost into consideration
29044 if arm_disable_literal_pool is enabled.
29045 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
29046 used when arm_disable_literal_pool is enabled.
29047 (arm_max_const_double_inline_cost): Remove use of
29048 arm_disable_literal_pool.
29049 (push_minipool_fix): Add assert.
29050 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
29051 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
29052 (no_literal_pool_sf_immediate): New.
29054 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29056 PR tree-optimization/80613
29057 * tree-ssa-dce.c (propagate_necessity): Remove cases for
29058 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
29060 2017-05-05 Richard Biener <rguenther@suse.de>
29062 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
29064 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
29066 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
29067 of this flag from insn conditions due to removal from r247495.
29069 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29071 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
29073 (arm_early_store_addr_dep_ptr): Likewise.
29074 * config/arm/aarch-common-protos.h
29075 (arm_early_load_addr_dep_ptr): Add prototype.
29076 (arm_early_store_addr_dep_ptr): Likewise.
29077 * config/arm/cortex-a53.md: Add new bypasses.
29079 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29081 * tree.c (next_type_uid): Change type to unsigned.
29082 (type_hash_canon): Decrement back next_type_uid if
29083 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
29084 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
29087 2017-05-04 Martin Sebor <msebor@redhat.com>
29089 * builtins.c: Fix a trivial typo in a comment.
29091 PR middle-end/79234
29092 * builtins.c (check_sizes): Adjust to handle reading past the end.
29093 Avoid printing excessive upper bound of ranges. Use %E to print
29094 tree nodes instead of converting them to %wu.
29095 (expand_builtin_memchr): New function.
29096 (compute_dest_size): Rename...
29097 (compute_objsize): ...to this.
29098 (expand_builtin_memcpy): Adjust.
29099 (expand_builtin_mempcpy): Adjust.
29100 (expand_builtin_strcat): Adjust.
29101 (expand_builtin_strcpy): Adjust.
29102 (check_strncat_sizes): Adjust.
29103 (expand_builtin_strncat): Adjust.
29104 (expand_builtin_strncpy): Adjust and simplify.
29105 (expand_builtin_memset): Adjust.
29106 (expand_builtin_bzero): Adjust.
29107 (expand_builtin_memcmp): Adjust.
29108 (expand_builtin): Handle memcmp.
29109 (maybe_emit_chk_warning): Check strncat just once.
29111 2017-05-04 Martin Sebor <msebor@redhat.com>
29113 PR preprocessor/79214
29114 PR middle-end/79222
29115 PR middle-end/79223
29116 * builtins.c (check_sizes): Add inlining context and issue
29117 warnings even when -Wno-system-headers is set.
29118 (check_strncat_sizes): Same.
29119 (expand_builtin_strncat): Same.
29120 (expand_builtin_memmove): New function.
29121 (expand_builtin_stpncpy): Same.
29122 (expand_builtin): Handle memmove and stpncpy.
29124 2017-05-04 Bin Cheng <bin.cheng@arm.com>
29126 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
29127 which is not used any more.
29129 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29131 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
29133 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29135 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
29136 (cortexa53_tunings): Likewise.
29137 (cortexa57_tunings): Likewise.
29138 (cortexa72_tunings): Likewise.
29139 (cortexa73_tunings): Likewise.
29141 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29143 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
29144 Set loop alignment to 8.
29146 2017-05-04 Martin Sebor <msebor@redhat.com>
29148 PR translation/80280
29149 * builtins.c (expand_builtin_object_size): Add missing quoting to
29150 %D and like directives.
29151 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
29152 (hsa_type_for_tree_type): Same.
29153 (verify_function_arguments): Same.
29154 * symtab.c (symbol_table::change_decl_assembler_name): Same.
29155 * varasm.c (get_section): Same.
29158 2017-05-04 Martin Sebor <msebor@redhat.com>
29160 PR translation/80280
29161 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
29163 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29165 * config/aarch64/aarch64.c (generic_addrcost_table):
29166 Change HI/TI mode setting.
29168 2017-05-04 Martin Jambor <mjambor@suse.cz>
29170 PR tree-optimization/80622
29171 * tree-sra.c (comes_initialized_p): New function.
29172 (build_accesses_from_assign): Only set write lazily when
29173 comes_initialized_p is false.
29174 (analyze_access_subtree): Use comes_initialized_p.
29175 (propagate_subaccesses_across_link): Assert !comes_initialized_p
29176 instead of testing for PARM_DECL.
29178 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29180 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
29181 constraint on operand 0 to allow more general addressing modes.
29182 Adjust output template.
29183 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
29185 * config/aarch64/aarch64-protos.h
29186 (aarch64_address_valid_for_prefetch_p): Declare prototype.
29187 * config/aarch64/constraints.md (Dp): New address constraint.
29188 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
29191 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
29193 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
29194 update use of estimate_ipcp_clone_size_and_time.
29195 (estimate_local_effects): Update use of
29196 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
29197 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
29198 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
29199 Return nonspecialized time.
29201 2017-05-04 Richard Biener <rguenther@suse.de>
29203 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
29204 for the last VUSE which def dominates the PHI. Directly call
29206 (get_continuation_for_phi_1): Remove.
29208 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
29210 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
29211 to explain the use of truncating division. Cap the number of
29212 iterations to the maximum given by nb_iterations_upper_bound,
29215 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
29217 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
29218 * configure: Regenerate.
29219 * config.in: Regenerate.
29220 * config/i386/driver-mingw32.c: new file.
29221 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
29222 * config.host: Link driver-mingw32.o on MinGW host.
29223 * doc/install.texi: Document new --enable-mingw-wildcard configure
29226 2017-05-04 Marek Polacek <polacek@redhat.com>
29228 PR tree-optimization/80612
29229 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
29231 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29232 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
29234 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
29235 (movt splitter): Likewise.
29236 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
29237 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
29238 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
29239 block for Thumb-1 with MOVT.
29240 (thumb2_legitimate_address_p): Move code block ...
29241 (can_avoid_literal_pool_for_label_p): ... into this new function.
29242 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
29244 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
29245 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
29246 "M-profile targets with the MOVT instruction".
29248 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29250 * config/arm/arm-builtins.c (arm_init_builtins): Rename
29251 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
29252 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
29254 2017-05-04 Martin Liska <mliska@suse.cz>
29256 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
29257 variable cond_code.
29259 2017-05-04 Richard Biener <rguenther@suse.de>
29261 * tree.c (array_at_struct_end_p): Handle arrays at struct
29262 end with flexarrays more conservatively. Refactor and treat
29263 arrays of arrays or aggregates more strict. Fix
29264 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
29265 * tree.c (array_at_struct_end_p): Adjust prototype.
29266 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
29267 * gimple-fold.c (get_range_strlen): Likewise.
29268 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
29270 2017-05-04 Richard Biener <rguenther@suse.de>
29272 PR tree-optimization/31130
29273 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
29275 (supports_overflow_infinity): Likewise.
29276 (is_negative_overflow_infinity): Likewise.
29277 (is_positive_overflow_infinity): Likewise.
29278 (is_overflow_infinity): Likewise.
29279 (stmt_overflow_infinity): Likewise.
29280 (overflow_infinity_range_p): Likewise.
29281 (usable_range_p): Remove as always returning true.
29282 (make_overflow_infinity): Remove.
29283 (negative_overflow_infinity): Likewise.
29284 (positive_overflow_infinity): Likewise.
29285 (avoid_overflow_infinity): Likewise.
29286 (set_value_range): Adjust accordingly.
29287 (set_value_range_to_nonnegative): Likewise, remove now unused
29288 overflow_infinity arg.
29289 (vrp_operand_equal_p): Adjust.
29290 (update_value_range): Likewise.
29291 (range_int_cst_singleton_p): Likewise.
29292 (operand_less_p): Likewise.
29293 (compare_values_warnv): Likewise.
29294 (extract_range_for_var_from_comparison_expr): Likewise.
29295 (vrp_int_const_binop): Likewise.
29296 (zero_nonzero_bits_from_vr): Likewise.
29297 (extract_range_from_multiplicative_op_1): Likewise.
29298 (extract_range_from_binary_expr_1): Likewise.
29299 (extract_range_from_unary_expr): Likewise.
29300 (extract_range_from_comparison): Likewise.
29301 (extract_range_basic): Likewise.
29302 (adjust_range_with_scev): Likewise.
29303 (compare_ranges): Likewise.
29304 (compare_range_with_value): Likewise.
29305 (dump_value_range): Likewise.
29306 (test_for_singularity): Likewise, remove strict_overflow_p parameter
29308 (simplify_cond_using_ranges): Adjust.
29310 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
29312 * brig-builtins.def: Added a builtin for class_f64.
29313 * builtin-types.def: Added a builtin type needed by class_f64.
29315 2017-05-03 Jason Merrill <jason@redhat.com>
29317 * timevar.def: Add TV_CONSTEXPR.
29319 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29321 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
29323 2017-05-03 Martin Jambor <mjambor@suse.cz>
29325 * ipa-prop.c (ipa_update_after_lto_read): Removed.
29326 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
29327 * ipa-cp.c (ipcp_propagate_stage): Do not call
29328 ipa_update_after_lto_read.
29329 * ipa-inline.c (ipa_inline): Likewise.
29331 2017-05-03 Martin Jambor <mjambor@suse.cz>
29333 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
29334 tag. Added a default constructor and a destructor.
29335 (ipa_edge_args_sum_t): New class;
29336 (ipa_edge_args_sum): Declare.
29337 (ipa_edge_args_vector): Remove declaration.
29338 (IPA_EDGE_REF): Use ipa_edge_args_sum.
29339 (ipa_free_edge_args_substructures): Remove declaration.
29340 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
29341 (ipa_edge_args_info_available_for_edge_p): Likewise.
29342 * ipa-prop.c (ipa_edge_args_vector): Removed.
29343 (edge_removal_hook_holder): Likewise.
29344 (edge_duplication_hook_holder): Likewise.
29345 (ipa_edge_args_sum): New variable.
29346 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
29347 ipa_edge_args_vector.
29348 (ipa_free_edge_args_substructures): Likewise.
29349 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
29350 ipa_edge_args_vector.
29351 (ipa_edge_removal_hook): Turned into method
29352 ipa_edge_args_sum_t::remove.
29353 (ipa_edge_duplication_hook): Turned into method
29354 ipa_edge_args_sum_t::duplicate.
29355 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
29356 registering edge hooks.
29357 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
29358 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
29359 ipa_edge_args_sum instead of ipa_edge_args_vector.
29360 * ipa-profile.c (ipa_profile): Likewise.
29362 2017-05-03 Martin Jambor <mjambor@suse.cz>
29364 * symbol-summary.h (function_summary): New method exists.
29365 (function_summary::symtab_removal): Deallocate through release.
29366 (call_summary): New class.
29367 (gt_ggc_mx): New overload.
29368 (gt_pch_nx): Likewise.
29369 (gt_pch_nx): Likewise.
29371 2017-05-03 Jeff Law <law@redhat.com>
29373 PR tree-optimization/78496
29374 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
29375 from simplify_cond_using_ranges. Split off code to walk
29376 backwards through casts into ...
29377 (simplify_cond_using_ranges_2): New function.
29378 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
29379 (execute_vrp): After identifying jump threads, call
29380 simplify_cond_using_ranges_2.
29382 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
29385 * ipa-inline.h (inline_summary): Add ctor.
29386 (create_ggc): Do not use ggc_cleared_alloc.
29388 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
29389 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29391 * gcc.c (handle_braces): Support escaping in switch matching
29393 * doc/invoke.texi (Spec Files): Document it.
29394 Remove superfluous @code markup in items.
29396 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29398 * diagnostic-show-locus.c (struct column_range): New struct.
29399 (get_affected_columns): New function.
29400 (get_printed_columns): New function.
29401 (struct correction): New struct.
29402 (correction::ensure_capacity): New function.
29403 (correction::ensure_terminated): New function.
29404 (struct line_corrections): New struct.
29405 (line_corrections::~line_corrections): New dtor.
29406 (line_corrections::add_hint): New function.
29407 (layout::print_trailing_fixits): Reimplement in terms of the new
29409 (selftest::test_overlapped_fixit_printing): New function.
29410 (selftest::diagnostic_show_locus_c_tests): Call it.
29412 2017-05-03 Nathan Sidwell <nathan@acm.org>
29414 Canonicalize canonical type hashing
29415 * tree.h (type_hash_canon_hash): Declare.
29416 * tree.c (type_hash_list, attribute_hash_list): Move into
29417 type_hash_canon_hash.
29418 (build_type_attribute_qual_variant): Break out hash code calc into
29419 type_hash_canon_hash.
29420 (type_hash_canon_hash): New. Generic type hash computation.
29421 (build_range_type_1, build_array_type_1, build_function_type,
29422 build_method_type_directly, build_offset_type, build_complex_type,
29423 make_vector_type): Call it.
29425 2017-05-03 Richard Biener <rguenther@suse.de>
29427 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
29428 When all DRs have unknown misaligned do not always peel
29429 when there is a store but apply the same costing model as if
29430 there were only loads.
29432 2017-05-03 Richard Biener <rguenther@suse.de>
29435 PR tree-optimization/80492
29436 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
29437 compare_base_decls returning dont-know properly.
29439 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
29441 * config/arm/iterators.md (CCSI): New mode iterator.
29442 (arch): New mode attribute.
29443 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
29444 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
29445 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
29446 code iterator for success result mode.
29447 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
29448 the corresponding new insn generators.
29450 2017-05-03 Bin Cheng <bin.cheng@arm.com>
29453 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29454 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29456 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
29458 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
29459 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
29460 (DDR_A): Wrap DDR argument in brackets.
29461 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
29462 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
29463 (DDR_REVERSED_P): Likewise.
29465 2017-05-03 Jakub Jelinek <jakub@redhat.com>
29467 PR tree-optimization/79472
29468 * tree-switch-conversion.c (struct switch_conv_info): Add
29469 contiguous_range and default_case_nonstandard fields.
29470 (collect_switch_conv_info): Compute contiguous_range and
29471 default_case_nonstandard fields, don't clear final_bb if
29472 contiguous_range and only the default case doesn't have the required
29474 (check_all_empty_except_final): Set default_case_nonstandard instead
29475 of failing if contiguous_range and the default case doesn't have empty
29477 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
29478 and only the default case doesn't have the required constants. Skip
29480 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
29481 if default_case_nonstandard.
29482 (build_constructors): Build constant 1 just once. Assert that default
29483 values aren't inserted in between cases if contiguous_range. Skip
29485 (build_arrays): Skip virtual phis.
29486 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
29487 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
29488 Handle virtual phis.
29489 (gen_inbound_check): Handle default_case_nonstandard case.
29490 (process_switch): Adjust check_final_bb caller. Call
29491 gather_default_values with the first non-default case instead of
29492 default case if default_case_nonstandard.
29494 2017-05-02 Nathan Sidwell <nathan@acm.org>
29496 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
29497 check. Fix formatting.
29499 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
29501 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
29502 errors when comparing specialized and unspecialized times.
29504 2017-05-02 David Malcolm <dmalcolm@redhat.com>
29506 * diagnostic-show-locus.c
29507 (layout::should_print_annotation_line_p): Make private.
29508 (layout::print_annotation_line): Make private.
29509 (layout::annotation_line_showed_range_p): Make private.
29510 (layout::show_ruler): Make private.
29511 (layout::print_source_line): Make private. Pass in line and
29512 line_width, rather than calling location_get_source_line. Drop
29514 (layout::print_leading_fixits): New method.
29515 (layout::print_any_fixits): Rename to...
29516 (layout::print_trailing_fixits): ...this, and make private.
29517 Don't print newline fixits.
29518 (diagnostic_show_locus): Move logic for printing one row into...
29519 (layout::print_line): ...this new function. Move the
29520 location_get_source_line call and error-handling from
29521 print_source_line to here. Call print_leading_fixits, and rename
29522 print_any_fixits to print_trailing_fixits.
29523 (selftest::test_fixit_insert_containing_newline): Update now that
29524 newlines are partially supported.
29525 (selftest::test_fixit_insert_containing_newline_2): New test.
29526 (selftest::test_fixit_replace_containing_newline): Update comments.
29527 (selftest::diagnostic_show_locus_c_tests): Call the new test.
29528 * edit-context.c (class added_line): New class.
29529 (class edited_line): Describe newline handling in comment.
29530 (edited_line::actually_edited_p): New method.
29531 (edited_line::print_content): Delete redundant decl.
29532 (edited_line::m_predecessors): New field.
29533 (edited_file::print_content): Call edited_line::print_content.
29534 (edited_file::print_diff): Update to support newlines.
29535 (edited_file::print_diff_hunk): Likewise.
29536 (edited_file::print_run_of_changed_lines): New function.
29537 (edited_file::print_diff_line): Convert to...
29538 (print_diff_line): ...this.
29539 (edited_file::get_effective_line_count): New function.
29540 (edited_line::edited_line): Initialize new field m_predecessors.
29541 (edited_line::~edited_line): Clean up m_predecessors.
29542 (edited_line::apply_fixit): Handle newlines.
29543 (edited_line::get_effective_line_count): New function.
29544 (edited_line::print_content): New function.
29545 (edited_line::print_diff_lines): New function.
29546 (selftest::test_applying_fixits_insert_containing_newline): New
29548 (selftest::test_applying_fixits_replace_containing_newline): New
29550 (selftest::insert_line): New function.
29551 (selftest::test_applying_fixits_multiple_lines): Add example of
29553 (selftest::edit_context_c_tests): Call the new tests.
29555 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29557 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
29558 parameter cand. Update dump information.
29559 (get_computation_cost): Update uses.
29561 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29563 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
29564 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
29565 (get_computation_at, rewrite_use_address): Update use of
29566 get_computation_aff.
29568 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29570 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
29571 (get_computation): Delete.
29572 (get_computation_cost): Implement like get_computation_cost_at.
29573 Use get_computation_at.
29574 (get_computation_cost_at): Delete.
29575 (rewrite_use_nonlinear_expr): Use get_computation_at.
29576 (rewrite_use_compare, remove_unused_ivs): Ditto.
29578 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29580 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
29582 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29584 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
29585 (ivopts_global_cost_for_size): Rename parameter and update uses.
29586 (iv_ca_recount_cost): Update uses.
29587 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
29588 candidates seperately in n_invs and n_cands.
29589 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
29591 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29593 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
29594 (find_inv_vars_cb): New.
29595 (find_depends): Renamed to ...
29596 (find_inv_vars): ... this.
29597 (add_candidate_1, force_var_cost): Call find_inv_vars.
29598 (split_address_cost, determine_group_iv_cost_cond): Ditto.
29600 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29602 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
29603 inv_vars. Add inv_exprs.
29604 (struct iv_cand): Rename depends_on to inv_vars.
29605 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
29606 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
29607 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
29608 (dump_cand): Dump inv_vars.
29609 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
29610 (record_invariant, find_depends, add_candidate_1): Ditto.
29611 (set_group_iv_cost, force_var_cost): Ditto.
29612 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
29613 (get_computation_cost_at, get_computation_cost): Ditto.
29614 (determine_group_iv_cost_generic): Ditto.
29615 (determine_group_iv_cost_address): Ditto.
29616 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
29617 (determine_group_iv_costs): Ditto.
29618 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
29619 (iv_ca_set_remove_invariants): Renamed to ...
29620 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
29621 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
29622 (iv_ca_set_add_invariants): Renamed to ...
29623 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
29624 (iv_ca_set_cp): Use iv_ca_set_add_invs.
29625 (iv_ca_has_deps): Support inv_vars and inv_exprs.
29626 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
29627 (create_new_ivs): Remove useless dump.
29629 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29631 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
29633 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
29634 (iv_ca_set_no_cp, create_new_iv): Ditto.
29636 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29638 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29640 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29642 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
29643 function tree_check2.
29645 2017-05-02 Martin Liska <mliska@suse.cz>
29647 * doc/gcov.texi: Add missing preposition.
29648 * gcov.c (function_info::function_info): Properly fill up
29649 all member variables.
29651 2017-05-02 Tamar Christina <tamar.christina@arm.com>
29653 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
29655 2017-05-02 Tamar Christina <tamar.christina@arm.com>
29657 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
29659 2017-05-02 Martin Liska <mliska@suse.cz>
29662 * lto-streamer-in.c (lto_read_tree_1): Remove
29663 LTO_STREAMER_DEBUG.
29664 * lto-streamer.c (struct tree_hash_entry): Likewise.
29665 (struct tree_entry_hasher): Likewise.
29666 (tree_entry_hasher::hash): Likewise.
29667 (tree_entry_hasher::equal): Likewise.
29668 (lto_streamer_init): Likewise.
29669 (lto_orig_address_map): Likewise.
29670 (lto_orig_address_get): Likewise.
29671 (lto_orig_address_remove): Likewise.
29672 * lto-streamer.h: Likewise.
29673 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
29674 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
29676 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
29678 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
29679 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
29680 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
29681 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
29682 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
29683 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
29684 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
29685 (mm_maskz_sub_ss): New intrinsics.
29686 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29687 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
29688 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
29689 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
29690 (__builtin_ia32_subss_mask_round): New builtins.
29691 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29692 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29693 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
29695 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
29696 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
29698 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
29701 2017-05-02 Martin Jambor <mjambor@suse.cz>
29703 PR tree-optimization/78687
29704 * tree-sra.c (access): New field parent.
29705 (process_subtree_disqualification): New function.
29706 (disqualify_candidate): Call it.
29707 (build_accesses_from_assign): Reset write flag if creating an
29709 (build_access_subtree): Fill in parent field and also prpagate
29710 down grp_write flag.
29711 (create_artificial_child_access): New parameter set_grp_write, set
29712 grp_write to its value.
29713 (propagate_subaccesses_across_link): Also propagate grp_write flag
29715 (propagate_all_subaccesses): Push the closest parent back to work
29716 queue if add_access_to_work_queue returned true.
29718 2017-05-02 Richard Biener <rguenther@suse.de>
29720 * common.opt (fstrict-overflow): Alias negative to fwrapv.
29721 * doc/invoke.texi (fstrict-overflow): Remove all traces of
29722 -fstrict-overflow documentation.
29723 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
29724 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
29725 flag_strict_overflow.
29726 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
29727 * lto-opts.c (lto_write_options): Do not stream it.
29728 * lto-wrapper.c (merge_and_complain): Do not handle it.
29729 * opts.c (default_options_table): Do not set -fstrict-overflow.
29730 (finish_options): Likewise do not clear it when sanitizing.
29731 * simplify-rtx.c (simplify_const_relational_operation): Do not
29732 test flag_strict_overflow.
29734 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29736 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
29737 using enabled attribute.
29738 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
29739 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
29740 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
29741 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
29742 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
29743 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
29744 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
29745 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
29746 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
29747 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
29749 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29751 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
29753 2017-05-02 Richard Biener <rguenther@suse.de>
29755 PR tree-optimization/80591
29757 2017-04-10 Richard Biener <rguenther@suse.de>
29759 * tree-ssa-structalias.c (find_func_aliases): Properly handle
29762 2017-05-02 Richard Biener <rguenther@suse.de>
29764 PR tree-optimization/80549
29765 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
29766 (cleanup_tree_cfg_noloop): Create forwarders to known loop
29767 headers if they do not have a preheader.
29769 2017-05-02 Martin Liska <mliska@suse.cz>
29772 * common.opt: Fix typo.
29773 * doc/invoke.texi: Likewise.
29775 2017-05-01 Jan Beulich <jbeulich@suse.com>
29777 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
29778 swapping, add (x,x,m,x,n) alternative.
29780 2017-05-01 Nathan Sidwell <nathan@acm.org>
29782 * calls.c (combine_pending_stack_adjustment_and_call): Remove
29783 unnecessary unadjusted_alignment check.
29785 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29788 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
29790 * gimplify.c (gimplify_cilk_detach): New function.
29791 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
29792 * tree-core.h: Document EXPR_CILK_SPAWN.
29793 * tree.h (EXPR_CILK_SPAWN): Define.
29795 2017-05-01 David Malcolm <dmalcolm@redhat.com>
29797 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
29798 to use new fixit_hint representation, using the "replace" logic.
29799 (get_line_span_for_fixit_hint): Likewise.
29800 (layout::print_any_fixits): Likewise.
29801 (selftest::test_one_liner_many_fixits): Rename to...
29802 (selftest::test_one_liner_many_fixits_1): ...this, and update
29803 comment and expected output to reflect that the multiple fix-it
29804 hints are now consolidated into one insertion.
29805 (selftest::test_one_liner_many_fixits_2): New test.
29806 (selftest::test_diagnostic_show_locus_one_liner): Update for
29808 (selftest::test_fixit_consolidation): Update for fix-it API
29810 * diagnostic.c (print_parseable_fixits): Likewise.
29811 * edit-context.c (edited_line::m_line_events): Convert from
29812 auto_vec <line_event *> to auto_vec <line_event>.
29813 (class line_event): Convert from abstract base class to a concrete
29814 class, taking over the role of replace_event.
29815 (class insert_event): Delete.
29816 (class replace_event): Rename to class line_event. Convert to
29818 (edit_context::add_fixits): Reimplement.
29819 (edit_context::apply_insert): Delete.
29820 (edit_context::apply_replace): Rename to...
29821 (edit_context::apply_fixit): ...this. Convert to half-open range.
29822 (edited_file::apply_insert): Delete.
29823 (edited_file::apply_replace): Rename to...
29824 (edited_file::apply_fixit): ...this.
29825 (edited_line::~edited_line): Drop deletion of events.
29826 (edited_line::apply_insert): Delete.
29827 (edited_line::apply_replace): Rename to...
29828 (edited_line::apply_fixit): ...this. Convert to half-open range.
29829 Update for change to type of m_line_events.
29830 * edit-context.h (edit_context::apply_insert): Delete.
29831 (edit_context::apply_replace): Rename to...
29832 (edit_context::apply_fixit): ...this.
29834 2017-05-01 Martin Sebor <msebor@redhat.com>
29836 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
29839 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
29842 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
29843 __get_cpuid_max returns 0.
29844 (__get_cpuid_count): Ditto.
29846 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
29848 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
29849 replacement expression is another instance of one of its arguments.
29851 2017-05-01 Jakub Jelinek <jakub@redhat.com>
29854 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
29855 check for stack push/pop autoinc.
29856 * config/i386/i386.c (ix86_agi_dependent): Return false
29857 if the only reason why modified_in_p returned true is that
29858 addr is SP based and set_insn is a push or pop.
29860 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
29862 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
29865 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
29868 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
29869 (account_size_time): Use two predicates - exec_pred and
29871 (evaluate_conditions_for_known_args): Compute both clause and
29873 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
29874 (inline_summary_t::duplicate): Update.
29875 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
29877 (compute_inline_parameters): Likewise.
29878 (estimate_edge_size_and_time): Update caluclation of time.
29879 (estimate_node_size_and_time): Compute both time and nonspecialized
29881 (estimate_ipcp_clone_size_and_time): Update.
29882 (inline_merge_summary): Update.
29883 (do_estimate_edge_time): Update.
29884 (do_estimate_edge_size): Update.
29885 (do_estimate_edge_hints): Update.
29886 (inline_read_section, inline_write_summary): Stream both new predicates.
29887 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
29889 (compute_inlined_call_time): Cleanup.
29890 (big_speedup_p): Update.
29891 (edge_badness): Update.
29892 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
29893 (size_time_entry): Replace predicate by exec_predicate and
29894 nonconst_predicate.
29895 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
29896 (estimate_edge_time): Return also nonspec_time.
29897 (reset_edge_growth_cache): Update.
29899 2017-04-29 Jakub Jelinek <jakub@redhat.com>
29901 PR rtl-optimization/80491
29902 * ifcvt.c (noce_process_if_block): When looking for x setter
29903 with missing else_bb, don't check only the insn right before
29904 cond_earliest, but look for the last insn that x is modified in
29905 within the same bb.
29907 PR rtl-optimization/80491
29908 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
29910 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
29912 PR tree-optimization/80487
29913 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
29915 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29917 PR tree-optimization/79697
29918 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
29919 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
29920 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
29922 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
29923 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
29925 2017-04-28 Martin Sebor <msebor@redhat.com>
29927 PR tree-optimization/80523
29928 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
29929 (init_target_to_host_charmap, target_to_host, target_strtol10): New
29931 (maybe_warn, format_directive, parse_directive): Use new functions.
29932 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
29934 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
29936 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
29938 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
29940 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
29941 target_header_dir): Set correctly.
29942 * configure: Regenerated.
29943 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29944 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29945 instead of SYSTEM_HEADER_DIR.
29947 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
29949 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
29950 (estimate_local_effects): Likewise.
29951 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
29952 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
29953 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
29954 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
29955 do_estimate_edge_time, estimate_edge_time): Likewise.
29956 * ipa-inline-analysis.c (estimate_node_size_and_time,
29957 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
29958 (estimate_time_after_inlining): Remove.
29960 2017-04-28 Martin Liska <mliska@suse.cz>
29962 * doc/gcov.texi: Enhance documentation of gcov.
29964 2017-04-28 Martin Liska <mliska@suse.cz>
29966 * doc/gcov.texi: Sort options in alphabetic order.
29967 * doc/gcov-dump.texi: Likewise.
29968 * doc/gcov-tool.texi: Likewise.
29969 * gcov.c (print_usage): Likewise.
29970 * gcov-dump.c (print_usage): Likewise.
29971 * gcov-tool.c (print_merge_usage_message): Likewise.
29972 (print_rewrite_usage_message): Likewise.
29973 (print_overlap_usage_message): Likewise.
29975 2017-04-28 Martin Liska <mliska@suse.cz>
29977 PR gcov-profile/53915
29978 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
29980 2017-04-28 Martin Liska <mliska@suse.cz>
29982 PR gcov-profile/79891
29983 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
29984 is marked by compiler as living on a line.
29985 (get_cycles_count): Remove usage of the union.
29986 (output_intermediate_file): Likewise.
29987 (find_source): Fix GNU coding style.
29988 (accumulate_line_counts): Remove old non-all block mode.
29989 (output_lines): Remove usage of the union.
29990 * profile.c (output_location): Include all BBs, even if
29991 belonging to a same line (and file) as a previous BB.
29993 2017-04-28 Martin Liska <mliska@suse.cz>
29995 * gcov.c (process_args): Handle new argument 'w'.
29996 (read_graph_file): Assign ID to BBs.
29997 (output_branch_count): Display BB # if verbose flag is set.
29998 (output_lines): Likewise for arcs.
29999 (print_usage): Add '--verbose' option help.
30000 * doc/gcov.texi: Document --verbose (-w) option.
30002 2017-04-28 Martin Liska <mliska@suse.cz>
30004 * gcov.c (struct block_location_info): New struct.
30005 (process_file): Fill up the new structure.
30006 (read_graph_file): Replace usage of encoding by the newly added
30008 (add_line_counts): Likewise.
30009 (accumulate_line_counts): Remove usage of the union.
30010 (function_info::function_info): New function.
30011 (function_info::~function_info): Likewise.
30012 (process_file): Call delete instead of release_function.
30013 (release_function): Release the function.
30014 (release_structures): Call delete instead of release_function.
30015 (solve_flow_graph): Replace usage of num_blocks.
30016 (find_exception_blocks): Likewise.
30017 (output_lines): Fix GNU coding style.
30019 2017-04-28 Martin Liska <mliska@suse.cz>
30022 * coverage.c (coverage_remove_note_file): New function.
30023 * coverage.h: Declare the function.
30024 * toplev.c (finalize): Clean if an error has been seen.
30026 2017-04-28 Martin Liska <mliska@suse.cz>
30028 PR gcov-profile/80031
30029 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
30030 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
30031 * gcov.c (read_graph_file): Read just number of blocks.
30032 * profile.c (branch_prob): Do not stream 0 flags per a basic
30035 2017-04-28 Martin Liska <mliska@suse.cz>
30037 * gcov-dump.c (tag_*): Add new argument to declarations.
30038 (dump_gcov_file): Likewise.
30039 (tag_blocks): Add and use new argument depth.
30040 (tag_arcs): Likewise.
30041 (tag_lines): Likewise.
30042 (tag_counters): Likewise.
30043 (tag_summary): Likewise.
30044 (dump_working_sets): Use depth to do a proper indentation.
30046 2017-04-28 Jakub Jelinek <jakub@redhat.com>
30049 * cgraph.h (symtab_node::debug_symtab): No longer inline.
30050 * symtab.c (symtab_node::debug_symtab): Move definition here.
30052 2017-04-28 Richard Biener <rguenther@suse.de>
30054 * lto-streamer.h (LTO_major_version): Bump to 7.
30056 2017-04-28 Richard Biener <rguenther@suse.de>
30058 * tree-vrp.c (assert_info): New struct.
30059 (add_assert_info): New helper.
30060 (register_edge_assert_for_2): Refactor to add asserts to a vector
30062 (register_edge_assert_for_1): Likewise.
30063 (register_edge_assert_for): Likewise.
30064 (finish_register_edge_assert_for): New helper actually registering
30065 asserts where live on edge.
30066 (find_conditional_asserts): Adjust.
30067 (find_switch_asserts): Likewise.
30068 (evrp_dom_walker::try_find_new_range): Generalize.
30069 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
30071 2017-04-27 Marek Polacek <polacek@redhat.com>
30074 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
30075 arg10 and arg11 to itype.
30077 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
30079 * doc/extend.texi (Object Size Checking): Improve grammar.
30081 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
30084 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
30085 that the logic for permitting reciprocal estimates matches that
30088 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30091 * tree.c (type_cache_hasher::equal): Only compare
30092 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
30093 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
30094 non-aggregate element types.
30095 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
30096 about the flag on ARRAY_TYPEs in the comment, formatting fix.
30098 2017-04-27 Richard Biener <rguenther@suse.de>
30100 PR middle-end/80533
30101 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
30102 stripping ARRAY_REFs from MEM_EXPR make sure we're not
30103 keeping a reference to a trailing array.
30105 2017-04-27 Richard Biener <rguenther@suse.de>
30107 PR middle-end/80539
30108 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
30109 being in loop-closed SSA form conservatively.
30110 (chrec_fold_multiply_poly_poly): Likewise.
30112 2017-04-27 Tamar Christina <tamar.christina@arm.com>
30114 PR middle-end/79665
30115 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
30116 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
30118 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30121 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
30122 (aarch64_function_arg_alignment): Return unsigned int again, but still
30123 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
30124 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30125 Don't emit -Wpsabi note.
30126 (aarch64_function_arg_boundary): Likewise.
30127 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30130 2017-04-26 Nathan Sidwell <nathan@acm.org>
30132 * tree.h (crc32_unsigned_n): Declare.
30133 (crc32_unsigned, crc32_unsigned): Make inline.
30134 * tree.c (crc32_unsigned_bits): Replace with ...
30135 (crc32_unsigned_n): ... this.
30136 (crc32_unsigned, crc32_byte): Remove.
30137 (crc32_string): Remove unnecessary braces.
30139 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30141 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
30142 * ipa-inline-analysis.c (MAX_TIME): Remove.
30143 (account_size_time): Use sreal for time.
30144 (dump_inline_summary): Update.
30145 (estimate_function_body_sizes): Update.
30146 (estimate_edge_size_and_time): Update.
30147 (estimate_calls_size_and_time): Update.
30148 (estimate_node_size_and_time): Update.
30149 (inline_merge_summary): Update.
30150 (inline_update_overall_summary): Update.
30151 (estimate_time_after_inlining): Update.
30152 (inline_read_section): Update.
30153 (inline_write_summary): Update.
30154 * ipa-inline.c (compute_uninlined_call_time): Update.
30155 (compute_inlined_call_time): Update.
30156 (recursive_inlining): Update.
30157 (inline_small_functions): Update.
30158 (dump_overall_stats): Update.
30159 * ipa-inline.h: Include sreal.h.
30160 (size_time_entry): Turn time to sreal.
30161 (inline_summary): Turn self_time nad time to sreal.
30163 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30165 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
30167 (sreal::stream_out, sreal::stream_in): New.
30168 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
30170 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30172 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
30175 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
30178 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
30180 (dimode_scalar_chain::compute_convert_gain): Ditto.
30181 (dimode_scalar_chain::make_vector_copies): Ditto.
30182 (dimode_scalar_chain::convert_reg): Ditto.
30183 (dimode_scalar_chain::convert_insn): Ditto.
30184 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
30185 (VI248_AVX512BW_1): New mode iterator.
30186 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
30187 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
30190 2017-04-25 Martin Sebor <msebor@redhat.com>
30192 PR tree-optimization/80497
30193 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
30194 constants are representable in HOST_WIDE_INT.
30195 (parse_directive): Ditto.
30197 2017-04-25 Martin Sebor <msebor@redhat.com>
30200 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
30201 (new_zero_array): Adjust signature.
30202 (dom_info::dom_init): Used unsigned rather that size_t.
30203 (dom_info::dom_info): Same.
30205 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30206 Jakub Jelinek <jakub@redhat.com>
30209 * config/arm/arm.c: Include gimple.h.
30210 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30211 returns negative, increment ncrn only if it returned positive.
30212 (arm_needs_doubleword_align): Return int instead of bool,
30213 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
30214 members, but if there is any such non-FIELD_DECL
30215 > PARM_BOUNDARY aligned decl, return -1 instead of false.
30216 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30217 returns negative, increment nregs only if it returned positive.
30218 (arm_setup_incoming_varargs): Likewise.
30219 (arm_function_arg_boundary): Emit -Wpsabi note if
30220 arm_needs_doubleword_align returns negative, return
30221 DOUBLEWORD_ALIGNMENT only if it returned positive.
30223 2017-04-25 Marek Polacek <polacek@redhat.com>
30226 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30227 first argument to type.
30229 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
30232 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
30233 type checks to test for compatibility instead of equality.
30235 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30236 Jakub Jelinek <jakub@redhat.com>
30239 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
30241 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
30242 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
30243 the alignment computation, but return their maximum in warn_alignment.
30244 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30245 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
30247 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
30248 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30251 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30253 * config/arc/simdext.md (dmpyh): Fix typo.
30255 2017-04-25 Richard Biener <rguenther@suse.de>
30257 PR tree-optimization/80492
30258 * alias.c (compare_base_decls): Handle registers with asm
30259 specification conservatively.
30260 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
30261 compare_base_decls returning dont-know properly.
30263 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30265 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
30266 (legitimate_offset_address_p): New function.
30267 (arc_legitimate_address_p): Use above function.
30269 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30271 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
30273 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30275 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
30276 ACCH registers whenever they are available.
30278 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30280 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
30281 double regs fix when not used.
30283 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30285 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
30287 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
30288 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
30290 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30292 * config/arc/arc.c (arc_output_addsi): Check for h-register class
30293 when emitting short ADD instructions.
30295 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30297 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
30299 (cmpsi_cc_c_insn): Likewise.
30300 (cbranchsi4_scratch): Compute proper instruction length using
30301 compact_hreg_operand.
30302 * config/arc/predicates.md (compact_hreg_operand): New predicate.
30304 2017-04-25 Richard Biener <rguenther@suse.de>
30306 PR middle-end/80509
30307 * passes.c (pass_manager::pass_manager): Initialize
30308 m_name_to_pass_map.
30310 2017-04-25 Richard Biener <rguenther@suse.de>
30312 PR tree-optimization/79201
30313 * tree-ssa-sink.c (statement_sink_location): Handle calls.
30315 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30318 * config/s390/vector.md: Split MEM->GPR vector moves for
30319 non-s_operand addresses.
30321 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30324 * config/s390/predicates.md (reload_const_wide_int_operand): New
30326 * config/s390/s390.md ("movti"): Remove d/P alternative.
30327 ("movti_bigconst"): New pattern definition.
30329 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30332 * s390-protos.h (s390_expand_cs_hqi): Removed.
30333 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
30334 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
30335 modes as well as CCZ1mode and CCZmode.
30336 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
30337 signature of s390_emit_compare_and_swap.
30338 (s390_expand_cs_hqi): Likewise, make static.
30339 (s390_expand_cs_tdsi): Generate an explicit compare before trying
30340 compare-and-swap, in some cases.
30341 (s390_expand_cs): Wrapper function.
30342 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
30344 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
30345 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
30346 patterns for small and large integers. Forbid symref memory operands.
30347 Move expander to s390.c. Require cc register.
30348 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
30349 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
30350 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
30351 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
30352 symref memory operands. Remove CC mode and call s390_match_ccmode
30354 ("atomic_exchange<mode>"): Allow and implement all integer modes.
30356 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30358 * config/s390/s390.md (define_peephole2): New peephole to help
30359 combining the load-and-test pattern with volatile memory.
30361 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30363 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
30364 with CCZmode for TARGET_Z196.
30366 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30368 PR rtl-optimization/80501
30369 * combine.c (make_compound_operation_int): Set subreg_code to SET
30370 even for AND with mask of the sign bit of mode.
30372 PR rtl-optimization/80500
30373 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
30374 sum's initial value.
30376 2017-04-25 Julian Brown <julian@codesourcery.com>
30377 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30379 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
30381 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
30383 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
30385 2017-04-25 Julian Brown <julian@codesourcery.com>
30386 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30388 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
30389 (thunderx2t99_sha): New Reservation.
30391 2017-04-25 Julian Brown <julian@codesourcery.com>
30392 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30394 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
30395 type for 1-element load.
30397 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
30399 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
30401 2017-04-24 Martin Jambor <mjambor@suse.cz>
30403 PR tree-optimization/80293
30404 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
30405 char arrays not totally scalarizable if it is false.
30406 (analyze_all_variable_accesses): Pass correct value in the new
30407 parameter. Add a statistics counter.
30409 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
30411 PR middle-end/79931
30412 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
30414 2017-04-24 Richard Biener <rguenther@suse.de>
30416 PR tree-optimization/80494
30417 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
30418 out for complex types.
30420 2017-04-24 Richard Biener <rguenther@suse.de>
30422 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
30423 * tree-ssa-sccvn.c (print_scc): Print SCC size.
30424 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
30425 (DFS): Adjust and never fail.
30426 (sccvn_dom_walker::fail): Remove.
30427 (sccvn_dom_walker::before_dom_children): Adjust.
30428 (run_scc_vn): Likewise and never fail.
30429 * tree-ssa-pre.c (pass_pre::execute): Adjust.
30430 (pass_fre::execute): Likewise.
30432 2017-04-24 Richard Biener <rguenther@suse.de>
30434 PR tree-optimization/79725
30435 * tree-ssa-sink.c (statement_sink_location): Return whether
30436 failure reason was zero uses. Move that check later.
30437 (sink_code_in_bb): Deal with zero uses by removing the stmt
30440 2017-04-24 Richard Biener <rguenther@suse.de>
30443 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
30444 pointer-based references.
30446 2017-04-24 Richard Biener <rguenther@suse.de>
30449 * pass_manager.h (pass_manager::operator new): Remove.
30450 (pass_manager::operator delete): Likewise.
30451 * passes.c (pass_manager::operator new): Remove.
30452 (pass_manager::operator delete): Likewise.
30453 (pass_manager::pass_manager): Zero individual pass members.
30455 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
30458 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
30459 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
30460 Check "XEXP (src, 1)" operand here.
30461 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
30462 Check "XEXP (src, 1)" operand here.
30463 (dimode_scalar_chain::make_vector_copies): Detect count register
30464 of a shift instruction. Zero extend count register from QImode
30465 to DImode to satisfy vector shift pattern count operand predicate.
30466 Substitute vector shift count operand with a DImode copy.
30467 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
30470 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
30472 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
30473 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
30474 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
30475 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
30476 (UNSPEC_NOREX_MEM): Remove definition.
30478 2017-04-21 Richard Biener <rguenther@suse.de>
30480 PR tree-optimization/79547
30481 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
30482 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
30483 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
30484 without any constraints.
30486 2017-04-21 Richard Biener <rguenther@suse.de>
30488 PR tree-optimization/78847
30489 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
30491 2017-04-21 Richard Biener <rguenther@suse.de>
30493 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
30494 (build_distinct_type_copy): Likewise.
30495 (build_variant_type_copy): Likewise.
30496 * tree.c (build_qualified_type): Pass down mem-stat info.
30497 (build_distinct_type_copy): Likewise.
30498 (build_variant_type_copy): Likewise.
30500 2017-04-21 Richard Biener <rguenther@suse.de>
30502 PR tree-optimization/80237
30503 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
30505 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
30506 for a simplified result.
30508 2016-04-21 Richard Biener <rguenther@suse.de>
30510 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
30511 sth as strict as a simple_iv but a chrec without symbols and an
30512 operand defined in the loop we are peeling (and not some subloop).
30513 (propagate_constants_for_unrolling): Propagate all constants.
30515 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
30518 * config/i386/i386.c (print_reg): Remove assert for disalowed
30519 regno values, call output_operand_lossage instead.
30521 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
30524 * config/i386/constraints.md (Yc): New register constraint.
30525 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
30526 Use Yc constraint for alternative 2 of operand 0. Remove
30527 preferred_for_speed attribute.
30529 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
30531 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
30532 lastprivate clauses in SIMT case.
30534 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
30536 * doc/invoke.texi (-Wextra-semi): Document new warning option.
30538 2017-04-20 Richard Biener <rguenther@suse.de>
30540 PR tree-optimization/57796
30541 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
30542 as N scalar stores.
30543 (vect_model_load_cost): Cost gathers as N scalar loads.
30545 2017-04-20 Richard Biener <rguenther@suse.de>
30547 * ggc-page.c (ggc_allocated_p): Rename to ...
30548 (safe_lookup_page_table_entry): ... this and return the lookup
30550 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
30552 2017-04-20 Richard Biener <rguenther@suse.de>
30554 PR tree-optimization/80453
30555 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
30556 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
30557 from the conditions.
30558 (vn_phi_eq): Pass them down.
30559 (vn_phi_lookup): Record them.
30560 (vn_phi_insert): Likewise.
30562 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
30564 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
30565 uninitialized variable warning to avoid buffer overrun.
30567 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
30570 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
30571 is suppressed for '{ 0 }' in C.
30573 2017-04-20 Jakub Jelinek <jakub@redhat.com>
30575 * BASE-VER: Set to 8.0.0.
30577 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
30579 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
30580 priority .init_array and .fini_array section with SECTION_NOTYPE
30583 2017-04-20 Jakub Jelinek <jakub@redhat.com>
30585 PR middle-end/80423
30586 * tree.h (build_array_type): Add typeless_storage default argument.
30587 * tree.c (type_cache_hasher::equal): Also compare
30588 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
30589 (build_array_type): Add typeless_storage argument, set
30590 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
30592 (build_nonshared_array_type): Adjust build_array_type_1 caller.
30593 (build_array_type): Likewise. Add typeless_storage argument.
30595 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
30596 Jakub Jelinek <jakub@redhat.com>
30598 PR tree-optimization/80426
30599 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
30600 operation on symbolic operands, also compute the overflow for the
30601 invariant part when the operation degenerates into a negation.
30603 2017-04-19 Jakub Jelinek <jakub@redhat.com>
30606 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
30607 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
30610 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
30612 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
30615 * config/avr/avr.c (tree.h): Include it.
30616 (cgraph.h): Include it.
30617 (avr_encode_section_info): Don't warn for uninitialized progmem
30618 variable if it's just an alias.
30620 2017-04-19 Richard Biener <rguenther@suse.de>
30623 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
30624 when needed by AutoPGO.
30626 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
30629 * doc/lto.texi: Remove an extra 'that'.
30631 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
30633 PR rtl-optimization/80429
30634 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
30635 are only used in debug insns.
30637 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
30638 Vladimir Makarov <vmakarov@redhat.com>
30640 * config/sparc/predicates.md (input_operand): Add comment. Return
30641 true for any memory operand when LRA is in progress.
30642 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
30644 2017-04-18 Jeff Law <law@redhat.com>
30647 * mips.md ({return,simple_return}_internal): Do not overwrite
30650 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30652 PR tree-optimization/80443
30653 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
30654 instead of adding 1, subtract -1 and similarly instead of subtracting
30657 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
30659 PR rtl-optimization/80357
30660 * haifa-sched.c (tmp_bitmap): New variable.
30661 (model_recompute): Handle duplicate use records.
30662 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
30663 (free_global_sched_pressure_data): Free it.
30665 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
30668 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
30669 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30670 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30671 instead of SYSTEM_HEADER_DIR.
30673 2017-04-18 Jeff Law <law@redhat.com>
30675 PR middle-end/80422
30676 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
30677 predecessors after walking up the insn chain.
30679 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30682 * dwarf2out.c (modified_type_die): Try harder not to emit internal
30683 sizetype type into debug info.
30685 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30688 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
30689 unneeded test for TARGET_UPPER_REGS_SF.
30690 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
30692 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30695 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
30696 instead of gsi_after_labels.
30698 2017-04-18 Jeff Law <law@redhat.com>
30700 * regcprop.c (maybe_mode_change): Avoid creating copies of the
30704 2017-04-13 Jeff Law <law@redhat.com>
30705 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30706 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30708 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
30711 * config/avr/avr.c (intl.h): Include it.
30712 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
30714 2017-04-18 Martin Liska <mliska@suse.cz>
30716 PR gcov-profile/78783
30717 * gcov-tool.c (gcov_output_files): Validate that destination
30718 file is either removed by the tool or by a user.
30720 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30721 Guy Benyei <guybe@mellanox.com>
30723 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
30724 block, and do not negate it, the stored id is already negative.
30726 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30728 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
30730 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
30733 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
30734 masks of options that should be turned off if the VSX vector
30735 options are turned off.
30736 (OTHER_P8_VECTOR_MASKS): Likewise.
30737 (OTHER_VSX_VECTOR_MASKS): Likewise.
30738 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
30739 rs6000_disable_incompatible_switches to validate no type switches
30741 (rs6000_incompatible_switch): New function to disallow turning on
30742 other vector options if -mno-vsx, -mno-power8-vector, or
30743 -mno-power9-vector are specified.
30745 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30747 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
30749 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30751 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
30752 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
30753 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
30754 (ARG_POINTER_CFA_OFFSET): Likewise.
30756 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30758 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
30759 conditions to take advantage of various optimizations.
30761 2017-04-13 Jeff Law <law@redhat.com>
30763 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30764 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30765 (zero_extendsidi2_dext): Likewise.
30767 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30770 * fold-const.c (fold_ternary_loc): Revert
30771 use op0 instead of fold_convert_loc (loc, type, arg0) part of
30774 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
30776 PR rtl-optimization/80343
30777 * lra-remat.c (update_scratch_ops): Assign original hard reg to
30778 new scratch pseudo.
30780 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
30783 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
30784 to ubsan_encode_value.
30786 2017-04-13 Jeff Law <law@redhat.com>
30788 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
30789 appearing in DEBUG_INSNs.
30791 2017-04-13 Martin Liska <mliska@suse.cz>
30793 PR gcov-profile/80413
30794 * gcov-io.c (gcov_write_string): Copy to buffer just when
30795 allocated size is greater than zero.
30797 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30800 * dwarf2out.c (decls_for_scope): Ignore declarations of
30801 current_function_decl in BLOCK_NONLOCALIZED_VARS.
30803 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
30806 * ipa-visibility.c (non_local_p): Fix typos.
30807 (localize_node): When localizing symbol in same comdat group,
30808 dissolve the group only when we know external symbols are going
30810 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
30812 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30814 PR tree-optimization/79390
30815 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
30816 order does not result in usable sequence, retry with reversed operand
30822 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
30823 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
30824 op0 instead of fold_convert_loc (loc, type, arg0).
30826 2017-04-12 Jeff Law <law@redhat.com>
30828 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
30829 has a delay slot in the generated code.
30831 * config/cris/cris.md (cris_preferred_reload_class): Return
30832 GENNONACR_REGS rather than GENERAL_REGS.
30834 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30837 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
30838 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
30839 signedness of the result type.
30841 2017-04-12 Richard Biener <rguenther@suse.de>
30842 Jeff Law <law@redhat.com>
30844 PR tree-optimization/80359
30845 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
30846 trim stores to TARGET_MEM_REFs.
30848 2017-04-12 Richard Biener <rguenther@suse.de>
30850 PR tree-optimization/79390
30851 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
30852 threading case even more.
30854 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
30857 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
30858 for quad_address_p for TImode, instead of just not indexed_address.
30860 2017-04-12 Richard Biener <rguenther@suse.de>
30861 Bernd Edlinger <bernd.edlinger@hotmail.de>
30863 PR middle-end/79671
30864 * alias.c (component_uses_parent_alias_set_from): Handle
30865 TYPE_TYPELESS_STORAGE.
30866 (get_alias_set): Likewise.
30867 * tree-core.h (tree_type_common): Add typeless_storage flag.
30868 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
30869 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
30870 for types containing members with TYPE_TYPELESS_STORAGE.
30871 (place_field): Likewise.
30872 (layout_type): Likewise for ARRAY_TYPE.
30873 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
30874 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
30875 TYPE_TYPELESS_STORAGE.
30876 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
30878 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30881 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30882 first argument to type.
30884 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30888 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
30889 CONST0_RTX (mode) rather than const0_rtx where appropriate.
30890 (rs6000_expand_binop_builtin): Likewise.
30891 (rs6000_expand_ternop_builtin): Likewise; also add missing
30892 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
30893 vshasigma built-ins.
30894 * doc/extend.texi: Document that vec_xxpermdi's third argument
30895 must be a constant.
30897 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
30899 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
30900 Use shift_const cost parameter when calculating gain of STV shifts.
30902 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
30904 PR rtl-optimization/70478
30905 * lra-constraints.c (process_alt_operands): Check memory for
30906 disfavoring memory insn operand.
30908 2017-04-11 Jakub Jelinek <jakub@redhat.com>
30910 PR middle-end/80100
30911 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
30912 left shift in unsigned HOST_WIDE_INT type.
30914 PR rtl-optimization/80385
30915 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
30916 (not (neg X)) into (plus X -1) for complex or non-integral modes.
30919 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
30920 if they have any depend clauses.
30922 2017-04-11 Martin Liska <mliska@suse.cz>
30925 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
30926 * ipa-split.c (split_function): Create a local comdat symbol
30927 if caller is in a comdat group.
30929 2017-04-11 Martin Liska <mliska@suse.cz>
30932 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
30935 2017-04-11 Martin Sebor <msebor@redhat.com>
30937 PR middle-end/80364
30938 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
30939 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
30941 (directive::set_width, directive::set_precision, format_character):
30943 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
30946 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
30949 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
30950 conflict, set target->arch_name instead of target->cpu_name.
30952 2017-04-11 Richard Biener <rguenther@suse.de>
30954 PR tree-optimization/80374
30955 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
30956 build_zero_cst, remove fold_convertible_p check again.
30958 2017-04-11 Martin Liska <mliska@suse.cz>
30961 * ubsan.c (instrument_object_size): Do not instrument register
30964 2017-04-11 Jakub Jelinek <jakub@redhat.com>
30967 * config/i386/i386-builtin-types.def
30968 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
30969 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
30970 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
30971 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
30972 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
30973 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
30974 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
30975 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
30976 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
30977 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
30978 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
30979 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
30980 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
30981 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
30982 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
30983 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
30984 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
30985 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
30986 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
30987 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
30988 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
30989 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
30990 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
30991 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
30992 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
30993 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
30994 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
30995 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
30996 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
30997 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
30998 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
30999 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
31000 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
31001 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
31002 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
31003 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
31004 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
31005 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
31006 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
31007 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
31008 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
31009 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
31010 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
31011 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
31012 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
31013 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
31015 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
31016 flag to second_arg_count, handle 4 argument function type _COUNT
31017 aliases, handle second_arg_count on second argument rather than last.
31019 2017-04-10 Jeff Law <law@redhat.com>
31021 PR tree-optimization/80374
31022 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
31023 record anything if we can not convert integer_zero_node to the
31026 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
31029 * config/rs6000/rs6000.c (rs6000_option_override_internal):
31030 Enhance special handling given to the TARGET_P9_MINMAX option in
31031 relation to certain other options.
31033 2017-04-10 Bin Cheng <bin.cheng@arm.com>
31035 PR tree-optimization/80153
31036 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
31037 remove POINTER_PLUS_EXPR's base part directly, rather than through
31040 2017-04-10 Richard Biener <rguenther@suse.de>
31041 Bin Cheng <bin.cheng@arm.com>
31043 PR tree-optimization/80153
31044 * tree-affine.c (aff_combination_to_tree): Get base pointer from
31045 the first element of pointer type aff_tree. Build result expr in
31047 (add_elt_to_tree): Convert to type unconditionally. Remove other
31048 fold_convert calls.
31049 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
31050 (rewrite_use_nonlinear_expr): Check invariant using iv information.
31052 2017-04-10 Richard Biener <rguenther@suse.de>
31054 * tree-ssa-structalias.c (find_func_aliases): Properly handle
31057 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
31059 PR rtl-optimization/70478
31060 * lra-constraints.c (curr_small_class_check): New.
31061 (update_and_check_small_class_inputs): New.
31062 (process_alt_operands): Update curr_small_class_check. Disfavor
31063 alternative insn memory operands. Check available regs for small
31066 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31069 * config/mips/mips.opt (-mvirt): Update description.
31070 * doc/invoke.texi (-mvirt): Likewise.
31072 2017-04-10 Richard Biener <rguenther@suse.de>
31074 PR middle-end/80362
31075 * fold-const.c (fold_binary_loc): Look at unstripped ops when
31076 looking for NEGATE_EXPR in -A / -B to A / B folding.
31078 2017-04-10 Martin Liska <mliska@suse.cz>
31080 PR gcov-profile/80224
31081 * gcov.c (print_usage): Fix usage string.
31082 (get_gcov_intermediate_filename): Remove.
31083 (output_gcov_file): Use both for normal and intermediate format.
31084 (generate_results): Do not initialize special file for
31085 intermediate format.
31087 2017-04-10 Richard Biener <rguenther@suse.de>
31089 PR tree-optimization/80304
31090 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
31093 2017-04-10 Nathan Sidwell <nathan@acm.org>
31096 * config/rs6000/rs6000.c (rs6000_vector_type): New.
31097 (rs6000_init_builtins): Use it.
31099 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31101 * config/arm/arm.md (<mrc>): Add mode to SET source.
31102 (<mrrc>): Likewise.
31104 2017-04-10 Richard Biener <rguenther@suse.de>
31106 PR middle-end/80344
31107 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
31109 2017-04-10 Jakub Jelinek <jakub@redhat.com>
31112 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
31113 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
31114 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
31115 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
31116 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
31117 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
31118 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
31119 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
31120 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
31121 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
31122 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
31123 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
31124 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
31125 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
31126 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
31127 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
31128 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
31129 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
31130 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
31131 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
31132 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
31133 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
31134 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
31136 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
31138 PR rtl-optimization/70478
31139 * lra-constraints.c: Reverse the last patch.
31141 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
31143 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
31144 Add comment for WCHAR_T.
31146 2017-04-08 Martin Liska <mliska@suse.cz>
31149 2017-04-07 Martin Liska <mliska@suse.cz>
31152 * ipa-split.c (split_function): Add function part to a same comdat
31155 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31158 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
31160 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
31162 * rs6000/rs6000.c (vec_load_pendulum): Rename...
31163 (vec_pairing): ...to this.
31164 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
31165 (rs6000_sched_init): Adjust for name change.
31166 (struct rs6000_sched_context): Likewise.
31167 (rs6000_init_sched_context): Likewise.
31168 (rs6000_set_sched_context): Likewise.
31170 2017-04-07 Jakub Jelinek <jakub@redhat.com>
31176 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
31178 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
31179 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
31180 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
31182 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
31184 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
31186 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31188 PR rtl-optimization/70703
31189 * ira-color.c (update_conflict_hard_regno_costs): Use
31190 int64_t instead of HOST_WIDE_INT.
31192 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31194 PR rtl-optimization/70478
31195 * lra-constraints.c (process_alt_operands): Disfavor alternative
31196 insn memory operands.
31198 2017-04-07 Jeff Law <law@redhat.com>
31200 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
31201 CALL and NOTE_INSN_CALL_ARG_LOCATION.
31203 2017-04-07 Martin Liska <mliska@suse.cz>
31206 * config/aarch64/aarch64.c (aarch64_process_target_attr):
31207 Show error message instead of an ICE.
31209 2017-04-07 Martin Liska <mliska@suse.cz>
31212 * ipa-split.c (split_function): Add function part to a same comdat
31215 2017-04-07 Richard Biener <rguenther@suse.de>
31217 PR middle-end/80341
31218 * tree.c (get_unwidened): Also handle ! for_type case for
31220 * convert.c (do_narrow): Split out from ...
31221 (convert_to_integer_1): ... here. Do not pass final truncation
31222 type to get_unwidened for TRUNC_DIV_EXPR.
31224 2017-04-07 Richard Biener <rguenther@suse.de>
31226 * tree-affine.c (wide_int_ext_for_comb): Take type rather
31228 (aff_combination_const): Adjust.
31229 (aff_combination_scale): Likewise.
31230 (aff_combination_add_elt): Likewise.
31231 (aff_combination_add_cst): Likewise.
31232 (aff_combination_convert): Likewise.
31233 (add_elt_to_tree): Likewise. Remove unused argument.
31234 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
31236 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
31238 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
31240 * config/arm/arm.c (arm_default_short_enums): Use
31241 ARM_DEFAULT_SHORT_ENUMS.
31242 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
31244 2017-04-06 Jakub Jelinek <jakub@redhat.com>
31247 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
31248 members with redundant out-of-class redeclaration.
31250 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31253 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
31254 * config/i386/i386.md (*zero_extendsidi2):
31255 Add (?*x,*x) and (?*v,*v) alternatives.
31257 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31260 * config/i386/i386.c (ix86_expand_builtin)
31261 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
31262 mode from insn data. Convert operands to insn operand mode.
31263 Copy operands that don't satisfy insn predicate to a register.
31265 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
31267 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
31270 2017-04-06 Richard Biener <rguenther@suse.de>
31272 PR tree-optimization/80334
31273 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
31274 preserve alignment of accesses.
31276 2017-04-06 Richard Biener <rguenther@suse.de>
31278 PR tree-optimization/80262
31279 * tree-sra.c (build_ref_for_offset): Preserve address-space
31281 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
31282 Drop useless address-space information on MEM_REF offsets.
31284 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
31286 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
31288 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31290 PR rtl-optimization/70703
31291 * ira-color.c (update_conflict_hard_regno_costs): Use
31292 HOST_WIDE_INT instead of long.
31294 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
31297 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
31298 not defined for x86_64 target. Add -mmmx target option when __SSE2__
31300 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
31301 for x86_64 target. Handle -m3dnowa option.
31303 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31305 PR rtl-optimization/70703
31306 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
31307 (update_conflict_hard_regno_costs): Use long instead of unsigned
31308 arithmetic for cost calculation.
31310 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31311 Bernd Edlinger <bernd.edlinger@hotmail.de>
31314 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
31317 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
31320 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
31321 ptr_mode with Pmode throughout.
31322 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
31323 into probe_stack_range and use DImode.
31325 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
31328 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
31329 call_eh_return is true.
31331 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31333 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
31334 Initialize last_match_fntype_index.
31336 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31339 * tree-nvr.c: Include internal-fn.h.
31340 (pass_return_slot::execute): Ignore internal calls without
31343 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31344 Richard Biener <rguenther@suse.de>
31347 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
31348 captures used multiple times, except for the last use.
31349 * generic-match-head.c: Include gimplify.h.
31351 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31353 PR tree-optimization/79390
31354 * target.h (struct noce_if_info): Declare.
31355 * targhooks.h (default_noce_conversion_profitable_p): Declare.
31356 * target.def (noce_conversion_profitable_p): New target hook.
31357 * ifcvt.h (struct noce_if_info): New type, moved from ...
31358 * ifcvt.c (struct noce_if_info): ... here.
31359 (noce_conversion_profitable_p): Renamed to ...
31360 (default_noce_conversion_profitable_p): ... this. No longer
31362 (noce_try_store_flag_constants, noce_try_addcc,
31363 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
31364 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
31365 instead of noce_conversion_profitable_p.
31366 * config/i386/i386.c: Include ifcvt.h.
31367 (ix86_option_override_internal): Don't override
31368 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
31369 (ix86_noce_conversion_profitable_p): New function.
31370 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
31371 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
31372 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
31373 * doc/tm.texi: Regenerated.
31375 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31377 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
31380 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31383 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
31384 instructions for small multiply cores.
31386 2017-04-04 Jeff Law <law@redhat.com>
31388 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
31390 (mips_expand_vec_perm_const): Initialize elements in orig_perm
31391 that are not set by the loop over the elements.
31393 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31396 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
31397 int mode, convert_modes it to mode as unsigned, otherwise use
31398 lowpart_subreg to mode rather than SImode.
31399 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
31400 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
31401 Use DImode instead of SImode for the shift count operand.
31402 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
31405 2017-04-04 Richard Biener <rguenther@suse.de>
31407 PR middle-end/80281
31408 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
31409 arithmetic done for the negate or the plus. Simplify.
31410 (A - (-B) -> A + B): Likewise.
31411 * fold-const.c (split_tree): Make sure to not negate pointers.
31413 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
31415 PR rtl-optimization/60818
31416 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
31417 a compare of comparisons with the thing compared if this results
31418 in a different machine mode.
31420 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
31422 * alias.c (base_alias_check): Fix typo in comment.
31423 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
31424 * cgraphunit.c (symbol_table::compile): Likewise.
31425 * collect2.c (maybe_run_lto_and_relink): Likewise.
31426 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
31427 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
31428 * config/avr/avr.c (avr_map_op_t): Likewise.
31429 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
31430 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
31431 * config/epiphany/epiphany.md (movcc): Likewise.
31432 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
31433 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
31435 * config/mips/mips.c (mips_save_restore_reg): Likewise.
31436 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
31437 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
31438 * config/sh/sh.c (sh_rtx_costs): Likewise.
31439 * fold-const.c (fold_truth_andor): Likewise.
31440 * genautomata.c (collapse_flag): Likewise.
31441 * gengtype.h (struct type::u::s): Likewise.
31442 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
31443 * input.c (FORMAT_AMOUNT): Likewise.
31444 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
31445 (known_aggs_to_agg_replacement_list): Likewise.
31446 * ipa-inline-analysis.c: Likewise.
31447 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
31448 * ipa-polymorphic-call.c
31449 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
31450 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
31451 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
31453 * modulo-sched.c (apply_reg_moves): Likewise.
31454 * omp-expand.c (build_omp_regions_1): Likewise.
31455 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
31456 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
31457 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
31458 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
31459 * value-prof.c: Likewise.
31460 * var-tracking.c (val_reset): Likewise.
31462 2017-04-03 Richard Biener <rguenther@suse.de>
31464 PR tree-optimization/80275
31465 * fold-const.c (split_address_to_core_and_offset): Handle
31468 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
31470 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
31471 descriptors is at least equal to that of functions.
31473 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
31475 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
31477 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
31480 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
31481 (mov<IMOD4:mode>): New expander.
31482 (*mov<IMOD4:mode>_internal): New insn and split pattern.
31484 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
31486 PR rtl-optimization/79405
31487 * fwprop.c (propagations_left): New variable.
31488 (forward_propagate_into): Decrement it.
31489 (fwprop_init): Initialize it.
31490 (fw_prop): If the variable has reached zero, stop propagating.
31491 (fwprop_addr): Ditto.
31493 2017-03-31 Jakub Jelinek <jakub@redhat.com>
31496 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
31497 a FUNCTION_DECL, pass it as decl instead of origin to
31500 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
31502 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
31505 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
31508 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
31509 TARGET_VSX_SMALL_INTEGER.
31511 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31513 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
31514 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
31516 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31518 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
31519 extraction from odd-numbered MSA register.
31521 2017-03-31 Jakub Jelinek <jakub@redhat.com>
31523 PR middle-end/80173
31524 * expmed.c (store_bit_field_1): Don't attempt to create
31525 a word subreg out of hard registers wider than word if they
31526 have HARD_REGNO_NREGS of 1 for their mode.
31528 PR middle-end/80163
31529 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
31530 conversions to integer types wider than word and pointer.
31533 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
31534 (rtx_equal_for_cselib_p): Pass 0 to it.
31535 * cselib.c (cselib_hasher::equal): Likewise.
31536 (rtx_equal_for_cselib_1): Add depth argument. If depth
31537 is 128, don't look up VALUE locs and punt. Increment
31538 depth in recursive calls when walking VALUE locs.
31540 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
31542 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
31543 (make_gcov_file_name): Use the canonical path name for generating
31545 (read_line): Fix handling of files with ascii null bytes.
31547 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
31549 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
31550 to initialise a vector register instead
31551 of using a const_int.
31553 2017-03-30 Jakub Jelinek <jakub@redhat.com>
31555 PR translation/80189
31556 * gimplify.c (omp_default_clause): Use %qs instead of %s in
31557 diagnostic messages.
31559 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
31562 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
31563 (dfp_diex_<mode>): Update mode of operand 1.
31564 * doc/extend.texi (dxex, dxexq): Document change to return type.
31565 (diex, diexq): Document change to argument type.
31567 2017-03-30 Martin Jambor <mjambor@suse.cz>
31570 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
31571 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
31572 it reflects the signature changes performed at the callee side.
31573 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
31574 to cgraph_build_function_type_skip_args.
31575 (build_function_decl_skip_args): Adjust call to the above function.
31577 2017-03-30 Jakub Jelinek <jakub@redhat.com>
31580 * config/i386/sse.md
31581 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
31582 register as dest whenever it is a MEM not rtx_equal_p to the
31583 corresponding dup operand, and when forcing into reg move the
31584 reg into the memory afterwards.
31585 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
31586 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
31587 for the force_reg mode.
31588 (avx512vl_vextractf128<mode>): Use register as dest either
31589 always when a MEM, or when it is a MEM not rtx_equal_p to the
31590 corresponding dup operand, or even not when it is a CONST_VECTOR
31591 depending on the mode and lo vs. hi.
31592 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
31594 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
31595 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
31596 Likewise. Require that operands[2] is even.
31597 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
31598 Remove extraneous parens. Require that operands[2] is a multiple
31600 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
31601 operands[0] is a MEM if <mask_applied>, the predicates/constraints
31602 disallow memory then.
31604 2017-03-30 Richard Biener <rguenther@suse.de>
31606 PR tree-optimization/77498
31607 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
31608 to non-constants over backedges.
31610 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
31612 PR rtl-optimization/80233
31613 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
31614 as last_combined_insn. Do not test for BARRIER_P separately.
31616 2017-03-29 Andreas Schwab <schwab@suse.de>
31619 * calls.c (prepare_call_address): Convert funexp to Pmode before
31620 copying to temp reg.
31622 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31624 PR tree-optimization/80158
31625 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
31626 Handle possible future case of more than one alternate
31628 (replace_rhs_if_not_dup): Likewise.
31629 (replace_one_candidate): Likewise.
31631 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
31633 PR rtl-optimization/80193
31634 * ira.c (ira): Do not check allocation for LRA.
31636 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
31638 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
31639 (nvptx_output_simt_exit): Declare.
31640 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
31641 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
31642 (init_softstack_frame): Move initialization of crtl->is_leaf to...
31643 (nvptx_declare_function_name): ...here. Emit declaration of local
31644 memory space buffer for omp_simt_enter insn.
31645 (nvptx_output_unisimt_switch): New.
31646 (nvptx_output_softstack_switch): New.
31647 (nvptx_output_simt_enter): New.
31648 (nvptx_output_simt_exit): New.
31649 * config/nvptx/nvptx.h (struct machine_function): New fields
31650 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
31651 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
31652 (UNSPECV_SIMT_EXIT): Ditto.
31653 (omp_simt_enter_insn): New insn.
31654 (omp_simt_enter): New expansion.
31655 (omp_simt_exit): New insn.
31656 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
31658 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
31659 (expand_GOMP_SIMT_ENTER_ALLOC): New.
31660 (expand_GOMP_SIMT_EXIT): New.
31661 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
31662 (GOMP_SIMT_ENTER_ALLOC): Ditto.
31663 (GOMP_SIMT_EXIT): Ditto.
31664 * target-insns.def (omp_simt_enter): New insn.
31665 (omp_simt_exit): Ditto.
31666 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
31668 (lower_rec_simd_input_clauses): Implement SIMT privatization.
31669 (lower_rec_input_clauses): Likewise.
31670 (lower_lastprivate_clauses): Handle SIMT privatization.
31672 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
31673 (ompdevlow_adjust_simt_enter): New.
31674 (find_simtpriv_var_op): New.
31675 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
31676 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
31678 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
31679 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
31680 (copy_decl_for_dup_finish): Ditto.
31682 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
31684 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
31687 * config/i386/i386.c (ix86_option_override_internal): Always
31688 allow -mpreferred-stack-boundary=3 for 64-bit targets.
31690 2017-03-28 Bin Cheng <bin.cheng@arm.com>
31692 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
31694 2017-03-28 Bin Cheng <bin.cheng@arm.com>
31696 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
31697 mark new edge's irreducible flag accordign to it.
31698 (vect_do_peeling): Check loop preheader edge's irreducible flag
31699 and pass it to function slpeel_add_loop_guard.
31701 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
31703 PR tree-optimization/80218
31704 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
31705 Update block frequencies and counts.
31707 2017-03-28 Richard Biener <rguenther@suse.de>
31709 PR tree-optimization/78644
31710 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
31711 of a simplification result we may not use it at all.
31713 2017-03-28 Richard Biener <rguenther@suse.de>
31716 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
31717 without arguments, generate default definition of a SSA name.
31719 2017-03-28 Richard Biener <rguenther@suse.de>
31721 PR middle-end/80222
31722 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
31723 TYPE_REF_CAN_ALIAS_ALL references.
31724 * fold-const.c (fold_indirect_ref_1): Likewise.
31726 2017-03-28 Martin Liska <mliska@suse.cz>
31729 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
31730 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
31732 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31733 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
31735 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
31736 (EXTRA_SPECS): Define.
31737 (SUBTARGET_EXTRA_SPECS): Likewise.
31738 (SUBTARGET_CPP_SPEC): Likewise.
31739 * config/arc/elf.h (EXTRA_SPECS): Renamed to
31740 SUBTARGET_EXTRA_SPECS.
31741 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
31743 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31745 * config/arc/simdext.md (vst64_insn): Update pattern.
31746 (vld32wh_insn): Likewise.
31747 (vld32wl_insn): Likewise.
31748 (vld64_insn): Likewise.
31749 (vld32_insn): Likewise.
31751 2017-03-28 Marek Polacek <polacek@redhat.com>
31754 * fold-const.c (fold_comparison): Use protected_set_expr_location
31755 instead of SET_EXPR_LOCATION.
31757 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
31759 * tree.c (add_expr): Avoid name lookup warning.
31761 2017-03-27 Jeff Law <law@redhat.com>
31763 PR tree-optimization/80216
31764 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
31765 function name. Limit recursion depth.
31766 (record_temporary_equivalences): Corresponding changes.
31768 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
31770 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
31773 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31776 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
31778 * cfgcleanup.c (reg_note_cfa_p): New array.
31779 (insns_have_identical_cfa_notes): New function.
31780 (old_insns_match_p): Don't cross-jump in between /f
31781 and non-/f instructions. If both i1 and i2 are frame related,
31782 verify all CFA notes, their order and content.
31784 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
31787 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
31788 HImode and SImode with zero extend to DImode to one insn.
31789 (bswap<mode>2_extenddi): Likewise.
31790 (bswapsi2_extenddi): Likewise.
31791 (bswaphi2_extendsi): Likewise.
31792 (bswaphi2): Combine bswap HImode and SImode into one insn.
31793 Separate memory insns from swapping register.
31794 (bswapsi2): Likewise.
31795 (bswap<mode>2): Likewise.
31796 (bswaphi2_internal): Delete, no longer used.
31797 (bswapsi2_internal): Likewise.
31798 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
31799 store, and gpr<-gpr swap insns.
31800 (bswap<mode>2_store): Likewise.
31801 (bswaphi2_reg): Register only splitter, combine with the splitter.
31802 (bswaphi2 splitter): Likewise.
31803 (bswapsi2_reg): Likewise.
31804 (bswapsi2 splitter): Likewise.
31805 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
31806 the insns into load, store, and register/register insns.
31807 (bswapdi2_ldbrx): Likewise.
31808 (bswapdi2_load): Likewise.
31809 (bswapdi2_store): Likewise.
31810 (bswapdi2_reg): Likewise.
31812 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
31814 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
31815 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
31817 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
31820 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
31822 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31823 special handling for target option conflicts between dform
31824 options (-mpower9-dform, -mpower9-dform-vector,
31825 -mpower9-dform-scalar) and -mno-direct-move.
31827 2017-03-27 Richard Biener <rguenther@suse.de>
31829 PR tree-optimization/80181
31830 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
31832 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31834 * config/arc/predicates.md (move_double_src_operand): Replace the
31835 call to move_double_src_operand with a call to address_operand.
31837 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31839 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
31840 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
31841 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
31843 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31845 * config/arc/predicates.md (long_immediate_loadstore_operand):
31846 Consider scaled addresses cases.
31848 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31850 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
31851 restored when in interrupt.
31852 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
31853 doesn't have delay slot.
31855 2017-03-27 Richard Biener <rguenther@suse.de>
31858 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
31859 inlined thunk clones.
31861 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31864 * asan.c (instrument_derefs): Copy over last operand from
31865 original COMPONENT_REF to the new COMPONENT_REF with
31866 DECL_BIT_FIELD_REPRESENTATIVE.
31867 * ubsan.c (instrument_object_size): Likewise.
31869 2017-03-27 Richard Biener <rguenther@suse.de>
31871 PR tree-optimization/80170
31872 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
31873 sure DR/SCEV didnt fold in constants we do not see when looking
31874 at the reference base alignment.
31876 2017-03-27 Richard Biener <rguenther@suse.de>
31878 PR middle-end/80171
31879 * gimple-fold.c (fold_ctor_reference): Properly guard against
31880 NULL return value from canonicalize_constructor_val.
31882 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
31885 * config/i386/i386.c (ix86_expand_builtin)
31886 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
31887 flags reg setting and flags reg using instructions.
31888 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
31889 clobbering instructions to zero extend op2.
31891 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
31893 * doc/install.texi (Configuration) <--with-aix-soname>:
31894 Update link to AIX ld.
31896 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
31898 PR rtl-optimization/80160
31899 PR rtl-optimization/80159
31900 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
31901 reg_alternate_class into account.
31903 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
31906 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
31907 to consider in curr_insn_transform.
31909 2017-03-24 Jakub Jelinek <jakub@redhat.com>
31911 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
31912 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
31913 and emit_mode_inner.
31915 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31917 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
31918 argument to the overloaded builtin variants. Use the new flag to
31919 deprecate certain builtin variants.
31920 * config/s390/s390-builtin-types.def: Add new builtin types.
31921 * config/s390/s390-builtins.h: Support new flags field for
31922 overloaded builtins.
31923 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
31924 (s390_macro_to_expand): Enable vector float data type.
31925 (s390_cpu_cpp_builtins_internal): Indicate support of the new
31926 builtins by incrementing the __VEC__ version number.
31927 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
31929 (s390_resolve_overloaded_builtin): Emit error messages depending
31930 on the builtin flags.
31931 * config/s390/s390.c (s390_expand_builtin): Support additional
31932 flags argument. Change error message to match the messages
31933 emitted in s390-c.c.
31934 * config/s390/s390.md: New UNSPEC_* constants.
31935 (op_type): Add new instruction types.
31936 * config/s390/vecintrin.h: Add new builtins and test data class
31938 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
31939 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
31940 (VEC_INEXACT, VEC_NOINEXACT): New constants.
31941 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
31942 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
31943 ("vec_mergel<mode>"): V_HW -> VEC_HW.
31945 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
31946 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
31947 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
31948 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
31950 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
31951 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
31952 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
31953 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
31955 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
31956 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
31957 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
31958 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
31959 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
31960 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
31961 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
31963 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
31964 ("vec_scatter_element<V_HW_4:mode>_DI")
31965 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
31966 ("vec_fpint<mode>", "vflls")
31967 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
31968 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
31969 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
31970 ("*vec_cmphe<mode>_cc"): ... these.
31972 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
31973 mode constant instead of magic value.
31975 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31977 * config/s390/s390.c (s390_expand_vec_compare): Support other
31978 vector floating point modes than just V2DF.
31979 (s390_expand_vcond): Likewise.
31980 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
31981 (s390_cannot_change_mode_class): Prevent mode changes between TF
31982 and V1TF in vector registers.
31983 * config/s390/s390.md (DF, SF): New mode attributes.
31984 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
31985 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
31986 SFmode support for VRs.
31987 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
31989 (VFT, VF_HW): New mode iterators.
31990 (vw, sdx): New mode attributes.
31991 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
31992 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
31993 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
31994 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
31995 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
31996 also the new vector floating point modes. Renaming to ...
31998 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
31999 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
32000 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
32001 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
32002 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
32003 ("vec_unordered<mode>"): ... these.
32005 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
32006 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
32007 ("*vec_extendv2df"): New insn definitions.
32009 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32011 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
32012 ("mulditi3_2", "*muldi3_sign"): New patterns.
32013 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
32014 rename the pattern definition.
32016 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32018 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
32020 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
32022 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32024 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
32025 instruction if possible.
32026 * config/s390/vector.md (vec_halfnumelts): New mode
32028 ("*vec_vllezlf<mode>"): New pattern.
32030 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32032 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
32033 ("popcountv4si2", "popcountv2di2"): Rename to ...
32034 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
32035 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
32037 ("popcount<mode>2_vxe"): New pattern.
32039 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32041 * common/config/s390/s390-common.c (processor_flags_table): Add
32043 * config.gcc: Add arch12.
32044 * config/s390/driver-native.c (s390_host_detect_local_cpu):
32045 Default to arch12 for unknown CPU model numbers.
32046 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
32047 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
32048 PROCESSOR_max sanity check.
32049 * config/s390/s390-opts.h (enum processor_type): Add
32051 * config/s390/s390.c (processor_table): Add arch12.
32052 (s390_expand_builtin): Add check for B_VXE flag.
32053 (s390_issue_rate): Add PROCESSOR_ARCH12.
32054 (s390_get_sched_attrmask): Likewise.
32055 (s390_get_unit_mask): Likewise.
32056 (s390_sched_score): Enable z13 scheduling for arch12.
32057 (s390_sched_reorder): Likewise.
32058 (s390_sched_variable_issue): Likewise.
32059 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
32061 (s390_tune_attr): Use z13 scheduling also for arch12.
32062 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
32063 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
32064 (TARGET_VXE_P): New macros.
32065 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
32066 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
32067 * config/s390/s390.opt: Add arch12 as processor_type.
32069 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32071 * config/s390/s390.md
32072 ("fixuns_truncdddi2", "fixuns_trunctddi2")
32073 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
32074 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
32076 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
32077 Rename expanders to ...
32079 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
32080 ("fixuns_truncdddi2_emu"): ... these.
32082 ("fixuns_trunc<mode>si2_emu"): New expander.
32084 ("*fixuns_truncdfdi2_z13"): Rename to ...
32085 ("*fixuns_truncdfdi2_vx"): ... this.
32087 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32089 * config/s390/2964.md: Remove the single element vector compare
32090 instructions which are no longer used.
32091 * config/s390/s390.c (s390_select_ccmode): Remove handling of
32093 (s390_canonicalize_comparison): Remove handling of DFmode
32095 (s390_expand_vec_compare_scalar): Remove function.
32096 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
32097 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
32099 ("*cmp<mode>_ccs"): Add wfcdb instruction.
32101 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32103 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
32105 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
32106 will anyway by matched by mov<mode>_64dfp.
32108 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32110 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
32111 vlef/vstef. Add missing operand to vleif.
32113 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32115 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
32116 pair for all vector types with 64 bit elements.
32117 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
32118 * config/s390/vector.md (V_HW_64): ... here.
32119 (V_128_NOSINGLE): New mode iterator.
32120 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
32121 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
32122 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
32123 ("*vec_load_pairv2di"): Change to ...
32124 ("*vec_load_pair<mode>"): ... this one.
32126 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32128 * config/s390/constraints.md: Add comments.
32129 (jKK): Reject element sizes > 8 bytes.
32130 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
32132 * config/s390/s390.md: Add the s_operand checks formerly in
32133 s390_split_ok_p to various splitters where they are still
32135 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
32136 for 128 bit vectors. Plus two splitters.
32138 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32140 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
32143 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32146 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
32147 error if the boundary argument is not constant.
32149 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32151 PR rtl-optimization/80112
32152 * loop-doloop.c (doloop_condition_get): Don't check condition
32153 if cmp isn't SET with IF_THEN_ELSE src.
32155 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32157 PR tree-optimization/80158
32158 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
32159 replacing a candidate statement, also replace it for the
32160 candidate's alternate interpretation.
32161 (replace_rhs_if_not_dup): Likewise.
32162 (replace_one_candidate): Likewise.
32164 2017-03-24 Richard Biener <rguenther@suse.de>
32166 PR tree-optimization/80167
32167 * graphite-isl-ast-to-gimple.c
32168 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
32170 (translate_isl_ast_to_gimple::get_rename): Likewise.
32172 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32174 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
32175 handling of certain combinations of target options, including the
32176 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
32177 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
32179 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32182 * config/arm/arm.md (*load_multiple): Add reload_completed to
32183 matching condition.
32185 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32186 Richard Biener <rguenth@suse.de>
32188 PR tree-optimization/79908
32189 PR tree-optimization/80136
32190 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32191 been cast away, gimplify_and_add suffices.
32193 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
32195 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
32197 2017-03-23 Richard Biener <rguenther@suse.de>
32199 PR tree-optimization/80032
32200 * gimplify.c (gimple_push_cleanup): Forced unconditional
32201 cleanups still have to go to the conditional_cleanups
32204 2017-03-22 Jakub Jelinek <jakub@redhat.com>
32206 PR tree-optimization/80072
32207 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
32209 (next_operand_entry_id): Change type to unsigned int.
32210 (sort_by_operand_rank): Make sure to return the right return value
32211 even if unsigned fields are bigger than INT_MAX.
32212 (struct oecount): Change cnt and id type to unsigned int.
32213 (oecount_hasher::equal): Formatting fix.
32214 (oecount_cmp): Make sure to return the right return value
32215 even if unsigned fields are bigger than INT_MAX.
32216 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
32219 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
32220 TREE_READONLY on result if writing it more than once.
32223 * doc/invoke.texi (-fsanitize=thread): Document that with
32224 -fnon-call-exceptions atomics are not able to throw
32228 * tsan.c: Include tree-eh.h.
32229 (instrument_builtin_call): Call maybe_clean_eh_stmt or
32230 maybe_clean_or_replace_eh_stmt where needed.
32231 (instrument_memory_accesses): Add cfg_changed argument.
32232 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
32233 if it returned true.
32234 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
32236 PR rtl-optimization/63191
32237 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
32238 wrapper function, moved the whole old content into ...
32239 (ix86_delegitimize_address_1): ... this. New inline function.
32240 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
32241 true as last argument instead of ix86_delegitimize_address.
32243 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32245 * config/aarch64/aarch64.c (generic_branch_cost): Copy
32246 cortexa57_branch_cost.
32248 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32250 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
32252 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32255 * doc/md.texi (Constraints): Document wA constraint.
32256 * config/rs6000/constraints.md (wA): New.
32257 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
32258 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
32259 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
32260 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
32262 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
32265 * gimplify.c (is_oacc_declared): New function.
32266 (oacc_default_clause): Use it to set default flags for acc declared
32267 variables inside parallel regions.
32268 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
32269 declared variables.
32270 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
32271 declare attribute to any decl as necessary.
32273 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
32276 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
32277 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
32278 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
32279 (arm_arch_lpae): This.
32280 * config/arm/arm.c (arm_arch7ve): Rename into ...
32281 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
32282 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
32285 2017-03-22 Martin Liska <mliska@suse.cz>
32288 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
32289 error message instead of an ICE.
32291 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32293 * doc/extend.texi (6.11 Additional Floating Types): Revise.
32295 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
32297 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
32299 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32302 2017-03-21 Martin Sebor <msebor@redhat.com>
32304 * doc/extend.texi: Use "cannot" instead of "can't."
32305 * doc/hostconfig.texi: Same.
32306 * doc/install.texi: Same.
32307 * doc/invoke.texi: Same.
32308 * doc/loop.texi: Same.
32309 * doc/md.texi: Same.
32310 * doc/objc.texi: Same.
32311 * doc/rtl.texi: Same.
32312 * doc/tm.texi: Same.
32313 * doc/tm.texi.in: Same.
32314 * doc/trouble.texi: Same.
32316 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
32319 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
32320 (collect_checksum_attributes): Set it.
32321 (die_checksum_ordered): Use it.
32323 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32325 PR tree-optimization/79908
32326 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
32327 change: For a VA_ARG whose LHS has been cast away, use
32328 force_gimple_operand to construct the side effects.
32330 2017-03-21 David Malcolm <dmalcolm@redhat.com>
32332 PR translation/80001
32333 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
32334 more amenable to translation.
32335 (oacc_loop_auto_partitions): Likewise.
32337 2017-03-21 Marek Polacek <polacek@redhat.com>
32338 Martin Sebor <msebor@redhat.com>
32340 PR tree-optimization/80109
32341 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
32342 on INTEGRAL_TYPE_P.
32344 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32345 Segher Boessenkool <segher@kernel.crashing.org>
32348 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
32349 check reg_used_between_p between insn and one of succ or succ2
32350 depending on if succ is artificial insn not inserted into insn
32353 2017-03-21 Martin Liska <mliska@suse.cz>
32355 PR gcov-profile/80081
32356 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
32357 * doc/gcc.texi: Include gcov-dump stuff.
32358 * doc/gcov-dump.texi: New file.
32360 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
32362 PR rtl-optimization/79150
32363 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
32364 conditional jump, if the jump is the last insn of the loop.
32366 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32367 Richard Biener <rguenth@suse.de>
32369 PR tree-optimization/79908
32370 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32371 been cast away, use force_gimple_operand to construct the side
32374 2017-03-21 Martin Liska <mliska@suse.cz>
32376 PR libfortran/79956
32377 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
32380 2017-03-21 Brad Spengler <spender@grsecurity.net>
32383 * plugin.c (htab_hash_plugin): New function.
32384 (add_new_plugin): Use it and adjust.
32385 (parse_plugin_arg_opt): Adjust.
32386 (init_one_plugin): Likewise.
32388 2017-03-21 Richard Biener <rguenther@suse.de>
32390 PR tree-optimization/80032
32391 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
32392 if set force the cleanup to happen unconditionally.
32393 (gimplify_target_expr): Push inserted clobbers with force_uncond
32394 to avoid them being removed by control-dependent DCE.
32396 2017-03-21 Richard Biener <rguenther@suse.de>
32398 PR tree-optimization/80122
32399 * tree-inline.c (copy_bb): Do not expans va-arg packs or
32400 va_arg_pack_len when the inlined call stmt requires pack
32402 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
32404 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32407 * tsan.c (instrument_builtin_call): If the memory model argument
32408 is not a constant, assume it is valid.
32411 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
32414 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
32416 PR rtl-optimization/79910
32417 * combine.c (can_combine_p): Do not allow combining an I0 or I1
32418 if its dest is used by an insn before I2 (other than the combined
32419 insns themselves, which are properly handled already).
32421 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
32424 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
32426 * combine.c (record_used_regs): New static function.
32427 (try_combine): Handle situations where there is an additional
32428 instruction between I2 and I3 which needs to have a LOG_LINK
32432 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
32434 * combine.c (try_combine): Delete redundant i1 test. Call
32435 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
32437 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32440 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
32441 alternatives 13/14.
32443 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32445 PR tree-optimization/80054
32446 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
32447 the optimization if a PHI or any of its arguments is not dominated
32448 by the candidate's basis. Use gphi* rather than gimple* as
32450 (replace_profitable_candidates): Clean up a gimple* variable that
32451 should be a gphi* variable.
32453 2017-03-20 Martin Sebor <msebor@redhat.com>
32456 * doc/extend.texi (attribute constructor): Document present limitation.
32458 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
32461 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
32462 __POWER9_VECTOR__ #ifdef control, change template definition to
32463 use Power9-specific built-in function.
32464 (vec_any_eq): Likewise.
32465 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
32466 to control outcomes from this test.
32467 (vector_ae_<mode>p): For VEC_F modes, likewise.
32469 2017-03-20 Ian Lance Taylor <iant@google.com>
32471 * config/i386/i386.c (ix86_function_regparm): Save an extra
32472 register for -fsplit-stack with DECL_STATIC_CHAIN.
32474 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
32477 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
32478 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
32480 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
32482 * config/riscv/riscv.c (riscv_print_operand): Use "fence
32484 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
32487 2017-03-20 Marek Polacek <polacek@redhat.com>
32490 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
32492 2017-03-20 Richard Biener <rguenther@suse.de>
32494 PR tree-optimization/80113
32495 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
32496 allocate extra SSA name for PHI def.
32497 (add_close_phis_to_outer_loops): Likewise.
32498 (add_close_phis_to_merge_points): Likewise.
32499 (copy_loop_close_phi_args): Likewise.
32500 (copy_cond_phi_nodes): Likewise.
32502 2017-03-20 Martin Liska <mliska@suse.cz>
32504 PR middle-end/79753
32505 * tree-chkp.c (chkp_build_returned_bound): Do not build
32506 returned bounds for a LHS that's not a BOUNDED_P type.
32508 2017-03-20 Martin Liska <mliska@suse.cz>
32512 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
32513 COMPLEX_CST and VECTOR_CST.
32515 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32518 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
32519 target operand. A new splitter adds the clobber statement in case
32520 the target operand is dead anyway.
32522 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
32524 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
32525 to age-old versions of binutils and glibc.
32527 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
32529 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
32531 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
32533 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
32535 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
32537 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
32538 requirement for binutils 2.13.
32540 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
32542 * combine.c (try_combine): Delete redundant i1 test. Call
32543 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
32545 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
32547 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
32548 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
32550 <riscv64-*-elf>: Re-arrange section
32551 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
32552 <riscv32-*-linux>: Likewise.
32553 <riscv64-*-elf>: Likewise
32554 <riscv64-*-linux>: Likewise.
32556 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
32559 * aarch64.opt(verbose-cost-dump): Fix typo.
32561 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
32564 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
32565 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
32567 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
32569 * reload.c (find_reloads): When reloading a nonoffsettable address,
32570 use RELOAD_OTHER for it and its address reloads.
32572 PR rtl-optimization/79910
32573 * combine.c (record_used_regs): New static function.
32574 (try_combine): Handle situations where there is an additional
32575 instruction between I2 and I3 which needs to have a LOG_LINK
32578 2017-03-17 Jeff Law <law@redhat.com>
32580 PR tree-optimization/71437
32581 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
32582 conditional in the hash table first.
32583 (vrp_dom_walker::before_dom_children): Extract condition from
32584 ASSERT_EXPR. Record condition, its inverion and any implied
32585 conditions as well.
32587 2017-03-17 Marek Polacek <polacek@redhat.com>
32588 Markus Trippelsdorf <markus@trippelsdorf.de>
32590 PR tree-optimization/80079
32591 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
32594 2017-03-17 Richard Biener <rguenther@suse.de>
32596 PR middle-end/80075
32597 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
32598 Properly verify the LHS before the RHS possibly claims to be
32600 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
32603 2017-03-17 Martin Jambor <mjambor@suse.cz>
32605 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
32606 (List of -O2 options): Likewise.
32607 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
32610 2017-03-17 Tom de Vries <tom@codesourcery.com>
32612 * gcov-dump.c (print_usage): Print bug_report_url.
32614 2017-03-17 Richard Biener <rguenther@suse.de>
32616 PR middle-end/80050
32617 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
32618 (parser::peek): Likewise.
32620 2017-03-17 Richard Biener <rguenther@suse.de>
32622 PR tree-optimization/80048
32623 * sese.c (free_sese_info): Properly release rename_map and
32624 copied_bb_map elements.
32626 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
32628 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
32629 Add linked-list forward and backlinks. Insert on
32630 construction, remove on destruction.
32631 (class pass_store_merging): Add m_stores_head field.
32632 (pass_store_merging::terminate_and_process_all_chains):
32633 Iterate over m_stores_head list.
32634 (pass_store_merging::terminate_all_aliasing_chains):
32636 (pass_store_merging::execute): Check for debug stmts first.
32637 Push new chains onto the m_stores_head stack.
32639 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
32642 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
32643 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
32644 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
32646 2017-03-16 Jeff Law <law@redhat.com>
32648 PR tree-optimization/71437
32649 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
32650 member function. Implementation moved into after_dom_children
32651 member function and into the threader's thread_outgoing_edges
32653 (dom_opt_dom_walker::after_dom_children): Simplify by moving
32654 some code into new thread_outgoing_edges.
32655 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
32656 definition. Simplify marker handling (do it here). Assume we always
32657 have the available expression and the const/copies tables.
32658 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
32660 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
32661 * tree-vrp.c (equiv_stack): No longer file scoped.
32662 (vrp_dom_walker): New class.
32663 (vrp_dom_walker::before_dom_children): New member function.
32664 (vrp_dom_walker::after_dom_children): Likewise.
32665 (identify_jump_threads): Setup domwalker. Use it rather than
32666 walking edges in a random order by hand. Simplify setup/finalization.
32667 (finalize_jump_threads): Remove.
32668 (vrp_finalize): Do not call identify_jump_threads here.
32669 (execute_vrp): Do it here instead and call thread_through_all_blocks
32672 PR tree-optimization/71437
32673 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
32675 (simplify_stmt_for_jump_threading): Add basic_block argument. All
32677 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
32678 (dom_opt_dom_walker::thread_across_edge): Remove
32679 handle_dominating_asserts argument. All callers changed.
32680 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
32681 changes. Remove calls to lhs_of_dominating_assert. Other
32682 uses of handle_dominating_asserts turn into unconditional code
32683 (simplify_control_stmt_condition_1): Likewise.
32684 (simplify_control_stmt_condition): Likewise.
32685 (thread_through_normal_block, thread_across_edge): Likewise.
32686 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
32687 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
32688 object if it is not an SSA_NAME.
32689 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
32690 before calling into the VRP specific simplifiers.
32691 (identify_jump_threads): Remove handle_dominating_asserts
32694 2017-03-16 Jakub Jelinek <jakub@redhat.com>
32697 * tree-diagnostic.c (default_tree_printer): No longer static.
32698 * tree-diagnostic.h (default_tree_printer): New prototype.
32700 2017-03-16 Tamar Christina <tamar.christina@arm.com>
32702 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
32703 Change ins into fmov.
32705 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32707 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
32708 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
32709 Use h_con constraint for operand 1.
32710 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
32711 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
32713 2017-03-15 Jeff Law <law@redhat.com>
32715 PR tree-optimization/71437
32716 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
32717 (record_temporary_equivalences): Use it.
32719 PR tree-optimization/71437
32720 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
32721 tree-ssa-scopedtables.
32722 (lookup_avail_expr, build_and_record_new_cond): Likewise.
32723 (record_conditions, record_cond, vuse_eq): Likewise.
32724 (record_edge_info): Adjust to API tweak of record_conditions.
32725 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
32726 (record_temporary_equivalences, optimize_stmt): Likewise.
32727 (eliminate_redundant_computations): Likewise.
32728 (record_equivalences_from_stmt): Likewise.
32729 * tree-ssa-scopedtables.c: Include options.h and params.h.
32730 (vuse_eq): New function, moved from tree-ssa-dom.c
32731 (build_and_record_new_cond): Likewise.
32732 (record_conditions): Likewise. Accept vector of conditions rather
32733 than edge_equivalence structure for first argument.
32734 for the first argument.
32735 (avail_exprs_stack::lookup_avail_expr): New member function, moved
32736 from tree-ssa-dom.c.
32737 (avail_exprs_stack::record_cond): Likewise.
32738 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
32739 from tree-ssa-dom.c.
32740 (avail_exprs_stack): Add new member functions lookup_avail_expr
32742 (record_conditions): Declare.
32744 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
32747 * lra-constraints.c (process_alt_operands): Increase reject for
32748 reloading an input/output operand.
32750 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
32753 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
32754 insns to convert from signed/unsigned char/short to IEEE 128-bit
32756 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
32758 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
32761 * config/i386/i386.c (ix86_vector_duplicate_value): Create
32762 subreg of inner mode for values already in registers.
32764 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
32766 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
32767 iteration reg is used after the loop.
32769 2017-03-14 Martin Sebor <msebor@redhat.com>
32771 PR tree-optimization/79800
32772 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
32773 precision in negative-positive range.
32774 (format_floating): Call non-const overload with adjusted precision.
32776 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
32779 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
32782 2017-03-14 Martin Sebor <msebor@redhat.com>
32784 PR middle-end/80020
32785 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
32786 * builtins.def (aligned_alloc): Use it.
32789 * Makefile.in (GTFILES): Add calls.c.
32790 * calls.c: Include "gt-calls.h".
32792 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
32794 PR rtl-optimization/79728
32795 * regs.h (struct target_regs): New field
32796 x_contains_allocatable_regs_of_mode.
32797 (contains_allocatable_regs_of_mode): New macro.
32798 * reginfo.c (init_reg_sets_1): Initialize it, and change
32799 contains_reg_of_mode so it includes global regs as well.
32800 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
32801 rather than contains_regs_of_mode.
32803 2017-03-14 Martin Liska <mliska@suse.cz>
32805 * doc/invoke.texi: Document options that can't be combined with
32806 -fcheck-pointer-bounds.
32808 2017-03-14 Martin Liska <mliska@suse.cz>
32810 PR middle-end/79831
32811 * doc/invoke.texi (-Wchkp): Document the option.
32813 2017-03-14 Martin Liska <mliska@suse.cz>
32815 * Makefile.in: Install gcov-dump.
32817 2017-03-14 Martin Liska <mliska@suse.cz>
32819 * multiple_target.c (expand_target_clones): Bail out for
32820 an invalid attribute.
32822 2017-03-14 Richard Biener <rguenther@suse.de>
32824 * alias.c (struct alias_set_entry): Pack properly.
32825 * cfgloop.h (struct loop): Likewise.
32826 * cse.c (struct set): Likewise.
32827 * ipa-utils.c (struct searchc_env): Likewise.
32828 * loop-invariant.c (struct invariant): Likewise.
32829 * lra-remat.c (struct cand): Likewise.
32830 * recog.c (struct change_t): Likewise.
32831 * rtl.h (struct address_info): Likewise.
32832 * symbol-summary.h (function_summary): Likewise.
32833 * tree-loop-distribution.c (struct partition): Likewise.
32834 * tree-object-size.c (struct object_size_info): Likewise.
32835 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
32836 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
32837 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
32838 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
32839 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
32840 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
32841 (struct _stmt_vec_info): Likewise.
32843 2017-03-14 Martin Liska <mliska@suse.cz>
32846 * multiple_target.c (create_dispatcher_calls): Check that
32847 a target can create a function dispatcher.
32849 2017-03-14 Martin Liska <mliska@suse.cz>
32852 * multiple_target.c (expand_target_clones): Drop local.local
32853 flag for default implementation.
32855 2017-03-14 Richard Biener <rguenther@suse.de>
32857 PR tree-optimization/80030
32858 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
32860 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
32862 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
32863 gcc_fallthrough() instead of __attribute__((fallthrough));
32865 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
32867 * doc/gcc.texi: Remove "up" link to (DIR).
32868 * doc/gccint.texi: Ditto.
32870 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
32872 * doc/install.texi (Specific) <avr>: Remove reference to
32875 2017-03-13 Jeff Law <law@redhat.com>
32877 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
32878 attribute rather than comments.
32880 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
32881 match_scratch operand is highest.
32883 2017-03-13 Martin Liska <mliska@suse.cz>
32885 PR middle-end/78339
32886 * ipa-pure-const.c (warn_function_noreturn): If the declarations
32887 is a CHKP clone, use original declaration.
32889 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32891 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
32892 (arc_conditional_register_usage): Use a different allocation order
32893 when optimizing for size.
32894 * common/config/arc/arc-common.c (arc_option_optimization_table):
32895 Section anchors default on when optimizing for size.
32897 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32899 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
32901 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32903 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
32904 * config/arc/arc.md (cpu_facility): Add cd variant.
32905 (*movqi_insn): Add code density variant.
32906 (*movhi_insn): Likewise.
32907 (*movqi_insn): Likewise.
32908 (*addsi3_mixed): Likewise.
32909 (subsi3_insn): Likewise.
32911 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32913 * config/arc/arc.md (movsi_cond_exec): Update constraint.
32915 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32917 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
32918 expressions with MINUS and UNARY ops.
32920 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32923 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
32925 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
32926 between vec_select and vector argument.
32927 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
32928 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
32929 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
32930 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
32931 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
32932 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
32934 2017-03-13 Richard Biener <rguenther@suse.de>
32937 * params.def (vect-max-peeling-for-alignment): Fix typo.
32939 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32941 * doc/install.texi (Specific) <mips-*-*>: Remove description of
32942 issue that only occurred with binutils below 2.18.
32944 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32946 * doc/install.texi (Specific) <cris-axis-elf>: No longer
32947 refer to binutils 2.11/2.12 minimum.
32949 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32951 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
32952 ftp.kernel.org and simplify binutils requirement.
32954 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
32956 * doc/invoke.texi (Warning Options): Fix spelling of link-time
32958 (Optimize Options): Ditto. Also remove redundancy.
32960 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32962 PR translation/79848
32963 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
32965 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
32966 to G_ to avoid double translation.
32968 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32970 PR translation/79923
32971 * auto-profile.c (get_combined_location): Convert leading
32972 character of diagnostics to lower case and remove trailing period.
32973 (read_profile): Likewise for various diagnostics.
32974 * config/arm/arm.c (arm_option_override): Remove trailing period
32975 from various diagnostics.
32976 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
32977 (msp430_expand_delay_cycles): Likewise.
32979 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32982 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
32983 full command-line argument, rather than just "str".
32984 (aarch64_validate_march): Likewise.
32985 (aarch64_validate_mtune): Likewise.
32987 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
32989 PR rtl-optimization/78911
32990 * lra-assigns.c (must_not_spill_p): New function.
32991 (spill_for): Use it.
32993 2017-03-10 Jakub Jelinek <jakub@redhat.com>
32995 PR tree-optimization/79981
32996 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
32997 ATOMIC_COMPARE_EXCHANGE ifn result.
32998 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
32999 IFN_ATOMIC_COMPARE_EXCHANGE.
33001 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33004 * opts.c (parse_sanitizer_options): Add missing question mark to
33005 "did you mean" message.
33007 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33009 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
33011 (VMULEUH_UNS): Likewise.
33012 (VMULOUB_UNS): Likewise.
33013 (VMULOUH_UNS): Likewise.
33014 * config/rs6000/rs6000.c (builtin_function_type): Remove
33015 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
33017 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33020 * read-rtl-function.c (function_reader::read_rtx_operand): Update
33021 x with result of extra_parsing_for_operand_code_0.
33022 (function_reader::extra_parsing_for_operand_code_0): Convert
33023 return type from void to rtx, returning x. When reading
33024 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
33025 larger size containing struct block_symbol.
33027 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
33029 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
33030 -mfloat128-hardware without -m64.
33032 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
33035 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
33036 entries to the case statement that marks unsigned arguments to
33037 overloaded functions.
33039 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33041 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
33042 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
33044 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
33047 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
33048 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
33050 2017-03-10 Martin Liska <mliska@suse.cz>
33054 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
33055 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
33058 2017-03-10 Olivier Hainque <hainque@adacore.com>
33060 * tree-switch-conversion (array_value_type): Start by resetting
33061 candidate type to it's main variant.
33063 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33065 PR rtl-optimization/79909
33066 * combine.c (try_combine): Use simplify_replace_rtx on individual
33067 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
33068 of the whole CALL_INSN_FUNCTION_USAGE.
33070 PR tree-optimization/79972
33071 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
33072 get_range_info on SSA_NAMEs. Formatting fixes.
33074 2017-03-10 Richard Biener <rguenther@suse.de>
33075 Jakub Jelinek <jakub@redhat.com>
33077 PR tree-optimization/77975
33078 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
33079 edge to be constant.
33080 (get_val_for): For constant x return it. Formatting fix.
33081 (loop_niter_by_eval): Avoid pointless looping if the next iteration
33082 would use the same bases as the current one.
33084 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33086 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
33087 instead of vec_select for V1TImode.
33088 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
33090 (VSX_LE_128): Add V1TI to this mode iterator.
33091 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
33092 (*vsx_le_perm_store_<mode>): Likewise.
33093 (pre-reload splitter for VSX stores): Likewise.
33094 (post-reload splitter for VSX stores): Likewise.
33095 (*vsx_xxpermdi2_le_<mode>): Likewise.
33096 (*vsx_lxvd2x2_le_<mode>): Likewise.
33097 (*vsx_stxvd2x2_le_<mode>): Likewise.
33099 2017-03-09 Michael Eager <eager@eagercon.com>
33101 Correct failures with --enable-checking=yes,rtl.
33103 * config/microblaze/microblaze.c (microblaze_expand_shift):
33104 Replace GET_CODE test with CONST_INT_P and INTVAL test with
33105 test for const0_rtx.
33106 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
33107 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
33109 2017-03-09 Richard Biener <rguenther@suse.de>
33111 PR tree-optimization/79977
33112 * graphite-scop-detection.c (scop_detection::merge_sese):
33113 Handle the case of extra exits to blocks dominating the entry.
33115 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
33117 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
33120 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
33122 PR rtl-optimization/79949
33123 * lra-constraints.c (process_alt_operands): Check memory when
33124 trying to predict a cycle. Print about the overall increase.
33126 2017-03-09 Richard Biener <rguenther@suse.de>
33128 PR middle-end/79971
33129 * gimple-expr.c (useless_type_conversion_p): Preserve
33130 TYPE_SATURATING for fixed-point types.
33132 2017-03-09 Richard Biener <rguenther@suse.de>
33135 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
33136 alignment of BLKmode params.
33138 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33141 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
33142 (VALL_NO_V2Q): Likewise.
33144 * config/aarch64/aarch64-simd.md
33145 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
33147 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
33148 VALL_NO_V2Q mode iterator.
33149 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
33151 2017-03-09 Martin Liska <mliska@suse.cz>
33153 PR tree-optimization/79631
33154 * tree-chkp-opt.c (chkp_is_constant_addr): Call
33155 tree_int_cst_sign_bit just for INTEGER constants.
33157 2017-03-09 Martin Liska <mliska@suse.cz>
33161 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
33164 2017-03-09 Marek Polacek <polacek@redhat.com>
33167 * tree.c (inchash::add_expr): Handle TREE_VEC.
33169 2017-03-09 Martin Liska <mliska@suse.cz>
33172 (chkp_narrow_size_and_offset): New function.
33173 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
33174 (void chkp_parse_bit_field_ref): New function.
33175 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
33176 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
33178 2017-03-09 Martin Liska <mliska@suse.cz>
33181 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
33182 (chkp_find_bounds_1): Remove gcc_unreachable.
33184 2017-03-09 Jakub Jelinek <jakub@redhat.com>
33187 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
33188 BUILT_IN_SYNC*, determine the access type from the size suffix and
33189 always build a MEM_REF with that type. Handle forgotten
33190 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
33193 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
33194 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
33195 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
33196 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
33197 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
33198 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
33199 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
33200 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
33201 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
33202 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
33203 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
33204 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
33205 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
33206 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
33207 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
33208 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
33209 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
33210 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
33211 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
33212 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
33213 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
33214 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
33215 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
33216 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
33217 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
33218 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
33219 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
33220 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
33221 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
33222 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
33223 definitions outside of __OPTIMIZE__ guarded section.
33226 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
33227 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
33228 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
33229 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
33232 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33234 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
33235 ("vfenez<mode>"): Add missing constraints.
33237 2017-03-08 Martin Sebor <msebor@redhat.com>
33240 * config/nds32/nds32.c (nds32_option_override):
33241 Fix misspelled diagnostic.
33243 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33246 * gimplify.c (gimplify_omp_for): Replace index var in outer
33247 taskloop statement with an artificial variable and add
33248 OMP_CLAUSE_PRIVATE clause for it.
33250 2017-03-08 Richard Biener <rguenther@suse.de>
33252 PR tree-optimization/79955
33253 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
33254 for accesses that are completely outside of the variable.
33256 2017-03-08 Andrew Haley <aph@redhat.com>
33258 PR tree-optimization/79943
33259 * tree-ssa-loop-split.c (compute_new_first_bound): When
33260 calculating the new upper bound, (END-BEG) should be added, not
33263 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33265 * config/avr/avr.md (setmemhi): Make sure match_dup
33266 operand number comes before match_scratch.
33268 2017-03-08 Richard Biener <rguenther@suse.de>
33270 PR tree-optimization/79920
33271 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
33272 with ncopies == 1 to ...
33273 (vect_transform_slp_perm_load): ... here. Properly compute
33274 all element loads by iterating VF times over the group. Do
33275 not handle ncopies (computed in a broken way) in
33276 vect_create_mask_and_perm.
33278 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33281 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
33282 is a uniform vector, use uniform_vector_p return value instead of
33283 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
33285 2017-03-07 Marek Polacek <polacek@redhat.com>
33287 PR middle-end/79809
33288 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
33289 (alloca_call_type): Likewise.
33291 2017-03-07 Martin Liska <mliska@suse.cz>
33293 * gcov.c (process_args): Put comment to correct location.
33295 2017-03-07 Martin Liska <mliska@suse.cz>
33297 PR middle-end/68270
33298 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
33299 Use array_at_struct_end_p instead of DECL_CHAIN (field).
33300 (chkp_narrow_bounds_for_field): Likewise.
33301 (chkp_parse_array_and_component_ref): Pass one more argument to
33304 2017-03-07 Richard Biener <rguenther@suse.de>
33306 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
33309 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
33311 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
33312 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
33314 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33317 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
33318 to end of description.
33319 (PARAM_MAX_STORES_TO_MERGE): Likewise.
33321 2017-03-07 Jakub Jelinek <jakub@redhat.com>
33323 PR rtl-optimization/79901
33324 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
33326 (*avx512f_<code><mode>3<mask_name>): ... this.
33327 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
33328 iterator instead of VI8_AVX2_AVX512BW.
33330 PR rtl-optimization/79901
33331 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
33332 min/max expander, expand it using expand_vec_cond_expr.
33335 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
33338 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33341 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
33342 to void * for PCH reasons.
33343 * dwarf2out.c (output_loc_operands, output_die): Cast
33344 v.val_vec.array to unsigned char *.
33346 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
33349 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
33352 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
33354 PR rtl-optimization/79571
33355 * lra-constraints.c (process_alt_operands): Calculate static
33356 reject and subtract it from overall when only addresses will be
33359 2017-03-06 Julia Koval <julia.koval@intel.com>
33362 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
33363 incoming stack boundary to 128 for 64-bit targets.
33365 2017-03-06 Richard Biener <rguenther@suse.de>
33367 PR tree-optimization/79894
33368 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
33369 to NULL after folding it.
33371 2017-03-06 Richard Biener <rguenther@suse.de>
33373 PR tree-optimization/79824
33374 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
33375 check disabling peeling for gaps.
33377 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
33379 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
33380 attributes): Document gettimeofday.
33382 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33384 * config/s390/s390.c (s390_option_override_internal): Set
33385 PARAM_MIN_VECT_LOOP_BOUND
33387 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33389 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
33390 * config/s390/s390.md: Likewise.
33392 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33395 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
33396 (<avx2_avx512>_perm<mode>): Rename to ...
33397 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
33399 (<avx512>_perm<mode>_mask): Rename to ...
33400 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
33402 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
33403 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
33404 instead of VI8F_256_512.
33405 (avx512f_perm<mode>): New define_expand.
33406 (avx512f_perm<mode>_mask): Likewise.
33407 (avx512f_perm<mode>_1<mask_name>): New define_insn.
33408 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
33410 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33412 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
33413 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
33415 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
33417 2017-03-06 Martin Liska <mliska@suse.cz>
33420 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
33421 when having a SSA NAME w/o VAR_DECL assigned to it.
33423 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33425 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
33426 msa_dpsub_<su>_d): Fix MODE for vec_select.
33428 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33430 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
33432 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
33434 2017-03-06 Richard Biener <rguenther@suse.de>
33436 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
33437 * plugin.c (register_plugin_info): Likewise.
33438 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
33440 2017-03-05 Jakub Jelinek <jakub@redhat.com>
33442 * config/i386/sse.md (sse_storehps, sse_storelps,
33443 avx_<castmode><avxsizesuffix>_<castmode>,
33444 avx512f_<castmode><avxsizesuffix>_<castmode>,
33445 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
33446 in condition that at least one operand is not a MEM.
33448 2017-03-03 Jakub Jelinek <jakub@redhat.com>
33450 PR middle-end/79805
33451 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
33452 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
33454 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
33455 gimple_call_nothrow_p flag based on whether original builtin can throw.
33456 If it can, emit following stmts on the fallthrough edge.
33457 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
33458 don't create new bb if inserting just debug stmts on the edge, try to
33459 insert them on the fallthru bb or just reset debug stmts.
33461 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
33464 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
33465 restore recog_data (including the operand rtxes inside it) around
33466 the call to get_insn_template.
33468 2017-03-03 Martin Sebor <msebor@redhat.com>
33470 PR tree-optimization/79699
33471 * context.c (context::~context): Free MPFR caches to avoid
33472 a memory leak on program exit.
33474 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33476 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
33477 Use wide_int::ulow () instead of .elt (0).
33479 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
33481 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
33482 (*pushxf): Limit oF constraint to 32bit targets and add oC
33483 constraint for 64bit targets.
33484 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
33485 (*pushdf): Change rmF constraint to rmC.
33487 2017-03-03 Martin Liska <mliska@suse.cz>
33489 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
33490 Remove unused variable.
33492 2017-03-03 Jakub Jelinek <jakub@redhat.com>
33495 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
33496 is a memory operand, increase num_memory.
33497 (ix86_expand_args_builtin): Likewise.
33499 2017-03-03 Jan Hubicka <jh@suse.cz>
33502 * ipa-devirt.c (maybe_record_node): Properly handle
33503 __cxa_pure_virtual visibility.
33505 2017-03-03 Martin Liska <mliska@suse.cz>
33507 PR tree-optimization/79803
33508 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
33510 (pass_loop_prefetch::execute): Disabled optimization if an
33511 assumption about L1 cache size is not met.
33513 2017-03-03 Martin Liska <mliska@suse.cz>
33515 PR rtl-optimization/79574
33516 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
33517 (hash_scan_set): Likewise.
33518 (dump_hash_table): Likewise.
33519 (hoist_code): Likewise.
33521 2017-03-03 Richard Biener <rguenther@suse.de>
33523 * fixed-value.c (fixed_from_string): Restore use of elt (1)
33524 in place of uhigh ().
33525 (fixed_convert_from_real): Likewise.
33527 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
33530 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
33532 2017-03-03 Richard Biener <rguenther@suse.de>
33534 PR middle-end/79818
33535 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
33536 TYPE_OVERFLOW_UNDEFINED check.
33538 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33540 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
33542 (vector_ae_<mode>_p): Likewise.
33543 (vector_nez_<mode>_p): Likewise.
33544 (vector_ne_v2di_p): Likewise.
33545 (vector_ae_v2di_p): Likewise.
33546 (vector_ne_<mode>_p): Likewise.
33547 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
33549 (vsx_tsqrt<mode>2_fe): Likewise.
33551 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
33554 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
33556 2017-03-02 Jakub Jelinek <jakub@redhat.com>
33558 PR rtl-optimization/79780
33559 * cprop.c (one_cprop_pass): When second and further conditional trap
33560 in a single basic block is turned into an unconditional trap, turn it
33561 into a deleted note to avoid RTL verification failures.
33563 2017-03-02 Richard Biener <rguenther@suse.de>
33565 * fold-const.c (const_binop): Use ulow () instead of elt (0).
33567 2017-03-02 Richard Biener <rguenther@suse.de>
33569 PR tree-optimization/79345
33571 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
33572 param and abort the walk, returning -1 if it is hit.
33573 (walk_aliased_vdefs): Take a limit param and pass it on.
33574 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
33575 defaulting to 0 and return a signed int.
33576 * tree-ssa-uninit.c (struct check_defs_data): New struct.
33577 (check_defs): New helper.
33578 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
33579 about uninitialized memory.
33580 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
33581 bogus uninitialized warning.
33582 (fixed_convert_from_real): Likewise.
33584 2017-03-02 Bin Cheng <bin.cheng@arm.com>
33586 PR tree-optimization/66768
33587 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
33588 iv_use if base object can't be determined.
33590 2017-03-02 Jakub Jelinek <jakub@redhat.com>
33592 PR tree-optimization/79345
33593 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
33594 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
33595 (get_pattern_stats): Initialize it.
33596 * genemit.c (gen_expand): Verify match_scratch numbers come after
33597 match_operand/match_dup numbers.
33598 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
33599 match_scratch numbers.
33600 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
33602 * config/s390/s390.md (trunctdsd2): Likewise.
33604 2017-03-02 Richard Biener <rguenther@suse.de>
33606 * wide-int.h (wide_int_storage::operator=): Implement in terms
33609 2017-03-02 Richard Biener <rguenther@suse.de>
33611 PR tree-optimization/79777
33612 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
33613 the to insert expression to sth existing.
33615 2017-03-01 Martin Sebor <msebor@redhat.com>
33617 PR middle-end/79692
33618 * gimple-ssa-sprintf.c
33619 (directive::known_width_and_precision): New function.
33620 (format_integer): Use it.
33621 (get_mpfr_format_length): Consider the full range of precision
33622 when computing %g output with the # flag. Set the likely byte
33623 count to 3 rather than 1 when precision is indeterminate.
33624 (format_floating): Correct the lower bound of precision.
33626 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33628 * doc/invoke.texi: Document default code model for 64-bit Linux.
33630 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33633 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
33634 udiv rather than div since input pattern is unsigned.
33636 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
33638 * config/i386/i386.c (print_reg): Warn for values of
33639 unsupported size in integer register.
33641 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
33644 * config/rs6000/predicates.md (current_file_function_operand): Do
33645 not allow self calls to be local if the function is replaceable.
33647 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
33650 * config/rs6000/altivec.h (vec_ctz and others): Change the
33651 preprocessor macro that controls conditional compilation from
33652 _ARCH_PWR9 to __POWER9_VECTOR__.
33653 (vec_all_ne): Change parameterization of __altivec_scalar_pred
33654 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
33655 control (instead of _ARCH_PWR9 control) so that template
33656 definition uses power9-specific function.
33657 (vec_any_eq): Likewise.
33658 (vec_all_ne): Change macro definition to use a power9-specific
33659 expansion under #ifdef __POWER9_VECTOR__ control (instead of
33660 _ARCH_PWR9 control).
33661 (vec_any_eq) Likewise.
33662 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
33663 expansion for CMPNEF to remove support for xvcmpnesp instruction.
33664 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
33665 support for xvcmpnedp instruction.
33666 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
33667 macro expansion so that Power9 implementation of vec_all_ne does
33668 not use the AltiVec predicate framework.
33669 (VCMPNEH_P): Likewise.
33670 (VCMPNEW_P): Likewise.
33671 (VCMPNED_P): Likewise.
33672 (VCMPNEFP_P): Likewise.
33673 (VCMPNEDP_P): Likewise.
33674 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
33675 implementation of vec_any_eq to not use AltiVec predicate
33677 (VCMPAEH_P): Likewise.
33678 (VCMPAEW_P): Likewise.
33679 (VCMPAED_P): Likewise.
33680 (VCMPAEFP_P): Likewise.
33681 (VCMPAEDP_P): Likewise.
33682 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
33683 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
33684 not use the AltiVec predicate framework.
33685 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
33686 of vec_any_eq to not use AltiVec predicate framework.
33687 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
33688 support for predefined __POWER9_VECTOR__ macro to indicate that
33689 Power9 instruction selection is enabled.
33690 (altivec_overloaded_builtins): Remove extraneous
33691 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
33692 function argument types RS6000_BTI_bool_V16QI and
33693 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
33694 entry for overloaded function argument types RS6000_BTI_bool_V4SI
33695 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
33696 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
33697 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
33698 Power9 for implementations of vec_cmpne. Change the signature for
33699 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
33700 (representing vec_all_ne) to remove the previously described first
33701 argument of type RS6000_BTI_INTSI, as this was an artifact of
33702 reliance on the AltiVec predicate framework, which is no longer
33703 used in the implementation of these functions. Add
33704 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
33705 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
33706 since, unlike the AltiVec predicate framework implementation, we
33707 do not share function descriptors between vec_alle and vec_anyeq.
33708 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
33709 set of modes that receive special treatment even when
33710 TARGET_P9_VECTOR is true. The special treatment emits code that
33711 does not depend on Power9 instructions.
33712 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
33713 define_expand to not rely on AltiVec predicate framework.
33714 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33716 (vector_ne_v2di_p): Change this define_expand to not rely on
33717 AltiVec predicate framework.
33718 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
33720 (vector_ne_<mode>_p): Change this define_expand to not rely on
33721 AltiVec predicate framework.
33722 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33724 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
33725 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
33726 define_insn pattern.
33727 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
33728 define_insn pattern because the xvcmpne<VSs>. instruction is not
33730 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
33731 instruction is not supported.
33733 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33735 * config/nvptx/nvptx.c: Include intl.h.
33737 2017-03-01 Martin Jambor <mjambor@suse.cz>
33740 * ipa-prop.h (ipa_bits): Removed field known.
33741 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
33742 to pointers. Adjusted their comments to warn about their sharing.
33743 (ipcp_transformation_summary): Change bits to a vector of pointers.
33744 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
33745 (ipa_get_ipa_bits_for_value): Declare.
33746 * tree-vrp.h (value_range): Mark as GTY((for_user)).
33747 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
33748 (ipa_bits_hash_table): Likewise.
33749 (ipa_vr_ggc_hash_traits): Likewise.
33750 (ipa_vr_hash_table): Likewise.
33751 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
33752 being pointers and vr_known being removed.
33753 (ipa_set_jf_unknown): Likewise.
33754 (ipa_get_ipa_bits_for_value): New function.
33755 (ipa_set_jfunc_bits): Likewise.
33756 (ipa_get_value_range): New overloaded functions.
33757 (ipa_set_jfunc_vr): Likewise.
33758 (ipa_compute_jump_functions_for_edge): Use the above functions to
33759 construct bits and vr parts of jump functions.
33760 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
33761 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33763 (ipcp_grow_transformations_if_necessary): Also allocate
33764 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33766 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
33767 them. Fix too long lines.
33768 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
33769 vr_known being removed.
33770 (ipa_read_jump_function): Use new setter functions to construct bits
33771 and vr parts of jump functions or set them to NULL.
33772 (write_ipcp_transformation_info): Adjust for bits being pointers.
33773 (read_ipcp_transformation_info): Likewise.
33774 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
33776 Include gt-ipa-prop.h.
33777 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
33779 (ipcp_store_bits_results): Likewise.
33780 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
33781 Do not write to existing jump functions but use a temporary instead.
33783 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33786 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
33787 attempt to use its first operand as BIT_FIELD_REF base.
33789 2017-03-01 Richard Biener <rguenther@suse.de>
33791 PR middle-end/79721
33792 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
33793 interpolating formula in wrapping arithmetic.
33794 (chrec_apply): Convert chrec_evaluate return value to wanted type.
33796 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33798 PR tree-optimization/79734
33799 * tree-vect-generic.c (expand_vector_condition): Optimize
33800 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
33801 Handle VEC_COND_EXPR where comparison has different inner width from
33802 type's inner width.
33804 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
33806 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
33807 markup, and similar issues. Remove @opindex entries for things
33808 that aren't options. Add missing -mmpy-option entries.
33810 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33812 PR tree-optimization/79737
33813 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
33814 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
33815 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
33816 instead of byte_size. Formatting fix.
33817 (shift_bytes_in_array_right): Formatting fix.
33819 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
33822 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
33823 condition on optimize for the leaf function test.
33825 2017-02-28 Martin Liska <mliska@suse.cz>
33828 * read-rtl-function.c (function_reader::handle_unknown_directive):
33829 Bail out when one uses -flto.
33831 2017-02-28 Martin Liska <mliska@suse.cz>
33833 * common.opt: Replace space with tabular for options of <number>
33835 * config/i386/i386.opt: Show <number> value for
33836 -mlarge-data-threshold.
33837 * opts.c (print_filtered_help): Do not display number in hexadecimal
33840 2017-02-28 Martin Liska <mliska@suse.cz>
33842 * common.opt: Fix --help=option -Q for options which are of
33845 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
33847 * config/i386/i386.c (print_reg): Error out for values
33848 of 8-bit size in invalid integer register.
33850 2017-02-28 Martin Sebor <msebor@redhat.com>
33852 PR tree-optimization/79691
33853 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
33855 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33858 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
33859 gcc_unreachable with output_operand_lossage.
33861 2017-02-28 Richard Biener <rguenther@suse.de>
33863 PR tree-optimization/79740
33864 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
33866 (visit_nary_op): Insert the nary into the hashtable if we
33867 pattern-matched sth.
33868 * tree-ssa-pre.c (eliminate_insert): Robustify.
33870 2017-02-28 Richard Biener <rguenther@suse.de>
33872 PR middle-end/79731
33873 * fold-const.c (decode_field_reference): Reject out-of-bound
33876 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33878 * config/i386/i386.c: Include intl.h.
33879 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
33880 instead of just cond ? "..." : "...".
33881 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
33882 * coverage.c (read_counts_file): Likewise.
33883 * omp-offload.c: Include intl.h.
33884 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
33885 of just cond ? "..." : "...".
33886 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
33887 of just cond ? "..." : "...".
33889 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
33892 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
33894 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
33895 'tune for' CPU name.
33896 * config/arm/arm-cpu-data.h: Regenerated.
33898 2017-02-28 Richard Biener <rguenther@suse.de>
33900 PR tree-optimization/79732
33901 * tree-inline.c (expand_call_inline): Do not shadow var.
33903 2017-02-28 Richard Biener <rguenther@suse.de>
33905 PR tree-optimization/79723
33906 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
33907 address-space properly.
33909 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
33911 * doc/optinfo.texi (Optimization groups): Fix option used for
33913 * doc/invoke.texi (-fopt-info): Document "omp".
33914 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
33915 (OPTGROUP_ALL): Add OPTGROUP_OMP.
33916 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
33917 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
33918 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
33920 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
33922 * dumpfile.c (optgroup_options): Instead of "openmp", associate
33923 OPTGROUP_OMP with "omp".
33925 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
33928 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
33929 for arithmetic shift of unsigned V2DI.
33931 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
33933 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
33934 arc/linux.h headers.
33935 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
33936 (LINK_SPEC): Likewise.
33937 (ARC_TLS_EXTRA_START_SPEC): Likewise.
33938 (EXTRA_SPECS): Likewise.
33939 (STARTFILE_SPEC): Likewise.
33940 (ENDFILE_SPEC): Likewise.
33941 (LIB_SPEC): Likewise.
33942 (TARGET_SDATA_DEFAULT): Likewise.
33943 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
33944 (MULTILIB_DEFAULTS): Likewise.
33945 (DWARF2_UNWIND_INFO): Likewise.
33946 * config/arc/big.h: New file.
33947 * config/arc/elf.h: Likewise.
33948 * config/arc/linux.h: Likewise.
33949 * config/arc/t-uClibc: Remove.
33951 2017-02-27 Bin Cheng <bin.cheng@arm.com>
33953 PR tree-optimization/77536
33954 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
33955 (tree_transform_and_unroll_loop): Use above function to compute the
33956 estimated niter of unrolled loop and use it when scaling profile.
33957 Also use count info rather than frequency if it's non-zero.
33958 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
33959 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
33960 (vect_transform_loop): Call above function.
33962 2017-02-27 Richard Biener <rguenther@suse.de>
33964 PR tree-optimization/45397
33965 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
33966 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
33967 (visit_nary_op): Add pattern matching for CSEing sign-changed
33968 or truncated operations with wider ones.
33970 2017-02-27 Richard Biener <rguenther@suse.de>
33972 PR tree-optimization/79690
33973 * tree-vect-stmts.c (vectorizable_store): Use vector type
33974 built from the DR with address-space.
33976 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
33978 * doc/invoke.texi (Optimize Options): Refine the description
33979 of asan-use-after-return.
33981 2017-02-25 Alan Modra <amodra@gmail.com>
33983 PR rtl-optimization/79584
33984 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
33985 base, not ad->base_term, the reg within base. Remove assertion
33986 that ad->base == ad->base_term. Replace gen_int_mode using
33987 bogus mode with const0_rtx.
33989 2017-02-25 Jakub Jelinek <jakub@redhat.com>
33991 PR middle-end/79396
33992 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
33993 FMA_EXPR like tcc_binary or tcc_unary.
33995 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
33998 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
34000 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
34001 (output_loc_operands): Handle DW_OP_call_ref and
34002 DW_OP_GNU_variable_value.
34003 (struct variable_value_struct): New type.
34004 (struct variable_value_hasher): Likewise.
34005 (variable_value_hash): New variable.
34006 (string_types): Remove.
34007 (copy_loc_descr): New function.
34008 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
34009 (prepend_loc_descr_to_each): New function.
34010 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
34011 instead of add_loc_descr_to_each if the first argument is single
34012 location list and the second has multiple.
34013 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
34014 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
34015 when looking for variable value which doesn't have other location info.
34016 (loc_list_from_tree): Formatting fix.
34017 (gen_array_type_die): Simplify DW_AT_string_length handling.
34018 (adjust_string_types): Remove.
34019 (gen_subprogram_die): Don't call adjust_string_types nor test/set
34020 string_types. Call resolve_variable_values.
34021 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
34022 (resolve_addr_in_expr): Likewise. Add A argument.
34023 (copy_deref_exprloc): Remove deref argument. Adjust for the
34024 original expression being DW_OP_GNU_variable_value with optionally
34025 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
34026 optionally after it.
34027 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
34028 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
34029 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
34030 (variable_value_hasher::hash, variable_value_hasher::equal): New
34032 (resolve_variable_value_in_expr, resolve_variable_value,
34033 resolve_variable_values, note_variable_value_in_expr,
34034 note_variable_value): New functions.
34035 (dwarf2out_early_finish): Call note_variable_value on all toplevel
34038 2017-02-24 Jakub Jelinek <jakub@redhat.com>
34041 * opts.h (handle_generated_option): Add GENERATED_P argument.
34042 * opts-common.c (handle_option): Adjust function comment.
34043 (handle_generated_option): Add GENERATED_P argument, pass it to
34045 (control_warning_option): Pass false to handle_generated_option
34047 * opts.c (maybe_default_option): Pass true to handle_generated_option
34049 * optc-gen.awk: Likewise.
34051 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34053 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
34054 a REG, look at the REG it is a SUBREG of.
34055 (splitter for cmpeqsi_t): Ditto.
34057 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34059 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
34060 the special USEs with the pattern of the insn, not the insn itself.
34062 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
34065 * doc/invoke.texi: Document -mload-store-pairs.
34067 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34068 Sandra Loosemore <sandra@codesourcery.com>
34070 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
34071 argument isn't a CONST_INT.
34072 (nios2_alternate_compare_const): Assert op is a CONST_INT.
34073 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
34074 (nios2_validate_compare): Bypass alternate compare logic if *op2
34075 is not a CONST_INT.
34076 (ldstwm_operation_p): Return false if first_base is not a REG or
34077 if first_offset is not a CONST_INT.
34079 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34081 * config/cris/cris.md: Use correct operand in a define_peephole2.
34083 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34085 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
34087 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34089 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
34090 this_insn if it is an INSN or JUMP_INSN.
34091 (force_offsettable): Look at base, not at addr.
34092 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
34093 on things that aren't necessarily CONST_INTs.
34095 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
34097 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
34098 -mfpmath=sse is the default also for x86-32 targets with SSE2
34099 instruction set when @option{-ffast-math} is enabled
34101 2017-02-24 Jeff Law <law@redhat.com>
34103 PR rtl-optimizatoin/79286
34104 * ira.c (update_equiv_regs): Drop may_trap_p exception to
34107 2017-02-24 Richard Biener <rguenther@suse.de>
34109 PR tree-optimization/79389
34110 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
34113 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
34115 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
34116 function comment to reflect reality.
34117 (loop_exits_before_overflow): Fix typo in function description.
34119 2017-02-24 Richard Biener <rguenther@suse.de>
34121 PR tree-optimization/79389
34122 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
34123 properly that a threading opportunity exists. Detect conditional
34124 copy/constant propagation opportunities.
34126 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
34128 * config/visium/visium.md (type): Add trap.
34129 (b): New mode attribute.
34130 (*btst): Rename into...
34131 (*btst<mode>): ...this and adjust.
34132 (*cbranchsi4_btst_insn): Rename into...
34133 (*cbranch<mode>4_btst_insn): ...this and adjust.
34134 (trap): New define_insn.
34136 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34138 PR tree-optimization/79389
34139 * ifcvt.c (struct noce_if_info): Add rev_cond field.
34140 (noce_reversed_cond_code): New function.
34141 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
34142 reversed_comparison_code. Formatting fix.
34143 (noce_try_store_flag): Test rev_cond != NULL in addition to
34144 reversed_comparison_code.
34145 (noce_try_store_flag_constants): Likewise.
34146 (noce_try_store_flag_mask): Likewise.
34147 (noce_try_addcc): Use rev_cond if non-NULL instead of
34148 reversed_comparison_code.
34149 (noce_try_cmove_arith): Likewise. Formatting fixes.
34150 (noce_try_minmax, noce_try_abs): Clear rev_cond.
34151 (noce_find_if_block): Initialize rev_cond.
34152 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
34153 instead of false as last argument never attempt to reverse it
34156 2017-02-23 Bin Cheng <bin.cheng@arm.com>
34158 PR tree-optimization/79663
34159 * tree-predcom.c (combine_chains): Process refs in reverse order
34160 only for ZERO length chains, and add explaining comment.
34162 2017-02-23 Jeff Law <law@redhat.com>
34164 PR tree-optimization/79578
34165 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
34166 in call to operand_equal_p.
34168 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
34171 * config/i386/cpuid.h: Fix another undefined behavior.
34173 2017-02-23 Richard Biener <rguenther@suse.de>
34175 PR tree-optimization/79683
34176 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
34177 vector types for data-refs.
34179 2017-02-23 Martin Liska <mliska@suse.cz>
34181 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
34183 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34185 PR middle-end/79665
34186 * internal-fn.c (get_range_pos_neg): Moved to ...
34187 * tree.c (get_range_pos_neg): ... here. No longer static.
34188 * tree.h (get_range_pos_neg): New prototype.
34189 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
34190 are known to be in between 0 and signed maximum inclusive, try to
34191 expand both unsigned and signed divmod and use the cheaper one from
34194 2017-02-22 Jeff Law <law@redhat.com>
34196 PR tree-optimization/79578
34197 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
34198 to compare base operands.
34200 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
34203 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
34204 gpc_reg_operand instead of fpr_reg_operand.
34206 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
34208 * config/mips/mips.c (mips_return_in_memory): Force FP
34209 vector types to be returned in memory for o32 ABI.
34211 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34213 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
34214 instead of DW_TAG_member for static data member declarations and don't
34215 set no_linkage_name for static inline data members.
34216 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
34219 2017-02-22 Martin Liska <mliska@suse.cz>
34221 * doc/invoke.texi: Replace inequality signs with square brackets
34224 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
34227 * lra-constraints.c (simplify_operand_subreg): Handle
34228 WORD_REGISTER_OPERATIONS targets.
34230 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34233 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
34234 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
34235 elimination by swapping fld*.
34237 2017-02-22 Richard Biener <rguenther@suse.de>
34239 PR tree-optimization/79673
34240 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
34241 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
34242 irrelevant address-space qualifiers and avoiding a
34243 ADDR_SPACE_CONVERT_EXPR from fold_convert.
34245 2017-02-22 Richard Biener <rguenther@suse.de>
34247 PR tree-optimization/79666
34248 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
34249 to not symbolically negate if that may introduce undefined
34252 2017-02-22 Martin Liska <mliska@suse.cz>
34255 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
34256 * data-streamer-out.c (streamer_write_gcov_count_stream):
34258 * value-prof.c (stream_out_histogram_value): Make assert more
34259 precise based on type of counter.
34261 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
34264 * config/i386/i386.md (standard_x87sse_constant_load splitter):
34265 Use nonimmediate_operand instead of memory_operand for operand 1.
34266 (float-extend standard_x87sse_constant_load splitter): Ditto.
34268 2017-02-21 Jeff Law <law@redhat.com>
34270 PR tree-optimization/79621
34271 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
34272 blocks with edges to themselves.
34274 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34277 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
34278 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
34279 Use gimple_call_builtin_p.
34282 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
34283 on temporarily removed DEBUG_INSNs.
34285 PR tree-optimization/79649
34286 * tree-loop-distribution.c (classify_partition): Give up on
34287 non-generic address space loads/stores.
34289 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
34291 * doc/loop.texi (Loop manipulation): Remove nonexistent
34292 tree_ssa_loop_version from the documentation.
34293 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
34295 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34298 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
34299 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
34300 * config/rs6000/rs6000.c: Include except.h.
34301 (rs6000_expand_split_stack_prologue): Call
34302 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
34304 2017-02-21 Martin Jambor <mjambor@suse.cz>
34307 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
34308 have been analyzed.
34310 2017-02-21 Martin Jambor <mjambor@suse.cz>
34312 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
34313 for backward compatibility only.
34314 * doc/invoke.texi (Option Summary): Remove all references to
34315 -fipa-cp-alignment.
34317 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
34321 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34323 * lra-constraints.c (curr_insn_transform): Handle
34324 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34326 2017-02-21 Martin Liska <mliska@suse.cz>
34328 * config/i386/i386.opt: Replace -masm-dialect with -masm.
34330 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
34332 PR translation/79638
34333 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
34335 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
34338 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
34339 (arm_function_ok_for_sibcall): Return false for an indirect call by
34340 descriptor if all the argument registers are used.
34341 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
34342 alignment of the function.
34344 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34346 PR tree-optimization/61441
34347 * simplify-rtx.c (simplify_const_unary_operation): For
34348 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
34349 the sNaN unmodified.
34351 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34353 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34354 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34355 instead of SYSTEM_HEADER_DIR.
34357 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
34358 Martin Liška <mliska@suse.cz>
34360 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
34361 Fix typos and grammar, use active voice, and clarify.
34363 2017-02-20 Marek Polacek <polacek@redhat.com>
34365 PR middle-end/79537
34366 * gimplify.c (gimplify_expr): Handle unused *&&L;.
34369 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
34371 2017-02-20 Jakub Jelinek <jakub@redhat.com>
34374 * config/i386/i386.c (ix86_expand_builtin): Handle
34375 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
34376 ix86_builtins_isa[fcode].isa as a requirement of those
34377 flags and any other flag in the bitmask.
34378 (ix86_init_mmx_sse_builtins): Use 0 instead of
34379 ~OPTION_MASK_ISA_64BIT as mask.
34380 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
34381 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
34382 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
34383 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
34385 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34388 * lra-constraints.c (split_reg): Check requested split mode
34389 is supported by the register.
34391 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34393 * lra-constraints.c (simplify_operand_subreg): Remove early
34396 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34399 * lra-constraints.c (curr_insn_transform): Tighten condition
34400 for converting SUBREG reloads from OP_OUT to OP_INOUT.
34402 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34405 * lra-constraints.c (curr_insn_transform): Handle
34406 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34408 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
34411 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
34413 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
34415 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
34418 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
34421 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34423 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
34426 2017-02-18 Jakub Jelinek <jakub@redhat.com>
34428 * final.c (last_columnnum, override_columnnum): New variables.
34429 (final_start_function): Set last_columnnum, pass it to begin_prologue
34430 hook and pass 0 to dwarf2out_begin_prologue.
34431 (final_scan_insn): Update override_columnnum. Pass last_columnnum
34432 to source_line debug hook.
34433 (notice_source_line): Compute last_columnnum and for debug_column_info
34434 return true on column changes.
34435 * debug.h (struct gcc_debug_hooks): Add column argument to
34436 source_line and begin_prologue hooks.
34437 (debug_nothing_int_charstar_int_bool): Remove prototype.
34438 (debug_nothing_int_int_charstar,
34439 debug_nothing_int_int_charstar_int_bool): New prototypes.
34440 (dwarf2out_begin_prologue): Add column argument.
34441 * debug.c (do_nothing_debug_hooks): Adjust source_line and
34442 begin_prologue hooks.
34443 (debug_nothing_int_charstar_int_bool): Remove.
34444 (debug_nothing_int_int_charstar,
34445 debug_nothing_int_int_charstar_int_bool): New functions.
34446 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
34447 through to dwarf2out_source_line.
34448 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
34449 (dwarf2out_source_line): Add column argument, emit it if requested.
34450 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
34452 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
34453 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
34454 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
34455 through to dwarf2out_begin_prologue.
34456 (vmsdbgout_source_line): Add column argument, pass it through to
34457 dwarf2out_source_line.
34458 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
34459 dbxout_source_line caller.
34460 (dbxout_source_line): Add column argument.
34462 * common.opt (gno-column-info, gcolumn-info): New options.
34463 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
34464 (check_die): Also test for multiple DW_AT_decl_column attributes.
34465 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
34466 DW_AT_decl_column if requested.
34467 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
34469 (gen_variable_die): Likewise.
34470 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
34471 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
34474 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
34475 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
34476 (ix86_handle_option): Handle OPT_m3dnowa.
34477 * doc/invoke.texi (-m3dnowa): Document.
34478 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
34479 -m3dnowa instead of -m3dnow -march=athlon.
34482 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
34483 instead of gcc_assert for K, r and R code checks. Formatting fixes.
34485 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34488 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
34489 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
34490 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
34491 generator for vsx_xxpermdi_<mode>_be.
34492 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
34493 force big-endian semantics.
34494 (vsx_xxpermdi_<mode>_be): New define_expand with same
34495 implementation as previous version of vsx_xxpermdi_<mode>.
34497 2017-02-17 Jakub Jelinek <jakub@redhat.com>
34499 PR tree-optimization/79327
34500 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
34501 variable, its initialization and use.
34503 2017-02-17 Julia Koval <julia.koval@intel.com>
34505 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
34506 (OPTION_MASK_ISA_PKU_UNSET): New.
34507 (ix86_handle_option): Handle -mrdpid.
34508 * config/i386/cpuid.h (bit_RDPID): New.
34509 * config/i386/driver-i386.c (host_detect_local_cpu):
34510 Detect RDPID feature.
34511 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
34512 * config/i386/i386-c.c (ix86_target_macros_internal):
34514 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
34515 (ix86_valid_target_attribute_inner_p): Add "rdpid".
34516 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
34517 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
34518 * config/i386/i386.md (define_insn "rdpid"): New.
34519 * config/i386/i386.opt Add -mrdpid.
34520 * config/i386/immintrin.h (_rdpid_u32): New.
34522 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
34524 PR rtl-optimization/79541
34525 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
34526 instead of transforming it into USE.
34528 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
34530 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
34531 If HONOR_SNANS (SFmode) force the input to a register.
34532 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
34533 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
34534 an frsp or similar insn.
34536 2017-02-17 Martin Liska <mliska@suse.cz>
34538 PR rtl-optimization/79577
34539 * params.def (selsched-max-sched-times): Increase minimum to 1.
34541 2017-02-17 Martin Liska <mliska@suse.cz>
34543 PR rtl-optimization/79574
34544 * gcse.c (want_to_gcse_p): Prevent integer overflow.
34546 2017-02-17 Martin Liska <mliska@suse.cz>
34548 PR tree-optimization/79529
34549 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
34550 ssa_defined_default_def_p to handle cases which are implicitly
34552 * tree-ssa.c (ssa_defined_default_def_p): New function.
34553 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
34554 which are implicitly defined.
34555 * tree-ssa.h (ssa_defined_default_def_p): Declare.
34557 2017-02-17 Richard Biener <rguenther@suse.de>
34559 PR middle-end/79576
34560 * params.def (max-ssa-name-query-depth): Limit to 10.
34562 2017-02-17 Richard Biener <rguenther@suse.de>
34564 PR tree-optimization/79552
34565 * tree-ssa-structalias.c (visit_loadstore): Properly verify
34568 2017-02-17 Richard Biener <rguenther@suse.de>
34571 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
34573 2017-02-17 Marek Polacek <polacek@redhat.com>
34575 PR middle-end/79536
34576 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
34577 (fold_negate_expr): New wrapper.
34579 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
34581 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
34582 Correct terminology and de-emphasize pre-standard behavior.
34584 2017-02-16 Alan Modra <amodra@gmail.com>
34586 PR rtl-optimization/79286
34587 * ira.c (def_dominates_uses): New function.
34588 (update_equiv_regs): Don't create an equivalence for insns that
34589 may trap where the register def does not dominate the use.
34591 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
34593 PR rtl-optimization/78127
34594 * lra.c (lra): Call lra_eliminate before finish the loop after
34597 2017-02-16 Richard Biener <rguenther@suse.de>
34599 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
34601 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
34602 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
34603 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
34604 (isl_val_int_from_wi): New function.
34605 (extract_affine_gmp): Rename to ...
34606 (extract_affine_wi): ... this, take a widest_int.
34607 (extract_affine_int): Just wrap extract_affine_wi.
34608 (add_param_constraints): Use isl_val_int_from_wi.
34609 (add_loop_constraints): Likewise, and extract_affine_wi.
34611 2017-02-15 Jeff Law <law@redhat.com>
34613 PR middle-end/79521
34614 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
34615 ira_init_register_move_cost_if_necessary.
34617 2017-02-15 Martin Sebor <msebor@redhat.com>
34619 PR middle-end/32003
34620 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
34621 removed in a prior commit.
34623 2017-02-15 Bin Cheng <bin.cheng@arm.com>
34625 PR tree-optimization/79347
34626 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
34627 counters during peeling.
34629 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
34631 * Makefile.in (site.exp): Remove "set ISLVER".
34633 2017-02-15 Jakub Jelinek <jakub@redhat.com>
34636 * real.c (real_from_integer): Call real_convert even for decimal.
34638 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
34641 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
34643 2017-02-14 Andrew Pinski <apinski@cavium.com>
34645 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
34646 cores and change the partno/implementer to be correct.
34647 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
34648 the 'B" as the implementer.
34649 * config/aarch64/aarch64-tune.md: Regenerate.
34651 2017-02-14 Carl Love <cel@us.ibm.com>
34653 * config/rs6000/rs6000.c: Add case statement entry to make the
34654 xvcvuxdsp built-in argument unsigned.
34655 * config/rs6000/vsx.md: Fix the source and return operand types so they
34656 match the instruction definitions from the ISA document. Fix typo
34657 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
34660 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
34663 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
34664 member early_clobber_alts.
34665 * lra-lives.c (reg_early_clobber_p): New.
34666 (process_bb_lives): Use it.
34667 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
34668 (debug_operand_data): Initialize early_clobber_alts.
34669 (setup_operand_alternative): Set up early_clobber_alts.
34670 (collect_non_operand_hard_regs): Ditto. Pass early clobber
34671 alternatives to new_insn_reg.
34672 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
34674 (lra_update_insn_regno_info): Pass the new arg.
34676 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34678 PR middle-end/79505
34679 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
34680 (new_oacc_loop_raw): Don't clear already cleared fields.
34683 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
34684 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
34685 _mm512_prefetch_i64gather_ps): New inline functions and macros.
34687 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
34690 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
34692 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
34695 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
34696 the extra instruction to the right place to store 128-bit constant
34699 2017-02-14 Martin Sebor <msebor@redhat.com>
34701 PR middle-end/79448
34702 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
34703 warning for strings of unknown length.
34705 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
34707 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
34709 2017-02-14 Jeff Law <law@redhat.com>
34712 * ira-costs.c (scan_one_insn): Initialize register move costs
34713 for pseudos seen in USE/CLOBBER insns.
34715 PR tree-optimization/79095
34716 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
34717 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
34718 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
34719 if the operands are known to be not equal, then the resulting range
34721 (intersect_ranges): If the new range is ~[0,0] and the old range is
34722 wide, then prefer ~[0,0].
34723 * tree-vrp.c (overflow_comparison_p_1): New function.
34724 (overflow_comparison_p): New function.
34725 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
34726 if NAME is used in an overflow test.
34727 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
34728 overflow check that can be expressed as an equality test, then adjust
34729 ops to be that equality test.
34731 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34733 * config/s390/s390-builtin-types.def: Remove flags argument.
34734 * config/s390/s390.c (s390_init_builtins): Likewise.
34736 2017-02-14 Martin Liska <mliska@suse.cz>
34738 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
34739 vector. Fix trailing white spaces.
34741 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
34743 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
34746 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34748 PR rtl-optimization/68664
34749 * config/arm/arm.c (arm_sched_can_speculate_insn):
34750 New function. Declare prototype.
34751 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34753 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34755 PR rtl-optimization/68664
34756 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
34758 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34760 2017-02-14 Amit Pawar <amit.pawar@amd.com>
34762 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
34763 max skip bytes for function, loop and jump.
34765 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34767 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
34768 ABS_EXPR for gimple dump.
34770 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34773 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
34775 PR tree-optimization/79408
34776 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
34777 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
34778 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
34779 also if rhs1 is INTEGER_CST.
34781 2017-02-14 Richard Biener <rguenther@suse.de>
34783 PR middle-end/79432
34784 * tree-into-ssa.c (insert_phi_nodes): When the function can
34785 have abnormal edges rewrite SSA names with broken use-def
34786 dominance out of SSA and register them for PHI insertion.
34788 2017-02-13 Martin Sebor <msebor@redhat.com>
34790 PR middle-end/79496
34791 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
34792 clearing info.nowrite flag when snprintf size argument is a range.
34794 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34796 * cprop.c (cprop_jump): Add missing space in string literal.
34797 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
34798 (get_constraint_for_component_ref): Likewise.
34799 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
34800 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
34801 * lra-constraints.c (process_alt_operands): Likewise.
34802 * ipa-inline.c (inline_small_functions): Likewise.
34803 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
34804 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
34805 * trans-mem.c (diagnose_tm_1_op): Likewise.
34806 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
34807 (grid_parallel_clauses_gridifiable): Likewise.
34809 * config/nvptx/mkoffload.c (process): Add space in between
34812 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
34813 "MOD4_SSE_REGS" and "ALL_REGS".
34815 * spellcheck.c (test_data): Add , in between "foo" and "food".
34817 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34820 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
34821 boundary crossing check and subsequent code generation agree.
34823 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34825 * config/aarch64/aarch64.c (has_memory_op): Delete.
34826 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
34829 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34831 PR rtl-optimization/79388
34832 PR rtl-optimization/79450
34833 * combine.c (distribute_notes): When removing TEM_INSN for which
34834 corresponding dest has last value recorded, invalidate that last
34837 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34839 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
34840 of explicit '@'. Add missing assembly comment marker on branch costs
34843 2017-02-13 Nathan Sidwell <nathan@acm.org>
34845 * gengtype-lex.l (<in_struct>): Add '/'.
34847 2017-02-13 Martin Liska <mliska@suse.cz>
34850 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
34852 2017-02-13 Richard Biener <rguenther@suse.de>
34854 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
34856 * configure: Re-generate.
34857 * config.in: Likewise.
34858 * graphite-dependences.c: Simplify as if
34859 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
34860 * graphite-isl-ast-to-gimple.c: Likewise.
34861 * graphite-optimize-isl.c: Likewise.
34862 * graphite-poly.c: Likewise.
34863 * graphite-sese-to-poly.c: Likewise.
34864 * graphite.h: Likewise.
34865 * toplev.c: Include isl/version.h and use isl_version () for
34866 printing the ISL version.
34867 * doc/install.texi: Update ISL requirement.
34869 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
34871 * doc/standards.texi (Standards): Update reference to
34874 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
34876 * doc/extend.texi (Named Address Spaces): sourceware.org now
34878 * doc/install.texi (Binaries): Ditto.
34881 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
34883 * doc/cpp.texi: Replace "stringify"/"stringification" with C
34884 standard terminology "stringize"/"stringizing" throughout.
34885 * doc/cppinternals.texi: Likewise.
34887 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
34889 * doc/extend.texi: Fix some spelling mistakes and typos.
34890 * doc/invoke.texi: Likewise.
34892 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34895 * params.def (inline-min-speedup) Change from 10 to 8.
34897 2017-02-11 Jakub Jelinek <jakub@redhat.com>
34899 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
34902 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34905 * ipa-inline-analysis.c (get_minimal_bb): New function.
34906 (record_modified): Use it.
34907 (remap_edge_change_prob): Handle also ancestor functions.
34909 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
34911 * doc/contrib.texi (Contributors): Remove broken link into
34912 the Mauve CVS repository.
34914 2017-02-11 Jakub Jelinek <jakub@redhat.com>
34916 PR middle-end/79454
34917 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
34918 result computation whenever lhs doesn't have vector mode, not
34919 just when it has BLKmode.
34921 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
34923 * doc/makefile.texi (profiledbootstrap): Refer to the
34924 installation instructions only in textual form.
34926 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34929 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
34931 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
34933 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
34934 (Specific): Update mingw-w64 reference.
34936 (Specific): Remove broken link to Renesas RX processor.
34938 2017-02-10 Richard Biener <rguenther@suse.de>
34940 * toplev.c (process_options): Do not mention obsolete graphite
34941 options when printing sorry message about missing graphite support.
34942 Mention -floop-nest-optimize.
34944 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
34946 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
34947 (vtst_p16): Likewise.
34948 (vtstq_p8): Likewise.
34949 (vtstq_p16): Likewise.
34951 (vtstq_p64): Likewise.
34952 * config/arm/arm_neon.h (vgetq_lane_p64): New.
34953 (vset_lane_p64): New.
34954 (vsetq_lane_p64): New.
34956 2017-02-10 Jakub Jelinek <jakub@redhat.com>
34958 PR tree-optimization/79411
34959 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
34960 stmt operands are SSA_NAMEs used in abnormal phis.
34961 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
34964 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
34967 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
34970 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
34972 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
34974 2017-02-09 Jakub Jelinek <jakub@redhat.com>
34976 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
34980 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
34981 not arbitrary TREE_CONSTANT.
34984 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
34985 "omp declare target link" attribute unless is_global_var.
34986 * omp-offload.c (find_link_var_op): Likewise.
34988 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
34989 Chung-Lin Tang <cltang@codesourcery.com>
34991 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
34993 (gimplify_adjust_omp_clauses): Don't delete TILE.
34994 (gimplify_omp_for): Deal with TILE.
34995 * internal-fn.c (expand_GOACC_TILE): New function.
34996 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
34998 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
34999 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
35001 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
35002 avoid DIV for outermost collapse var.
35003 (expand_oacc_for): Insert tile element loop as needed. Adjust.
35004 Remove out of date comments, fix whitespace.
35005 * omp-general.c (omp_extract_for_data): Deal with tiling.
35006 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
35007 adjust OLF_DIM_BASE value.
35008 (struct omp_for_data): Add tiling field.
35009 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
35010 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
35011 for auto loops. Remove default auto determining, moved to
35012 oacc_loop_fixed_partitions.
35013 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
35014 stmts, add e_mask field.
35015 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
35016 (oacc_thread_numbers): Use oacc_dim_call.
35017 (oacc_xform_tile): New.
35018 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
35019 (finish_oacc_loop): Adjust for ifns vector.
35020 (oacc_loop_discover_walk): Append loop abstraction sites to list,
35021 add case for GOACC_TILE fns.
35022 (oacc_loop_xform_loop): Delete.
35023 (oacc_loop_process): Iterate over call list directly, and add
35024 handling for GOACC_TILE fns.
35025 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
35027 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
35028 vector partitioning to outer loops. Assign 2 partitions to loops
35029 when available. Add TILE handling.
35030 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
35031 (execite_oacc_device_lower): Process GOACC_TILE fns,
35032 ignore unknown specs.
35033 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
35034 * tree.c (omp_clause_num_ops): Adjust TILE ops.
35035 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
35037 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
35039 * configure.ac (ACX_BUGURL): Update.
35040 * configure: Regenerate.
35042 2017-02-09 Richard Biener <rguenther@suse.de>
35044 PR tree-optimization/69823
35045 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
35046 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
35048 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35050 * config/arc/arc-c.def: Add __NPS400__ definition.
35051 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
35052 (TARGET_NPS400): Define.
35054 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35056 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
35058 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
35059 pointer, arch_info.
35060 (arc_cpu_types): Fill the arch_info field with a pointer into the
35061 arc_arch_types table.
35062 (arc_selected_cpu): Declare.
35063 * config/arc/arc.c (arc_selected_cpu): Make global.
35064 (arc_selected_arch): Delete.
35065 (arc_base_cpu): Delete.
35066 (arc_override_options): Remove references to deleted variables,
35067 update access to arch information.
35068 (ARC_OPT): Update access to arch information.
35069 (ARC_OPTX): Likewise.
35070 * config/arc/arc.h (arc_base_cpu): Remove declaration.
35071 (TARGET_ARC600): Update access to arch information.
35072 (TARGET_ARC601): Likewise.
35073 (TARGET_ARC700): Likewise.
35074 (TARGET_EM): Likewise.
35075 (TARGET_HS): Likewise.
35076 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
35079 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
35082 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
35083 condition/operands for integer GE/LE/GEU/LEU operations.
35085 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
35087 PR translation/79397
35088 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
35091 2017-02-08 Martin Jambor <mjambor@suse.cz>
35094 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
35095 whether allocation happened.
35096 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
35097 nothing was allocated.
35099 2017-02-08 Jakub Jelinek <jakub@redhat.com>
35101 PR tree-optimization/79408
35102 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
35103 constant, but SSA_NAME with a known integer range, use the minimum
35104 of that range instead of op1 to determine if modulo can be replaced
35105 with its first operand.
35107 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35109 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
35111 2017-02-08 Richard Biener <rguenther@suse.de>
35113 PR tree-optimization/71824
35114 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35115 Check all loops contained in the merged region.
35117 2017-02-07 Andrew Pinski <apinski@cavium.com>
35119 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
35121 2017-02-07 Andrew Pinski <apinski@cavium.com>
35123 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
35124 (thunderxt88): Likewise.
35125 (thunderxt81): Disable LSE and change v8.1 to v8.
35126 (thunderxt83): Likewise.
35128 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35129 Richard Biener <rguenther@suse.de>
35131 PR middle-end/79399
35132 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
35133 type from int to size_t.
35134 * ira-costs.c (struct_costs_size): Change type from int to size_t.
35136 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35138 PR rtl-optimization/79386
35139 * cprop.c (bypass_conditional_jumps): Initialize
35140 bypass_last_basic_block already before splitting bbs after
35141 unconditional traps...
35142 (bypass_conditional_jumps): ... rather than here.
35145 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
35146 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
35147 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
35148 fix -masm=intel patterns.
35150 2017-02-07 Richard Biener <rguenther@suse.de>
35152 PR tree-optimization/79256
35153 PR middle-end/79278
35154 * builtins.c (get_object_alignment_2): Use min_align_of_type
35155 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
35156 and ADJUST_FIELD_ALIGN.
35158 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
35160 * doc/tm.texi: Regenerate.
35161 * stor-layout.c (layout_decl): Adjust.
35162 (update_alignment_for_field): Likewise.
35163 (place_field): Likewise.
35164 (min_align_of_type): Likewise.
35165 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
35166 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
35167 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
35168 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
35169 * config/frv/frv.c (frv_adjust_field_align): Likewise.
35170 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
35171 * config/i386/i386.c (x86_field_alignment): Likewise.
35172 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
35173 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
35174 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
35175 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
35176 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
35177 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
35181 2017-01-30 Richard Biener <rguenther@suse.de>
35183 PR tree-optimization/79256
35184 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35185 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35188 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
35190 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
35191 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
35192 builtins to SImode and emit a zero-extend, if necessary.
35194 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35196 * docs/invoke.texi (RISC-V Options): Alphabetize.
35198 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35200 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
35203 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35205 * config/riscv/riscv.c: New file.
35206 * common/config/riscv/riscv-common.c: Likewise.
35207 * config.gcc: Likewise.
35208 * config/riscv/constraints.md: Likewise.
35209 * config/riscv/elf.h: Likewise.
35210 * config/riscv/generic.md: Likewise.
35211 * config/riscv/linux.h: Likewise.
35212 * config/riscv/multilib-generator: Likewise.
35213 * config/riscv/peephole.md: Likewise.
35214 * config/riscv/pic.md: Likewise.
35215 * config/riscv/predicates.md: Likewise.
35216 * config/riscv/riscv-builtins.c: Likewise.
35217 * config/riscv/riscv-c.c: Likewise.
35218 * config/riscv/riscv-ftypes.def: Likewise.
35219 * config/riscv/riscv-modes.def: Likewise.
35220 * config/riscv/riscv-opts.h: Likewise.
35221 * config/riscv/riscv-protos.h: Likewise.
35222 * config/riscv/riscv.h: Likewise.
35223 * config/riscv/riscv.md: Likewise.
35224 * config/riscv/riscv.opt: Likewise.
35225 * config/riscv/sync.md: Likewise.
35226 * config/riscv/t-elf-multilib: Likewise.
35227 * config/riscv/t-linux: Likewise.
35228 * config/riscv/t-linux-multilib: Likewise.
35229 * config/riscv/t-riscv: Likewise.
35230 * configure.ac: Likewise.
35231 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
35232 Waterman as RISC-V maintainers.
35233 * doc/install.texi: Add RISC-V entries.
35234 * doc/invoke.texi: Add RISC-V options section.
35235 * doc/md.texi: Add RISC-V constraints section.
35236 * configure: Regenerated.
35238 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
35241 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
35242 false values to be constant vectors with all 0 or all 1 bits set.
35243 (vcondu<mode><mode>): Likewise.
35244 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
35246 (fpmask_comparison_operator): Update comment.
35247 (vecint_comparison_operator): New predicate.
35248 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
35249 vector conditionals when the true and false values are constant
35250 vectors with all 0 bits or all 1 bits set.
35252 2017-02-06 Martin Sebor <msebor@redhat.com>
35254 PR tree-optimization/79376
35255 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
35257 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
35259 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
35260 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
35261 to simplify split condition.
35263 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35265 * omp-expand.c (oxpand_omp_atomic_fetch_op,
35266 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
35269 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
35271 PR rtl-optimization/68664
35272 * target.def (can_speculate_insn): New hook.
35273 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
35274 * doc/tm.texi: Regenerate.
35275 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
35276 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
35277 (rs6000_sched_can_speculate_insn): New function.
35279 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35281 PR tree-optimization/79284
35282 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
35283 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
35284 vectorizable_mask_load_store, vectorizable_operation,
35285 vect_is_simple_cond, get_same_sized_vectype): Use it instead
35286 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
35287 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
35288 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
35289 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
35290 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
35291 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
35292 is_gimple_assign (stmt). Replace another such test with
35293 is_gimple_assign (stmt).
35295 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35298 * config/avr/avr.c (rtl-iter.h): Include it.
35299 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
35300 (avr_legitimate_combined_insn): ...and implementation.
35302 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
35304 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
35305 * config/s390/s390.c (s390_const_operand_ok)
35306 (s390_canonicalize_comparison, s390_extract_part)
35307 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
35308 (s390_contiguous_bitmask_p, s390_rtx_costs)
35309 (legitimize_pic_address): Likewise.
35310 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
35311 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
35312 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
35313 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
35314 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
35316 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35318 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
35319 REGNO($0) == REGNO($1).
35321 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35323 * config/s390/linux.h(SIZE_TYPE): Add comment.
35325 2017-02-06 Julian Brown <julian@codesourcery.com>
35326 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35327 Virendra Pathak <virendra.pathak@broadcom.com>
35329 * config/aarch64/aarch64-cores.def: Change the scheduler
35331 * config/aarch64/aarch64.md: Include thunderx2t99.md.
35332 * config/aarch64/thunderx2t99.md: New file.
35334 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35336 * doc/standards.texi (Go Language): Update link to language
35339 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35341 * tree-eh.c (lower_resx): Sanitize profile.
35342 (cleanup_empty_eh_move_lp): Likewise.
35344 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35347 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
35349 * cfgloopmanip.h (loop_version): Update prototype.
35350 * modulo-sched.c (sms_schedule): Update call of loop_version.
35351 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
35352 * tree-parloops.c (gen_parallel_loop): Likewise.
35353 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
35354 * tree-ssa-loop-split.c (split_loop): Likewise.
35355 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
35356 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
35358 2017-02-05 Martin Liska <mliska@suse.cz>
35361 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
35363 (print_operand_address): Initialize a struct to zero.
35365 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35367 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
35368 garbage collector only in textual form.
35370 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35372 * doc/extend.texi (x86 specific memory model extensions for
35373 transactional memory): Simplify a phrase.
35375 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
35378 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
35379 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
35380 (atomic_storedi_1): Likewise.
35382 2017-02-04 Jakub Jelinek <jakub@redhat.com>
35384 PR tree-optimization/79338
35385 * tree-parloops.c (gather_scalar_reductions): Don't call
35386 vect_analyze_loop_form for loop->inner before destroying loop's
35389 2017-02-03 Martin Sebor <msebor@redhat.com>
35391 PR tree-optimization/79327
35392 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
35393 when precision has resulted in leading zeros.
35394 (format_integer): Adjust the likely counter to assume an unknown
35395 argument that may be zero is non-zero.
35397 2017-02-03 Jason Merrill <jason@redhat.com>
35400 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
35401 avoid copying non-taken branch.
35403 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35405 PR tree-optimization/79340
35406 * tree-vect-loop.c (vectorizable_reduction): Release
35407 vec_defs elements after safe_splicing them into other vectors.
35410 PR tree-optimization/79327
35411 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
35412 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
35414 (format_integer): Use wide_int_to_tree instead of build_int_cst
35415 + to_?hwi. If argmin is NULL, just set argmin and argmax to
35416 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
35417 of shortest and longest sequence.
35419 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
35421 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
35422 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
35424 2017-02-03 Walter Lee <walt@tilera.com>
35427 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
35428 after initial stackframe link reg save.
35429 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
35431 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35434 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
35435 wu for stxssp alternative.
35437 2017-02-03 Martin Sebor <msebor@redhat.com>
35439 PR tree-optimization/79352
35440 * gimple-fold.c (get_range_strlen): Add argument.
35441 (get_range_strlen): Change return type to bool.
35442 (get_maxval_strlen): Pass in a dummy argument.
35443 * gimple-fold.h (get_range_strlen): Change return type to bool.
35444 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
35445 * tree.h (array_at_struct_end_p): Add argument.
35446 * tree.c (array_at_struct_end_p): Handle it.
35448 2017-02-03 Martin Liska <mliska@suse.cz>
35451 * multiple_target.c (create_dispatcher_calls): Redirect edge
35452 from a caller of a dispatcher.
35453 (expand_target_clones): Make the clones local.
35454 (ipa_target_clone): Do both target clones and resolvers.
35455 (ipa_dispatcher_calls): Remove the pass.
35456 (pass_dispatcher_calls::gate): Likewise.
35457 (make_pass_dispatcher_calls): Likewise.
35458 * passes.def (pass_target_clone): Put as very first IPA early
35461 2017-02-03 Martin Liska <mliska@suse.cz>
35463 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
35464 in case of a function with ifunc attribute.
35466 2017-02-03 Martin Liska <mliska@suse.cz>
35468 * cgraph.c (cgraph_node::dump): Dump function version info.
35469 * symtab.c (symtab_node::dump_base): Add missing new line.
35471 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
35473 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
35474 (ifcombine_ifandif): Use it.
35476 2017-02-03 Martin Liska <mliska@suse.cz>
35478 * doc/invoke.texi: Document default value for
35479 use-after-scope-direct-emission-threshold.
35481 2017-02-03 Martin Liska <mliska@suse.cz>
35483 PR tree-optimization/79339
35484 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
35485 (format_floating): Likewise.
35487 2017-02-03 Martin Liska <mliska@suse.cz>
35490 * ipa-prop.c (ipa_node_params_t::insert): Remove current
35492 (ipa_node_params_t::remove): Likewise.
35493 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
35494 initialization from removed ipa_node_params_t::insert.
35495 (ipa_node_params::~ipa_node_params): Move from removed
35496 ipa_node_params_t::release.
35497 * symbol-summary.h (symbol_summary::m_released): New member.
35498 Do not release a summary twice. Do not allow to call finalizer
35499 for types of a summary that live in GGC memory.
35501 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35503 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
35506 2017-02-02 Martin Sebor <msebor@redhat.com>
35508 PR middle-end/79275
35509 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
35510 (format_string): Tighten up the range of output for non-constant
35511 strings and correct the expected range for wide non-constant strings.
35513 2017-02-02 Martin Sebor <msebor@redhat.com>
35515 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
35517 PR middle-end/32003
35518 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
35520 (-fdump-tree-@var): Add to index and document how to come up
35521 with pass-specific option and dump file names.
35522 (-fdump-passes): Clarify where to look for output.
35524 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
35526 PR middle-end/77445
35527 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
35528 statistics of the analyzed path; allow threading for speed when
35529 any of BBs along the path are optimized for speed.
35531 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
35533 PR middle-end/78468
35534 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
35535 settings of the virtual registers.
35538 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
35540 * explow.c (get_dynamic_stack_size): Take known alignment of stack
35541 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
35544 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35546 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
35547 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
35549 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35551 * config/s390/s390.md: Add missing comments with the expanded
35553 * config/s390/vector.md: Likewise.
35554 * config/s390/vx-builtins.md: Likewise.
35556 2017-02-02 Jakub Jelinek <jakub@redhat.com>
35559 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
35560 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
35561 conditions on a single line.
35563 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35565 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35566 __S390_VX__ to __VX__.
35568 2017-02-01 Andrew Pinski <apinski@cavium.com>
35570 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
35571 stmt_info to record_stmt_cost.
35572 (vect_get_known_peeling_cost): Pass stmt_info if known to
35574 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
35575 cpu_vector_cost field into
35576 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
35577 field into vec_int_stmt_cost and vec_fp_stmt_cost.
35578 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
35579 splitting of scalar_stmt_cost and vec_stmt_cost.
35580 (thunderx_vector_cost): Likewise.
35581 (cortexa57_vector_cost): LIkewise.
35582 (exynosm1_vector_cost): Likewise.
35583 (xgene1_vector_cost): Likewise.
35584 (thunderx2t99_vector_cost): Improve after the splitting of the two
35586 (aarch64_builtin_vectorization_cost): Update for the splitting of
35587 scalar_stmt_cost and vec_stmt_cost.
35589 2017-02-01 Torvald Riegel <triegel@redhat.com>
35590 Richard Henderson <rth@redhat.com>
35592 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
35593 conditional on existance of a fast atomic load.
35594 * optabs-query.c (can_atomic_load_p): New function.
35595 * optabs-query.h (can_atomic_load_p): Declare it.
35596 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
35597 no fast atomic load is available for the particular size of access.
35598 (expand_atomic_compare_and_swap): Likewise.
35599 (expand_atomic_load): Likewise.
35600 (expand_atomic_store): Likewise.
35601 (expand_atomic_fetch_op): Likewise.
35602 * testsuite/lib/target-supports.exp
35603 (check_effective_target_sync_int_128): Remove x86 because it provides
35604 no fast atomic load.
35605 (check_effective_target_sync_int_128_runtime): Likewise.
35607 2017-02-01 Richard Biener <rguenther@suse.de>
35609 * graphite.c: Include tree-vectorizer.h for find_loop_location.
35610 (graphite_transform_loops): Provide opt-info for optimized nests.
35611 * tree-parloop.c (parallelize_loops): Provide opt-info for
35612 parallelized loops.
35614 2017-02-01 Richard Biener <rguenther@suse.de>
35616 PR middle-end/79315
35617 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
35618 was not set before.
35620 2017-02-01 Richard Biener <rguenther@suse.de>
35622 PR tree-optimization/71824
35623 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35624 Verify the loops are valid in the merged SESE region.
35625 (scop_detection::can_represent_loop_1): Check analyzing the
35626 evolution of the number of iterations in the region succeeds.
35628 2017-01-31 Ian Lance Taylor <iant@golang.org>
35630 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
35631 REG_ARGS_SIZE note to 32-bit push insns and call insn.
35633 2017-01-31 David Malcolm <dmalcolm@redhat.com>
35635 PR preprocessor/79210
35636 * input.c (get_substring_ranges_for_loc): Replace line_width
35637 assertion with error-handling.
35639 2017-01-31 Richard Biener <rguenther@suse.de>
35641 PR tree-optimization/77318
35642 * graphite-sese-to-poly.c (extract_affine): Fix assert.
35643 (create_pw_aff_from_tree): Take loop parameter.
35644 (add_condition_to_pbb): Pass loop of the condition to
35645 create_pw_aff_from_tree.
35647 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35649 * config/s390/s390.c (s390_asan_shadow_offset): New function.
35650 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
35652 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
35656 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
35658 (convert_int_to_float128): Likewise.
35659 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
35660 (convert_int_to_float128): Likewise.
35661 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
35662 (UNSPEC_IEEE128_CONVERT): Likewise.
35663 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
35664 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
35665 Use local variables for IBM extended format.
35666 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
35667 (fix_trunc<mode>si2_fprs): Likewise.
35668 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
35669 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
35670 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
35671 to know that we can now have integers of all sizes in vector
35673 (fix<uns>_<mode>di2_hw): Likewise.
35674 (float<uns>_<mode>si2_hw): Likewise.
35675 (fix_<mode>si2_hw): Likewise.
35676 (fixuns_<mode>si2_hw): Likewise.
35677 (float<uns>_<mode>di2_hw): Likewise.
35678 (float_<mode>di2_hw): Likewise.
35679 (float_<mode>si2_hw): Likewise.
35680 (floatuns_<mode>di2_hw): Likewise.
35681 (floatuns_<mode>si2_hw): Likewise.
35682 (xscvqp<su>wz_<mode>): Delete, no longer used.
35683 (xscvqp<su>dz_<mode>): Likewise.
35684 (xscv<su>dqp_<mode>): Likewise.
35685 (ieee128_mfvsrd_64bit): Likewise.
35686 (ieee128_mfvsrd_32bit): Likewise.
35687 (ieee128_mfvsrwz): Likewise.
35688 (ieee128_mtvsrw): Likewise.
35689 (ieee128_mtvsrd_64bit): Likewise.
35690 (ieee128_mtvsrd_32bit): Likewise.
35692 2017-01-31 Martin Liska <mliska@suse.cz>
35695 * ipa-prop.c (ipa_free_all_node_params): Call release method
35696 instead of ~sumbol_summary to not to trigger double times
35699 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
35701 PR tree-optimization/71691
35702 * bitmap.h (class auto_bitmap): New.
35703 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
35704 is_maybe_undefined instead of ssa_undefined_value_p.
35706 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35708 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35709 __S390_ARCH_LEVEL__ to __ARCH__.
35711 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35713 PR tree-optimization/79267
35714 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
35715 if should_remove_lhs_p is true.
35717 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
35720 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
35721 (add_alignment_attribute): New.
35722 (base_type_die): Add alignment attribute.
35723 (subrange_type_die): Likewise.
35724 (modified_type_die): Likewise.
35725 (gen_array_type_die): Likewise.
35726 (gen_descr_array_type_die: Likewise.
35727 (gen_enumeration_type_die): Likewise.
35728 (gen_subprogram_die): Likewise.
35729 (gen_variable_die): Likewise.
35730 (gen_field_die): Likewise.
35731 (gen_ptr_to_mbr_type_die): Likewise.
35732 (gen_struct_or_union_type_die): Likewise.
35733 (gen_subroutine_type_die): Likewise.
35734 (gen_typedef_die): Likewise.
35735 (base_type_cmp): Compare alignment attribute.
35737 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35740 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
35741 (setb_unsigned) New pattern for setb with CCUNS.
35742 * config/rs6000/rs6000.c (expand_block_compare): Use a different
35743 subfc./subfe sequence to avoid overflow problems. Generate a
35744 shorter sequence with cmpld/setb for power9.
35745 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
35746 for generating subfc. instruction.
35747 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
35748 now uses this instruction.
35750 2017-01-30 Ian Lance Taylor <iant@google.com>
35753 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
35754 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
35756 2017-01-30 Martin Sebor <msebor@redhat.com>
35758 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
35759 Move constant to the right of a relational operator.
35760 (get_mpfr_format_length, format_character, format_string): Ditto.
35761 (should_warn_p, maybe_warn): Same.
35763 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
35765 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
35768 * asan.c (get_translation_unit_decl): Remove function.
35769 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
35771 2017-01-30 Martin Liska <mliska@suse.cz>
35773 PR gcov-profile/79259
35774 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
35775 -fprofile-generate.
35777 2017-01-30 Martin Liska <mliska@suse.cz>
35780 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
35781 Initialize variables with NULL value.
35783 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
35786 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
35788 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
35790 2017-01-30 Richard Biener <rguenther@suse.de>
35792 PR tree-optimization/79276
35793 * tree-vrp.c (process_assert_insertions): Properly adjust common
35794 when removing a duplicate.
35796 2017-01-30 Richard Biener <rguenther@suse.de>
35798 PR tree-optimization/79256
35799 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35800 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35802 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
35804 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
35807 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
35808 ("*r<noxa>sbg_<mode>_sll_bitmask")
35809 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
35810 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
35811 Use contiguous_bitmask_nowrap_operand.
35813 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35816 * config/rs6000/altivec.h (vec_xl): Revise #define.
35817 (vec_xst): Likewise.
35819 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
35821 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
35823 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
35825 PR rtl-optimization/79194
35826 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
35827 traps before call to bypass_conditional_jumps.
35829 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35831 PR tree-optimization/71374
35832 * lra-constraints.c (check_conflict_input_operands): New.
35833 (match_reload): Use it.
35835 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35838 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
35839 account to calculate conflict_set.
35841 2017-01-27 Bin Cheng <bin.cheng@arm.com>
35843 PR rtl-optimization/78559
35844 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
35845 other_insn in combine.
35847 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
35849 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
35850 uint16_type_node for BT_UINT16.
35852 2017-01-27 David Malcolm <dmalcolm@redhat.com>
35854 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
35855 "RTL Tests" to menu.
35856 (GIMPLE Tests): New node.
35857 (RTL Tests): New node.
35859 2017-01-27 Richard Biener <rguenther@suse.de>
35861 PR tree-optimization/79245
35862 * tree-loop-distribution.c (distribute_loop): Apply cost
35863 modeling also to detected patterns.
35865 2017-01-27 Richard Biener <rguenther@suse.de>
35867 PR tree-optimization/71433
35868 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
35869 (compare_assert_loc): New function.
35870 (process_assert_insertions): Sort and optimize assert locations
35871 to remove duplicates and push down identical assertions on
35872 edges to their destination block.
35874 2017-01-27 Richard Biener <rguenther@suse.de>
35876 PR tree-optimization/79244
35877 * tree-vrp.c (remove_range_assertions): Forcefully propagate
35878 out SSA names even if abnormal.
35880 2017-01-27 Jakub Jelinek <jakub@redhat.com>
35882 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
35883 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
35884 instead of MPFR_RNDN.
35886 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
35889 * arm.c (arm_option_override): Don't call build_target_option_node
35890 until after doing all option overrides.
35891 (arm_valid_target_attribute_tree): Likewise.
35893 2017-01-27 Martin Liska <mliska@suse.cz>
35895 * doc/invoke.texi (-fprofile-arcs): Document profiling support
35896 for {cd}tors and C++ {cd}tors.
35898 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
35900 * config/s390/s390.md ("*setmem_long_and")
35901 ("*setmem_long_and_31z"): Use zero_extend instead of and.
35903 2017-01-26 Martin Sebor <msebor@redhat.com>
35905 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
35908 2017-01-26 Martin Sebor <msebor@redhat.com>
35910 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
35911 HAVE_DFmode before using XFmode or DFmode.
35912 (parse_directive): Avoid using the z length modifier to avoid
35913 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
35915 PR middle-end/78703
35916 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
35917 to accept adjustment as an array.
35918 (get_int_range): New function.
35919 (struct directive): Make width and prec arrays.
35920 (directive::set_width, directive::set_precision): Call get_int_range.
35921 (format_integer, format_floating): Handle width and precision ranges.
35922 (format_string, parse_directive): Same.
35924 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35927 * dwarf2out.c (generate_skeleton_bottom_up): For children with
35928 comdat_type_p set, just clone them, but keep the children in the
35932 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
35933 which have direct callers with -fvar-tracking-assignments enabled
35935 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
35936 inside of type units.
35938 2017-01-26 Martin Sebor <msebor@redhat.com>
35940 PR middle-end/78703
35941 * gimple-ssa-sprintf.c (struct result_range): Add likely and
35943 (struct format_result): Replace number_chars, number_chars_min,
35944 and number_chars_max with a single member of struct result_range.
35946 (format_result::operator+=): Adjust.
35947 (struct fmtresult): Remove bounded. Handle likely and unlikely
35949 (fmtresult::adjust_for_width_or_precision): New function.
35950 (fmtresult:type_max_digits): New function.
35951 (bytes_remaining): Handle likely and unlikely counters.
35952 (min_bytes_remaining): Remove.
35953 (format_percent): Simplify.
35954 (format_integer, format_floating): Set likely and unlikely counters.
35955 (get_string_length, format_character, format_string): Same.
35956 (format_plain, should_warn_p): New function.
35957 (maybe_warn): Call should_warn_p. Update diagnostic messages
35958 and handle those for all directives, including plain strings.
35959 (format_directive): Handle likely and unlikely counters.
35960 Remove unnecessary quoting from diagnostics. Add an informational
35962 (add_bytes): Remove.
35963 (pass_sprintf_length::compute_format_length): Simplify.
35964 (try_substitute_return_value): Handle likely and unlikely counters.
35966 2017-01-26 Carl Love <cel@us.ibm.com>
35968 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
35969 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
35971 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
35974 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
35975 endianess for subregs into account.
35976 * lra-constraints.c (lra_constraints): Do risky transformations
35977 always on the first iteration.
35978 * lra-lives.c (check_pseudos_live_through_calls): Add arg
35979 last_call_used_reg_set.
35980 (process_bb_lives): Define and use last_call_used_reg_set.
35981 * lra.c (lra): Always continue after lra_constraints on the first
35984 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
35986 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
35988 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
35990 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35992 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
35993 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
35994 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
35995 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
35996 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
35997 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
35998 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
35999 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
36000 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
36002 2017-01-26 Marek Polacek <polacek@redhat.com>
36005 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
36006 for the third operand.
36008 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36010 PR middle-end/79236
36011 * omp-low.c (struct omp_context): Add simt_stmt field.
36012 (scan_omp_for): Return omp_context *.
36013 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
36014 context to the _simt_ SIMD stmt.
36015 (lower_omp_for): For combined SIMD with sibling _simt_
36016 SIMD, make sure to use the same decls in _looptemp_
36017 clauses as in the sibling.
36019 2017-01-26 David Sherwood <david.sherwood@arm.com>
36021 PR middle-end/79212
36022 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
36025 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36028 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
36029 emit fld b; fld a; if possible.
36031 * brig-builtins.def: Update copyright years.
36032 * config/arm/arm_acle_builtins.def: Update copyright years.
36034 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
36037 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
36038 constraint instead of o for the stxsd instruction.
36040 2017-01-25 Carl Love <cel@us.ibm.com>
36042 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
36043 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
36045 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
36047 * doc/invoke.texi (C++ Dialect Options): Fix typo.
36049 2017-01-25 Richard Biener <rguenther@suse.de>
36051 PR tree-optimization/69264
36052 * target.def (vector_alignment_reachable): Improve documentation.
36053 * doc/tm.texi: Regenerate.
36054 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
36056 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
36057 earlier changes with respect to TYPE_USER_ALIGN.
36058 (vector_alignment_reachable_p): Likewise. Improve dumping.
36060 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36063 * config/arm/arm.md (xordi3): Force constant operand into a register
36066 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36068 * doc/invoke.texi (-fstore-merging): Correct default optimization
36069 levels at which it is enabled.
36070 (-O): Move -fstore-merging from list to...
36073 2017-01-25 Richard Biener <rguenther@suse.de>
36076 * omp-expand.c: Include debug.h.
36077 (expand_omp_taskreg): Make sure to generate early debug before
36078 outlining anything from a function.
36079 (expand_omp_target): Likewise.
36080 (grid_expand_target_grid_body): Likewise.
36082 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
36085 * asan.c (get_translation_unit_decl): New function.
36086 (asan_add_global): Extract modules file name from globals
36087 TRANSLATION_UNIT_DECL name.
36089 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
36092 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
36093 for long calls with APCS frame and VFP.
36095 2017-01-24 David Malcolm <dmalcolm@redhat.com>
36097 * cfg.c (original_copy_tables_initialized_p): New function.
36098 * cfg.h (original_copy_tables_initialized_p): New decl.
36099 * cfgrtl.c (relink_block_chain): Guard the call to
36100 free_original_copy_tables with a call to
36101 original_copy_tables_initialized_p.
36102 * cgraph.h (symtab_node::native_rtl_p): New decl.
36103 * cgraphunit.c (symtab_node::native_rtl_p): New function.
36104 (symtab_node::needed_p): Don't assert for early assembly output
36105 for __RTL functions.
36106 (cgraph_node::finalize_function): Set "force_output" for __RTL
36108 (cgraph_node::analyze): Bail out early for __RTL functions.
36109 (analyze_functions): Update assertion to support __RTL functions.
36110 (cgraph_node::expand): Bail out early for __RTL functions.
36111 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
36113 * function.h (struct function): Update comment for field
36115 * gimple-expr.c: Include "tree-pass.h".
36116 (gimple_has_body_p): Return false for __RTL functions.
36117 * Makefile.in (OBJS): Add run-rtl-passes.o.
36118 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
36120 (gcc::pass_manager::get_clean_slate): New accessor.
36121 * passes.c: Include "insn-addr.h".
36122 (should_skip_pass_p): Add logging. Update logic for running
36123 "expand" to be compatible with both __GIMPLE and __RTL. Guard
36124 property-provider override so it is only done for gimple passes.
36126 (skip_pass): New function.
36127 (execute_one_pass): Call skip_pass when skipping passes.
36128 * read-md.c (md_reader::read_char): Support filtering
36129 the input to a subset of line numbers.
36130 (md_reader::md_reader): Initialize fields
36131 m_first_line and m_last_line.
36132 (md_reader::read_file_fragment): New function.
36133 * read-md.h (md_reader::read_file_fragment): New decl.
36134 (md_reader::m_first_line): New field.
36135 (md_reader::m_last_line): New field.
36136 * read-rtl-function.c (function_reader::create_function): Only
36137 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
36138 curr_properties. Set DECL_INITIAL to a dummy block.
36139 (read_rtl_function_body_from_file_range): New function.
36140 * read-rtl-function.h (read_rtl_function_body_from_file_range):
36142 * run-rtl-passes.c: New file.
36143 * run-rtl-passes.h: New file.
36145 2017-01-24 Jeff Law <law@redhat.com>
36147 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
36150 2017-01-24 Bin Cheng <bin.cheng@arm.com>
36152 PR tree-optimization/79159
36153 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
36154 (record_nonwrapping_iv): Improve boundary using above function if no
36155 value range information.
36157 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
36158 Martin Jambor <mjambor@suse.cz>
36160 * brig-builtins.def: New file.
36161 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
36162 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
36163 (DEF_HSAIL_SAT_BUILTIN): Likewise.
36164 (DEF_HSAIL_INTR_BUILTIN): Likewise.
36165 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
36166 * builtin-types.def (BT_INT8): New.
36167 (BT_INT16): Likewise.
36168 (BT_UINT8): Likewise.
36169 (BT_UINT16): Likewise.
36170 (BT_FN_ULONG): Likewise.
36171 (BT_FN_UINT_INT): Likewise.
36172 (BT_FN_UINT_ULONG): Likewise.
36173 (BT_FN_UINT_LONG): Likewise.
36174 (BT_FN_UINT_PTR): Likewise.
36175 (BT_FN_ULONG_PTR): Likewise.
36176 (BT_FN_INT8_FLOAT): Likewise.
36177 (BT_FN_INT16_FLOAT): Likewise.
36178 (BT_FN_UINT32_FLOAT): Likewise.
36179 (BT_FN_UINT16_FLOAT): Likewise.
36180 (BT_FN_UINT8_FLOAT): Likewise.
36181 (BT_FN_UINT64_FLOAT): Likewise.
36182 (BT_FN_UINT16_UINT32): Likewise.
36183 (BT_FN_UINT32_UINT16): Likewise.
36184 (BT_FN_UINT16_UINT16_UINT16): Likewise.
36185 (BT_FN_INT_PTR_INT): Likewise.
36186 (BT_FN_UINT_PTR_UINT): Likewise.
36187 (BT_FN_LONG_PTR_LONG): Likewise.
36188 (BT_FN_ULONG_PTR_ULONG): Likewise.
36189 (BT_FN_VOID_UINT64_UINT64): Likewise.
36190 (BT_FN_UINT8_UINT8_UINT8): Likewise.
36191 (BT_FN_INT8_INT8_INT8): Likewise.
36192 (BT_FN_INT16_INT16_INT16): Likewise.
36193 (BT_FN_INT_INT_INT): Likewise.
36194 (BT_FN_UINT_FLOAT_UINT): Likewise.
36195 (BT_FN_FLOAT_UINT_UINT): Likewise.
36196 (BT_FN_ULONG_UINT_UINT): Likewise.
36197 (BT_FN_ULONG_UINT_PTR): Likewise.
36198 (BT_FN_ULONG_ULONG_ULONG): Likewise.
36199 (BT_FN_UINT_UINT_UINT): Likewise.
36200 (BT_FN_VOID_UINT_PTR): Likewise.
36201 (BT_FN_UINT_UINT_PTR: Likewise.
36202 (BT_FN_UINT32_UINT64_PTR): Likewise.
36203 (BT_FN_INT_INT_UINT_UINT): Likewise.
36204 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
36205 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
36206 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
36207 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
36208 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
36209 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
36210 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
36211 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
36212 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
36213 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
36214 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
36215 * doc/frontends.texi: List BRIG FE.
36216 * doc/install.texi (Testing): Add BRIG tesring requirements.
36217 * doc/invoke.texi (Overall Options): Mention BRIG.
36218 * doc/standards.texi (Standards): Doucment BRIG HSA version.
36220 2017-01-24 Richard Biener <rguenther@suse.de>
36222 PR translation/79208
36223 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
36225 2017-01-24 Martin Jambor <mjambor@suse.cz>
36228 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
36229 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
36230 and known_contexts.
36232 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
36234 PR middle-end/79123
36235 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
36236 casts from signed to unsigned really don't have a range.
36238 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
36240 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
36241 GMP_RNDx for compatiblity.
36243 2017-01-24 Martin Liska <mliska@suse.cz>
36246 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
36247 that would prevent us to call alloca with -1 as argument.
36249 2017-01-24 Jakub Jelinek <jakub@redhat.com>
36251 * dwarf2out.c (output_compilation_unit_header, output_file_names):
36252 Avoid -Wformat-security warning.
36254 2017-01-23 Andrew Pinski <apinski@cavium.com>
36256 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
36259 2017-01-23 Martin Sebor <msebor@redhat.com>
36261 PR middle-end/78703
36262 * gimple-ssa-sprintf.c (warn_level): New global.
36263 (format_integer): Use it here and throughout the rest of the file.
36264 Use the same switch to compute sign as base.
36265 (maybe_warn): New function.
36266 (format_directive): Factor out warnings into maybe_warn.
36267 Add debugging output. Use warn_level.
36268 (add_bytes): Use warn_level.
36269 (pass_sprintf_length::compute_format_length): Add debugging output.
36270 (try_substitute_return_value): Same.
36271 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
36273 PR middle-end/78703
36274 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
36275 (struct fmtresult, format_integer, format_floating): Adjust.
36276 (fmtresult::fmtresult): Set max correctly in two argument ctor.
36277 (get_string_length, format_string,format_directive): Same.
36278 (pass_sprintf_length::compute_format_length): Same.
36279 (try_substitute_return_value): Simplify slightly.
36281 PR middle-end/78703
36282 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
36283 (fmtresult::operator+=): Outlined.
36284 (struct fmtresult): Add ctors.
36285 (struct conversion_spec): Rename...
36286 (struct directive): ...to this. Add and remove data members.
36287 (directive::set_width, directive::set_precision): New functions.
36288 (format_percent): Use fmtresult ctor.
36289 (get_width_and_precision): Remove.
36290 (format_integer): Make naming changes. Avoid computing width and
36292 (format_floating): Same. Adjust indentation.
36293 (format_character, format_none): New functions.
36294 (format_string): Moved character handling to format_character.
36295 (format_directive): Remove arguments, change return type.
36296 (parse_directive): New function.
36297 (pass_sprintf_length::compute_format_length): Move directive
36298 parsing to parse_directive.
36300 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36302 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
36303 (assign_assembler_name_if_needed): ... this.
36304 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
36305 (assign_assembler_name_if_needed): ... this.
36306 (free_lang_data_in_cgraph): Adjust callers.
36307 * cgraphunit.c (cgraph_node::analyze): Likewise.
36308 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
36311 2017-01-23 Richard Biener <rguenther@suse.de>
36313 PR tree-optimization/79088
36314 PR tree-optimization/79188
36315 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
36316 resetting loop bounds after last path deletion. Reset loop
36317 bounds of the target loop, make code match the comments.
36318 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
36319 Make sure loops need no fixups.
36321 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
36323 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
36324 exponent support with double type for first argument.
36325 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
36326 type returned by __builtin_vec_extract_sig,
36327 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
36328 functions from "vector int" to "vector unsigned int" or from
36329 "vector long long int" to "vector unsigned long long int".
36330 Changed type returned by __builtin_vec_extract_exp,
36331 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
36332 functions from "vector int" to "vector unsigned int" or from
36333 "vector long long int" to "vector unsigned long long int".
36334 Changed return type of __builtin_vec_test_data_class,
36335 __builtin_vec_test_data_class_sp, and
36336 __builtin_vec_test_data_class_dp from "vector int" to
36337 "vector bool int" or from "vector long long int" to "vector bool
36338 long long int" and changed second argument type from "unsigned
36339 int" to "int". Added new overloaded function forms "vector float
36340 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
36341 "vector float __builtin_vec_insert_exp_sp (vector float, vector
36342 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
36343 double, vector unsigned long long int)" and "vector double
36344 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
36345 long int)". Changed return type of
36346 __builtin_scalar_test_data_class and
36347 __builtin_scalar_test_data_class_sp and
36348 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
36349 int" and changed second argument from "unsigned int" to "int".
36350 Changed type returned by __builtin_scalar_test_neg,
36351 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
36352 from "int" to "bool int". Added new overloaded function form
36353 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
36354 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
36355 exponent double-precision with floating point first argument.
36356 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
36357 documentation of scalar_test_data_class, scalar_test_neg,
36358 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
36359 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
36360 vec_test_data_class built-in functions to reflect refinements in
36361 their type signatures.
36363 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
36365 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
36367 (aarch64_elf_asm_destructor): Likewise.
36369 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
36371 PR rtl-optimization/78634
36372 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
36373 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
36374 * ifcvt.c (noce_try_cmove): Add missing cost check.
36376 PR rtl-optimization/71724
36377 * combine.c (if_then_else_cond): Look for situations where it is
36378 beneficial to undo the work of one of the recursive calls.
36380 2017-01-23 Bin Cheng <bin.cheng@arm.com>
36382 PR tree-optimization/70754
36383 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
36384 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
36385 combined stmt before it if not NULL.
36386 (combine_chains): Process refs reversely and compute dominance point
36389 2017-01-23 Martin Liska <mliska@suse.cz>
36391 PR tree-optimization/79196
36392 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
36393 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
36395 (strlen_optimize_stmt): Call the renamed function.
36397 2017-01-23 Michael Matz <matz@suse.de>
36399 PR tree-optimization/78384
36400 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
36402 2017-01-23 Richard Biener <rguenther@suse.de>
36404 PR tree-optimization/79186
36405 * tree-vrp.c (register_new_assert_for): Make sure we've seen
36406 both incoming edges before moving an assert.
36408 2017-01-23 Martin Jambor <mjambor@suse.cz>
36410 * ipa-prop.c (load_from_param_1): Removed.
36411 (load_from_unmodified_param): Bits from load_from_param_1 put back
36413 (load_from_param): Removed.
36414 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
36415 with stmt. Reverted back to use of load_from_unmodified_param.
36417 2017-01-23 Martin Jambor <mjambor@suse.cz>
36420 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
36421 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
36422 field a pointer to garbage collected vector, mark lattices and
36423 ipcp_orig_node with GTY((skip)).
36424 (ipa_get_param_count): Adjust to descriptors being a pointer.
36425 (ipa_get_param): Likewise.
36426 (ipa_get_type): Likewise.
36427 (ipa_get_param_move_cost): Likewise.
36428 (ipa_set_param_used): Likewise.
36429 (ipa_get_controlled_uses): Likewise.
36430 (ipa_set_controlled_uses): Likewise.
36431 (ipa_is_param_used): Likewise.
36432 (ipa_node_params_t): Move into garbage collector. New methods insert
36434 (ipa_node_params_sum): Annotate wth GTY(()).
36435 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
36437 (ipa_load_from_parm_agg): Adjust declaration.
36438 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
36439 * ipa-profile.c (ipa_profile): Likewise.
36440 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
36441 (ipa_populate_param_decls): Make descriptors parameter garbage
36443 (ipa_dump_param): Adjust to descriptors being a pointer.
36444 (ipa_alloc_node_params): Likewise.
36445 (ipa_initialize_node_params): Likewise.
36446 (load_from_param_1): Make descriptors parameter garbage collected.
36447 (load_from_unmodified_param): Likewise.
36448 (load_from_param): Likewise.
36449 (ipa_load_from_parm_agg): Likewise.
36450 (ipa_node_params::~ipa_node_params): Removed.
36451 (ipa_free_all_node_params): Remove call to delete operator.
36452 (ipa_node_params_t::insert): New.
36453 (ipa_node_params_t::remove): Likewise.
36454 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
36455 copy known_csts and known_contexts vectors.
36456 (ipa_read_node_info): Adjust to descriptors being a pointer.
36457 (ipcp_modif_dom_walker): Make m_descriptors field garbage
36459 (ipcp_transform_function): Make descriptors variable garbage
36462 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
36464 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
36465 * config/i386/avx512dqintrin.h: Ditto.
36466 * config/i386/avx512fintrin.h: Ditto.
36467 * config/i386/i386.c: Handle new builtins.
36468 * config/i386/i386-builtin.def: Add new builtins.
36469 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
36470 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
36472 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36473 Martin Liska <mliska@suse.cz>
36475 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
36476 * asan.c (asan_expand_poison_ifn): Support stores and use
36477 appropriate ASAN report function.
36478 * internal-fn.c (expand_ASAN_POISON_USE): New function.
36479 * internal-fn.def (ASAN_POISON_USE): Declare.
36480 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
36481 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
36482 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
36483 ASAN_POISON calls w/o LHS.
36484 * tree-ssa.c (execute_update_addresses_taken): Create clobber
36485 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
36486 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
36487 * gimplify.c (asan_poison_variables): Add attribute
36488 use_after_scope_memory to variables that really needs to live
36490 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
36491 having the attribute.
36493 2017-01-23 Martin Liska <mliska@suse.cz>
36495 * asan.c (create_asan_shadow_var): New function.
36496 (asan_expand_poison_ifn): Likewise.
36497 * asan.h (asan_expand_poison_ifn): New declaration.
36498 * internal-fn.c (expand_ASAN_POISON): Likewise.
36499 * internal-fn.def (ASAN_POISON): New builtin.
36500 * sanopt.c (pass_sanopt::execute): Expand
36501 asan_expand_poison_ifn.
36502 * tree-inline.c (copy_decl_for_dup_finish): Make function
36504 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
36505 * tree-ssa.c (is_asan_mark_p): New function.
36506 (execute_update_addresses_taken): Rewrite local variables
36507 (identified just by use-after-scope as addressable) into SSA.
36509 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
36511 * doc/install.texi (Specific): opensource.apple.com uses https
36512 now. Remove trailing slash.
36514 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
36516 * README.Portability: Remove note on an Irix compatibility issue.
36518 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
36520 * gcov.c (INCLUDE_ALGORITHM): Define.
36521 (INCLUDE_VECTOR): Define.
36522 No longer include <vector> and <algorithm> directly.
36524 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
36526 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
36528 * doc/invoke.texi (Code Gen Options): Ditto.
36530 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
36533 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
36535 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
36537 rtl-optimization/79125
36538 * cprop.c (local_cprop_pass): Handle cases where we make an
36539 unconditional trap.
36541 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
36545 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
36546 read from, for big endian.
36548 2017-01-20 Jiong Wang <jiong.wang@arm.com>
36550 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
36551 register pauth builtins for LP64 only.
36553 2017-01-20 Marek Polacek <polacek@redhat.com>
36556 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
36559 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
36561 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
36563 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
36564 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
36565 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
36567 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36570 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
36571 in big-endian mode when they are not a single duplicated value.
36573 2017-01-20 Richard Biener <rguenther@suse.de>
36575 * BASE-VER: Bump to 7.0.1.
36577 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
36579 * omp-low.c (omplow_simd_context): New struct. Use it...
36580 (lower_rec_simd_input_clauses): ...here and...
36581 (lower_rec_input_clauses): ...here to hold common data. Adjust all
36582 references to idx, lane, max_vf, is_simt.
36584 2017-01-20 Graham Markall <graham.markall@embecosm.com>
36586 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
36589 2017-01-20 Martin Jambor <mjambor@suse.cz>
36591 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
36592 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
36594 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
36595 (GTFILES): Rename hsa.c to hsa-common.c.
36596 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
36597 * hsa-dump.c: Likewise.
36598 * hsa-gen.c: Likewise.
36599 * hsa-regalloc.c: Likewise.
36600 * ipa-hsa.c: Likewise.
36601 * omp-expand.c: Likewise.
36602 * omp-low.c: Likewise.
36603 * toplev.c: Likewise.
36605 2017-01-20 Marek Polacek <polacek@redhat.com>
36608 * doc/invoke.texi: Document -Wduplicated-branches.
36609 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
36610 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
36611 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
36612 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
36613 return 0 only when not OEP_LEXICOGRAPHIC.
36614 (fold_build_cleanup_point_expr): Use the expression
36615 location when building CLEANUP_POINT_EXPR.
36616 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
36617 * tree.c (add_expr): Handle error_mark_node.
36619 2017-01-20 Martin Liska <mliska@suse.cz>
36622 * tree-profile.c (init_ic_make_global_vars): Do not call
36624 (gimple_init_gcov_profiler): Likewise.
36626 2017-01-20 Martin Liska <mliska@suse.cz>
36629 * cgraph.h (maybe_create_reference): Remove argument and
36631 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
36633 * ipa-cp.c (create_specialized_node): Likewise.
36634 * symtab.c (symtab_node::maybe_create_reference): Handle
36635 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
36637 2017-01-20 Martin Liska <mliska@suse.cz>
36639 * read-rtl-function.c (function_reader::create_function): Use
36640 build_decl instread of build_decl_stat.
36642 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
36644 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
36645 * config/i386/avx512dqintrin.h: Ditto.
36646 * config/i386/avx512fintrin.h: Ditto.
36647 * config/i386/i386-builtin-types.def: Add new types.
36648 * config/i386/i386.c: Handle new types.
36649 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
36650 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
36651 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
36652 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
36653 (__builtin_ia32_kshiftridi): New.
36654 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
36656 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
36660 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
36661 define to rs6000_init_stack_protect_guard.
36662 (rs6000_init_stack_protect_guard): New function.
36664 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36665 Yunqiang Su <yunqiang.su@imgtec.com>
36667 * config.gcc (supported_defaults): Add madd4.
36668 (with_madd4): Add validation.
36669 (all_defaults): Add madd4.
36670 * config/mips/mips.opt (mmadd4): New option.
36671 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
36673 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
36675 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
36676 (ISA_HAS_FUSED_MADD4): Likewise.
36677 * doc/invoke.texi (-mmadd4): Document the new option.
36678 * doc/install.texi (--with-madd4): Document the new option.
36680 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36682 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
36683 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
36684 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
36685 (aarch64_init_pauth_hint_builtins): New.
36686 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
36687 (aarch64_expand_builtin): Expand new builtins.
36689 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36691 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
36692 * combine-stack-adj.c (no_unhandled_cfa): Handle
36693 REG_CFA_TOGGLE_RA_MANGLE.
36694 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
36695 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
36696 info for return address signing.
36697 (aarch64_expand_epilogue): Likewise.
36699 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36701 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
36702 * config/aarch64/aarch64-protos.h
36703 (aarch64_return_address_signing_enabled): New declaration.
36704 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
36706 (aarch64_expand_prologue): Sign return address before it's pushed onto
36708 (aarch64_expand_epilogue): Authenticate return address fetched from
36710 (aarch64_override_options): Sanity check for ILP32 and ISA level.
36711 (aarch64_attributes): New function attributes for "sign-return-address".
36712 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
36713 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
36714 ("*do_return"): Generate combined instructions according to key index.
36715 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
36716 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
36718 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
36719 * config/aarch64/aarch64.opt (msign-return-address=): New.
36720 * doc/extend.texi (AArch64 Function Attributes): Documents
36721 "sign-return-address=".
36722 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
36724 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36726 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
36727 overall option summary.
36729 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36731 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
36732 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
36733 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
36734 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
36736 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
36738 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
36739 -mpower9-minmax by default for -mcpu=power9.
36740 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
36741 128-bit floating point.
36743 2017-01-20 Alan Modra <amodra@gmail.com>
36745 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
36746 optimizing for size.
36748 2017-01-20 Alan Modra <amodra@gmail.com>
36751 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
36752 for strcmp and strncmp from corresponding builtin decl.
36754 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36756 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
36757 instead of i386/rtems-64.h.
36758 * config/i386/rtems-64.h: Remove.
36760 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36764 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
36766 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
36768 2017-01-19 Tamar Christina <tamar.christina@arm.com>
36770 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
36771 Change int to HOST_WIDE_INT.
36772 * config/aarch64/aarch64-protos.h
36773 (aarch64_simd_gen_const_vector_dup): Likewise.
36774 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
36776 2017-01-19 David Malcolm <dmalcolm@redhat.com>
36778 * langhooks-def.h (lhd_type_for_size): New decl.
36779 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
36780 * langhooks.c (lhd_type_for_size): New function, taken from
36783 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
36785 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
36786 define_bypass for CR latency.
36787 (power9-cracked-alu): Update bypass latency and remove power9-branch.
36788 (power9-alu2): Add define_bypass for CR latency.
36790 (power9-mul): Update insn latency.
36791 (power9-mul-compare): Update insn latency, bypass latency and remove
36794 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36796 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
36798 * config/aarch64/aarch64.md
36799 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
36800 aarch64_nopcrelative_literal_loads.
36801 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
36803 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
36805 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
36806 TARGET_LOONGSON_3A.
36807 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
36809 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
36812 * config.gcc (supported_defaults): Add lxc1-sxc1.
36813 (with_lxc1_sxc1): Add validation.
36814 (all_defaults): Add lxc1-sxc1.
36815 * config/mips/mips.opt (mlxc1-sxc1): New option.
36816 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
36818 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
36819 __mips_no_lxc1_sxc1.
36820 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
36821 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
36822 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
36824 2017-01-19 Richard Biener <rguenther@suse.de>
36826 PR tree-optimization/72488
36827 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
36828 sure to restore SSA info.
36829 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
36831 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
36833 PR rtl-optimization/79121
36834 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
36835 of the inner type when shifting an extended value.
36837 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
36840 * symtab.c (symtab_node::equal_address_to): Fix comparing of
36841 interposable aliases.
36843 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
36846 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
36847 Use the evmergelohi instruction.
36848 (mov_si<mode>_e500_subreg4_2_le): Likewise.
36849 (mov_sitf_e500_subreg8_2_be): Likewise.
36850 (mov_sitf_e500_subreg12_2_le): Likewise.
36851 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
36852 (mov_si<mode>_e500_subreg4_2_be): Likewise.
36853 (mov_sitf_e500_subreg8_2_le): Likewise.
36854 (mov_sitf_e500_subreg12_2_be): Likewise.
36856 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36858 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
36859 attribute from vecsimple to vecperm.
36860 (altivec_vbpermq2): Likewise.
36862 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36865 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
36867 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36868 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
36869 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
36870 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
36871 case where N arg is SIZE_MAX.
36872 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
36873 (cmpstrsi): Add pattern.
36875 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
36877 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
36878 __builtin_vec_revb builtins.
36879 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
36880 built-in functions to support generation of the ISA 3.0 XXBR<x>
36881 vector byte reverse instructions.
36882 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
36883 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
36884 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
36885 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
36886 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
36887 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
36888 (P9V_BUILTIN_VEC_REVB): Likewise.
36889 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
36890 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
36891 (p9_xxbrq_v16qi): Likewise.
36892 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
36893 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
36894 (p9_xxbrh_v8hi): Likewise.
36895 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
36896 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
36897 vec_revb built-in functions.
36899 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
36901 PR rtl-optimization/78952
36902 * config/i386/i386.md (any_extract): New code iterator.
36903 (*insvqi_2): Use any_extract for source operand.
36904 (*insvqi_3): Use any_shiftrt for source operand.
36906 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
36908 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
36910 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
36912 2017-01-18 Matthias Klose <doko@ubuntu.com>
36914 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
36916 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36918 * config/rs6000/altivec.h (vec_bperm): Change #define.
36919 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
36920 (altivec_vbpermq2): New define_insn.
36921 (altivec_vbpermd): Likewise.
36922 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
36923 function interface.
36924 (VBPERMD): Likewise.
36925 (VBPERM): New polymorphic function interface.
36926 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
36927 Add entries for P9V_BUILTIN_VEC_VBPERM.
36928 * doc/extend.texi: Add interfaces for vec_bperm.
36930 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36932 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
36933 first letter of error messages.
36934 (s390_resolve_overloaded_builtin): Likewise.
36935 * config/s390/s390.c (s390_expand_builtin): Likewise.
36936 (s390_invalid_arg_for_unprototyped_fn): Likewise.
36937 (s390_valid_target_attribute_inner_p): Likewise.
36938 * config/s390/s390.md ("tabort"): Likewise.
36940 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
36942 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
36943 (ISA_AVOID_DIV_HILO): New macro.
36944 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
36945 (ISA_HAS_DDIV): Likewise.
36947 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36949 * doc/invoke.texi (fabi-version): Correct number of occurrences.
36951 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36953 * doc/invoke.texi (fabi-version): Spelling fix.
36955 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36958 * doc/invoke.texi (fabi-version): Mention mangling fix for
36961 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36964 * doc/invoke.texi (fabi-version): Document discriminator mangling.
36966 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
36969 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
36970 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
36972 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
36974 (stack_protect_test): Ditto.
36975 * config/rs6000/rs6000.opt (mstack-protector-guard=,
36976 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
36978 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
36979 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
36980 -mstack-protector-guard-offset=.
36981 (RS/6000 and PowerPC Options): Ditto.
36983 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
36985 * config/i386/i386.h (MASK_CLASS_P): New define.
36986 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
36987 there are no registers from different register sets also when
36988 mask registers are used. Update function comment.
36989 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
36990 to (*k/*r) and (*k/*km) alternatives.
36992 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
36994 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
36995 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
36996 (EH_RETURN_HANDLER_RTX): New define.
36997 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
36998 Force frame pointer in EH return functions.
36999 (aarch64_expand_epilogue): Add barrier for eh_return.
37000 (aarch64_final_eh_return_addr): Remove.
37001 (aarch64_eh_return_handler_rtx): New function.
37002 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
37004 (aarch64_eh_return_handler_rtx): New prototype.
37006 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37008 * config/rs6000/altivec.h (vec_rlmi): New #define.
37009 (vec_vrlnm): Likewise.
37010 (vec_rlnm): Likewise.
37011 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
37012 (UNSPEC_VRLNM): Likewise.
37013 (VIlong): New mode iterator.
37014 (altivec_vrl<VI_char>mi): New define_insn.
37015 (altivec_vrl<VI_char>nm): Likewise.
37016 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
37018 (VRLDNM): Likewise.
37019 (RLNM): New polymorphic function entry.
37020 (VRLWMI): New monomorphic function entry.
37021 (VRLDMI): Likewise.
37022 (RLMI): New polymorphic function entry.
37023 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
37024 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
37025 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
37028 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37031 * dwarf2out.c (field_byte_offset): Restore the
37032 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
37033 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
37034 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
37037 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37040 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
37042 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37045 * dwarf2out.c (add_data_member_location_attribute): For constant
37046 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
37047 instead of DW_AT_data_member_location, DW_AT_bit_offset and
37048 DW_AT_byte_size attributes.
37050 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37052 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
37053 after forcing to constant memory when the code model is medium.
37055 2017-01-17 Julia Koval <julia.koval@intel.com>
37058 * config/i386/avx512fintrin.h
37059 (_mm512_i32gather_ps): Change __addr type to void const*.
37060 (_mm512_mask_i32gather_ps): Ditto.
37061 (_mm512_i32gather_pd): Ditto.
37062 (_mm512_mask_i32gather_pd): Ditto.
37063 (_mm512_i64gather_ps): Ditto.
37064 (_mm512_mask_i64gather_ps): Ditto.
37065 (_mm512_i64gather_pd): Ditto.
37066 (_mm512_mask_i64gather_pd): Ditto.
37067 (_mm512_i32gather_epi32): Ditto.
37068 (_mm512_mask_i32gather_epi32): Ditto.
37069 (_mm512_i32gather_epi64): Ditto.
37070 (_mm512_mask_i32gather_epi64): Ditto.
37071 (_mm512_i64gather_epi32): Ditto.
37072 (_mm512_mask_i64gather_epi32): Ditto.
37073 (_mm512_i64gather_epi64): Ditto.
37074 (_mm512_mask_i64gather_epi64): Ditto.
37075 (_mm512_i32scatter_ps): Change __addr type to void*.
37076 (_mm512_mask_i32scatter_ps): Ditto.
37077 (_mm512_i32scatter_pd): Ditto.
37078 (_mm512_mask_i32scatter_pd): Ditto.
37079 (_mm512_i64scatter_ps): Ditto.
37080 (_mm512_mask_i64scatter_ps): Ditto.
37081 (_mm512_i64scatter_pd): Ditto.
37082 (_mm512_mask_i64scatter_pd): Ditto.
37083 (_mm512_i32scatter_epi32): Ditto.
37084 (_mm512_mask_i32scatter_epi32): Ditto.
37085 (_mm512_i32scatter_epi64): Ditto.
37086 (_mm512_mask_i32scatter_epi64): Ditto.
37087 (_mm512_i64scatter_epi32): Ditto.
37088 (_mm512_mask_i64scatter_epi32): Ditto.
37089 (_mm512_i64scatter_epi64): Ditto.
37090 (_mm512_mask_i64scatter_epi64): Ditto.
37091 * config/i386/avx512pfintrin.h
37092 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
37093 (_mm512_mask_prefetch_i32gather_ps): Ditto.
37094 (_mm512_mask_prefetch_i64gather_pd): Ditto.
37095 (_mm512_mask_prefetch_i64gather_ps): Ditto.
37096 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
37097 (_mm512_prefetch_i32scatter_ps): Ditto.
37098 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
37099 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
37100 (_mm512_prefetch_i64scatter_pd): Ditto.
37101 (_mm512_prefetch_i64scatter_ps): Ditto.
37102 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
37103 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
37104 * config/i386/avx512vlintrin.h
37105 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
37106 (_mm_mmask_i32gather_ps): Ditto.
37107 (_mm256_mmask_i32gather_pd): Ditto.
37108 (_mm_mmask_i32gather_pd): Ditto.
37109 (_mm256_mmask_i64gather_ps): Ditto.
37110 (_mm_mmask_i64gather_ps): Ditto.
37111 (_mm256_mmask_i64gather_pd): Ditto.
37112 (_mm_mmask_i64gather_pd): Ditto.
37113 (_mm256_mmask_i32gather_epi32): Ditto.
37114 (_mm_mmask_i32gather_epi32): Ditto.
37115 (_mm256_mmask_i32gather_epi64): Ditto.
37116 (_mm_mmask_i32gather_epi64): Ditto.
37117 (_mm256_mmask_i64gather_epi32): Ditto.
37118 (_mm_mmask_i64gather_epi32): Ditto.
37119 (_mm256_mmask_i64gather_epi64): Ditto.
37120 (_mm_mmask_i64gather_epi64): Ditto.
37121 (_mm256_i32scatter_ps): Change __addr type to void*.
37122 (_mm256_mask_i32scatter_ps): Ditto.
37123 (_mm_i32scatter_ps): Ditto.
37124 (_mm_mask_i32scatter_ps): Ditto.
37125 (_mm256_i32scatter_pd): Ditto.
37126 (_mm256_mask_i32scatter_pd): Ditto.
37127 (_mm_i32scatter_pd): Ditto.
37128 (_mm_mask_i32scatter_pd): Ditto.
37129 (_mm256_i64scatter_ps): Ditto.
37130 (_mm256_mask_i64scatter_ps): Ditto.
37131 (_mm_i64scatter_ps): Ditto.
37132 (_mm_mask_i64scatter_ps): Ditto.
37133 (_mm256_i64scatter_pd): Ditto.
37134 (_mm256_mask_i64scatter_pd): Ditto.
37135 (_mm_i64scatter_pd): Ditto.
37136 (_mm_mask_i64scatter_pd): Ditto.
37137 (_mm256_i32scatter_epi32): Ditto.
37138 (_mm256_mask_i32scatter_epi32): Ditto.
37139 (_mm_i32scatter_epi32): Ditto.
37140 (_mm_mask_i32scatter_epi32): Ditto.
37141 (_mm256_i32scatter_epi64): Ditto.
37142 (_mm256_mask_i32scatter_epi64): Ditto.
37143 (_mm_i32scatter_epi64): Ditto.
37144 (_mm_mask_i32scatter_epi64): Ditto.
37145 (_mm256_i64scatter_epi32): Ditto.
37146 (_mm256_mask_i64scatter_epi32): Ditto.
37147 (_mm_i64scatter_epi32): Ditto.
37148 (_mm_mask_i64scatter_epi32): Ditto.
37149 (_mm256_i64scatter_epi64): Ditto.
37150 (_mm256_mask_i64scatter_epi64): Ditto.
37151 (_mm_i64scatter_epi64): Ditto.
37152 (_mm_mask_i64scatter_epi64): Ditto.
37153 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
37154 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
37155 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
37156 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
37157 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
37158 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
37159 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
37160 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
37161 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
37162 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
37163 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
37164 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
37165 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
37166 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
37167 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
37168 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
37169 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
37170 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
37171 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
37172 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
37173 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
37174 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
37175 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
37176 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
37177 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
37178 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
37179 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
37180 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
37181 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
37182 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
37183 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
37184 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
37185 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
37186 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
37187 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
37188 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
37189 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
37190 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
37191 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
37192 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
37193 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
37194 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
37195 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
37196 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
37197 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
37198 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
37199 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
37200 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
37201 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
37202 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
37203 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
37204 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
37205 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
37206 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
37207 definitions accordingly.
37209 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
37210 Kuan-Lin Chen <kuanlinchentw@gmail.com>
37213 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
37216 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
37219 * ira-conflicts.c (ira_build_conflicts): Update total conflict
37220 hard regs for inner regno.
37222 2017-01-17 Martin Liska <mliska@suse.cz>
37225 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
37226 assumption and add comment.
37228 2017-01-17 Nathan Sidwell <nathan@acm.org>
37230 * ipa-visibility.c (localize_node): New function, broken out of ...
37231 (function_and_variable_visibility): ... here. Call it.
37233 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
37235 PR middle-end/77445
37236 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
37237 correctly set frequency of oudgoing edge.
37238 (duplicate_thread_path): Fix profile updating.
37240 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37243 * configure.ac: Add GCC_BASE_VER.
37244 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
37245 version from BASE-VER file.
37246 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
37247 (gcc.o): Depend on $(BASEVER).
37248 * common.opt (dumpfullversion): New option.
37249 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
37250 * doc/invoke.texi: Document -dumpfullversion.
37251 * doc/install.texi: Document --with-gcc-major-version-only.
37252 * configure: Regenerated.
37254 2017-01-17 Richard Biener <rguenther@suse.de>
37256 PR tree-optimization/71433
37257 * tree-vrp.c (register_new_assert_for): Merge same asserts
37258 on all incoming edges.
37259 (process_assert_insertions_for): Handle insertions at the
37262 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
37264 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
37265 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
37267 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
37270 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
37273 2017-01-17 Alan Modra <amodra@gmail.com>
37276 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
37277 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
37278 symbolic stack limit when pic.
37280 2017-01-16 Martin Sebor <msebor@redhat.com>
37282 PR tree-optimization/78608
37283 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
37285 2017-01-16 Jeff Law <law@redhat.com>
37288 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
37289 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
37290 for several include directories that may be relative to sysroot.
37291 * config/i386/x-mingw32 (gplus_includedir): Define.
37292 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
37293 (native_system_includedir): Likewise.
37294 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
37295 override if TARGET_SYSTEM_ROOT is defined.
37296 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
37298 PR tree-optimization/79090
37299 PR tree-optimization/33562
37300 PR tree-optimization/61912
37301 PR tree-optimization/77485
37302 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
37303 and computed trims into the dump file.
37305 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37307 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
37309 2017-01-16 Jakub Jelinek <jakub@redhat.com>
37312 * gimplify.c (gimplify_init_constructor): If want_value and
37313 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
37317 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
37318 sequence. Formatting fixes.
37319 (doloop_optimize): Formatting fixes.
37322 * gcc.c (debug_level_greater_than_spec_func): New function.
37323 (static_spec_functions): Add debug-level-gt spec function.
37324 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
37326 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37327 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37328 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
37329 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
37330 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
37331 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
37333 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
37335 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
37336 QImode fixups to general and mask registers only.
37338 2017-01-16 Carl Love <cel@us.ibm.com>
37340 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
37341 for built-in functions
37342 vector signed char vec_nabs (vector signed char)
37343 vector signed short vec_nabs (vector signed short)
37344 vector signed int vec_nabs (vector signed int)
37345 vector signed long long vec_nabs (vector signed long long)
37346 vector float vec_nabs (vector float)
37347 vector double vec_nabs (vector double)
37348 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
37350 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
37351 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
37352 * doc/extend.texi: Update the documentation file for the new built-in
37355 2017-01-16 Martin Sebor <msebor@redhat.com>
37357 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
37360 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37362 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
37363 UNSPEC_VSX__XXSPLTD to require special splat handling.
37365 2017-01-16 David Malcolm <dmalcolm@redhat.com>
37368 * system.h: Poison strndup.
37370 2017-01-16 Alan Modra <amodra@gmail.com>
37373 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
37376 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
37378 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
37380 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
37382 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
37383 call recog here. Assert that INSN_CODE (insn) is non-negative.
37385 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37388 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
37390 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
37391 in the currently scheduled RTL fragment.
37393 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37395 PR rtl-optimization/78751
37396 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
37399 2017-01-14 Jeff Law <law@redhat.com>
37401 PR tree-optimization/79090
37402 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
37403 variable length stores.
37404 (compute_trims): Delete dead assignment to *trim_tail.
37405 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
37408 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
37410 PR rtl-optimization/78626
37411 PR rtl-optimization/78727
37412 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
37413 of a block, and split such blocks after everything else is finished.
37415 2017-01-14 Alan Modra <amodra@gmail.com>
37418 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
37419 target legitimate_combined_insn.
37420 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
37421 (rs6000_legitimate_combined_insn): New function.
37422 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
37424 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
37425 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
37426 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
37428 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
37430 * doc/frontends.texi (G++ and GCC): Remove references to Java.
37432 2017-01-13 Jeff Law <law@redhat.com>
37434 PR tree-optimization/33562
37435 PR tree-optimization/61912
37436 PR tree-optimization/77485
37437 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
37439 (delete_dead_assignment): Likewise.
37440 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
37441 statement to delete_dead_call and delete_dead_assignment.
37443 2017-01-13 David Malcolm <dmalcolm@redhat.com>
37446 * substring-locations.c (format_warning_va): Strengthen case 1 so
37447 that both endpoints of the substring must be within the format
37448 range for just the substring to be printed.
37450 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
37452 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
37453 * config/i386/i386.c (ix86_target_string): Add missing options
37454 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
37455 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
37456 flags_other and ix86_target_other to flags2_other. Display unknown
37458 (ix86_valid_target_attribute_inner_p): Add missing options and
37459 reorder options by implied ISAs, as in ix86_target_string.
37461 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
37463 * hash-table.h (hash_table::too_empty_p): New function.
37464 (hash_table::expand): Use it.
37465 (hash_table::traverse): Likewise.
37466 (hash_table::empty_slot): Use sizeof (value_type) instead of
37467 sizeof (PTR) to convert bytes to elements. Shrink the table
37468 if the current size is excessive for the current number of
37471 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
37473 * ira-costs.c (record_reg_classes): Break from the inner loop
37474 early once alt_fail is known to be true. Update outer loop
37475 handling accordingly.
37477 2017-01-13 Jeff Law <law@redhat.com>
37479 * tree-ssa-dse.c (decrement_count): New function.
37480 (increment_start_addr, maybe_trim_memstar_call): Likewise.
37481 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
37482 when we know the partially dead statement is a mem* function.
37484 PR tree-optimization/61912
37485 PR tree-optimization/77485
37486 * tree-ssa-dse.c: Include expr.h.
37487 (maybe_trim_constructor_store): New function.
37488 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
37490 PR tree-optimization/33562
37491 PR tree-optimization/61912
37492 PR tree-optimization/77485
37493 * doc/invoke.texi: Document new dse-max-object-size param.
37494 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
37495 * tree-ssa-dse.c: Include params.h.
37496 (dse_store_status): New enum.
37497 (initialize_ao_ref_for_dse): New, partially extracted from
37499 (valid_ao_ref_for_dse, normalize_ref): New.
37500 (setup_live_bytes_from_ref, compute_trims): Likewise.
37501 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
37502 (maybe_trim_partially_dead_store): Likewise.
37503 (maybe_trim_complex_store): Likewise.
37504 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
37505 Track what bytes live from the original store. Return tri-state
37506 for dead, partially dead or live.
37507 (dse_dom_walker): Add constructor, destructor and new private members.
37508 (delete_dead_call, delete_dead_assignment): New extracted from
37510 (dse_optimize_stmt): Make a member of dse_dom_walker.
37511 Use initialize_ao_ref_for_dse.
37513 PR tree-optimization/33562
37514 PR tree-optimization/61912
37515 PR tree-optimization/77485
37516 * sbitmap.h (bitmap_count_bits): Prototype.
37517 (bitmap_clear_range, bitmap_set_range): Likewise.
37518 * sbitmap.c (bitmap_clear_range): New function.
37519 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
37521 2017-01-13 Martin Liska <mliska@suse.cz>
37524 * function.c (set_cfun): Add new argument force.
37525 * function.h (set_cfun): Likewise.
37526 * ipa-inline-transform.c (inline_call): Use the function when
37527 strict alising from is dropped for function we inline to.
37529 2017-01-13 Richard Biener <rguenther@suse.de>
37531 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
37532 for dumping GIMPLE INTEGER_CSTs.
37534 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
37536 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
37537 to 201112L since C++17.
37539 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
37542 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
37543 if -fsanitize=kernel-address is present.
37545 2017-01-13 Richard Biener <rguenther@suse.de>
37547 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
37548 as _Literal ( type ) number in case usual suffixes do not
37549 preserve all information.
37551 2017-01-13 Richard Biener <rguenther@suse.de>
37553 PR tree-optimization/77283
37554 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
37555 and ssa-iterators.h.
37556 (is_feasible_trace): Implement a cost model based on joiner
37559 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
37562 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
37563 char or short to __float128/_Float128 directly.
37565 2017-01-12 Martin Sebor <msebor@redhat.com>
37567 to -Wformat-overflow.
37568 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
37569 (min_bytes_remaining): Same.
37570 (get_string_length): Same.
37571 (format_string): Same.
37572 (format_directive): Same.
37574 (pass_sprintf_length::handle_gimple_call): Same.
37576 2017-01-12 Jakub Jelinek <jakub@redhat.com>
37578 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
37579 info.nowrite calls with no lhs that can't throw. Return bool
37580 whether gsi_remove has been called or not.
37581 (pass_sprintf_length::handle_gimple_call): Return bool whether
37582 try_substitute_return_value called gsi_remove. Formatting fix.
37583 (pass_sprintf_length::execute): Don't use gsi_remove if
37584 handle_gimple_call returned true.
37587 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
37588 be removed due to side-effects, don't remove following barrier nor
37589 turn the successor edge into fallthru edge.
37591 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37594 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
37595 element-reversing loads and stores as not swappable.
37597 2017-01-12 Nathan Sidwell <nathan@acm.org>
37598 Nicolai Stange <nicstange@gmail.com>
37600 * combine.c (try_combine): Don't ignore result of overlap checking
37601 loop. Combine overlap & asm check into single loop.
37603 2017-01-12 Richard Biener <rguenther@suse.de>
37605 * tree-pretty-print.c (dump_generic_node): Provide -gimple
37606 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
37608 2017-01-12 Richard Biener <rguenther@suse.de>
37610 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
37611 and TS_TARGET_OPTION directly derive from TS_BASE.
37612 * tree-core.h (tree_optimization_option): Derive from tree_base.
37613 (tree_target_option): Likewise.
37615 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
37617 * config/i386/i386.c (memory_address_length): Increase len
37618 only when rip_relative_addr_p returns false.
37620 2017-01-11 Julia Koval <julia.koval@intel.com>
37622 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
37623 (OPTION_MASK_ISA_SGX_SET): New.
37624 (ix86_handle_option): Handle OPT_msgx.
37625 * config.gcc: Added sgxintrin.h.
37626 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
37627 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
37628 * config/i386/i386.c (ix86_target_string): Add -msgx.
37630 (ix86_option_override_internal): Handle new options.
37631 (ix86_valid_target_attribute_inner_p): Add sgx.
37632 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
37633 * config/i386/i386.opt: Add msgx.
37634 * config/i386/sgxintrin.h: New file.
37635 * config/i386/x86intrin.h: Add sgxintrin.h.
37637 2017-01-11 Jakub Jelinek <jakub@redhat.com>
37640 * fold-const.c (maybe_nonzero_address): Return 1 for function
37642 (tree_single_nonzero_warnv_p): Don't handle function local objects
37646 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
37649 2017-01-11 David Malcolm <dmalcolm@redhat.com>
37652 * opts.c: Include "spellcheck.h"
37653 (struct string_fragment): New struct.
37654 (struct edit_distance_traits<const string_fragment &>): New
37656 (get_closest_sanitizer_option): New function.
37657 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
37659 2017-01-11 Jakub Jelinek <jakub@redhat.com>
37661 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
37663 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
37664 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
37665 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
37666 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
37667 for initial die_offset if dwarf_split_debug_info.
37668 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
37669 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
37671 (output_skeleton_debug_sections): Formatting fix. Use
37672 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
37673 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
37675 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
37677 * config/arm/cortex-a53.md: Add bypasses for
37678 cortex_a53_r2f_cvt.
37679 (cortex_a53_r2f): Only use for transfers.
37680 (cortex_a53_f2r): Likewise.
37681 (cortex_a53_r2f_cvt): Add reservation for conversions.
37682 (cortex_a53_f2r_cvt): Likewise.
37684 2017-01-11 Tamar Christina <tamar.christina@arm.com>
37686 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
37687 to all inlined functions, change static to extern.
37689 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
37692 * config/arm/arm.c (legitimize_pic_address): Handle reference to
37694 (arm_assemble_integer): Likewise.
37696 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
37698 * config.gcc: Use new awk script to check CPU, FPU and architecture
37699 parameters for --with-... options.
37700 * config/arm/parsecpu.awk: New file
37701 * config/arm/arm-cpus.in: New file.
37702 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
37704 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
37706 * config/arm/t-arm: Update dependency rules.
37707 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
37708 of processing .def files.
37709 * config/arm/genopt.sh: Deleted.
37710 * config/arm/gentune.sh: Deleted.
37711 * config/arm/arm-cores.def: Deleted.
37712 * config/arm/arm-arches.def: Deleted.
37713 * config/arm/arm-fpus.def: Deleted.
37714 * config/arm/arm-tune.md: Regenerated.
37715 * config/arm/arm-tables.opt: Regenerated.
37716 * config/arm/arm-cpu.h: New generated file.
37717 * config/arm/arm-cpu-data.h: New generated file.
37718 * config/arm/arm-cpu-cdata.h: New generated file.
37720 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
37723 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
37725 (input_varpool_node): Unpack dynamically_initialized bit.
37727 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
37729 PR rtl-optimization/79032
37730 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
37731 the alignment of the adjusted memory reference against that of MODE,
37732 instead of the alignment of the original memory reference.
37734 2017-01-11 Martin Jambor <mjambor@suse.cz>
37736 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
37738 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
37739 decorated functions.
37741 2017-01-11 Richard Biener <rguenther@suse.de>
37743 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
37744 set range/nonnull info for PHI results. Do not set it on
37745 stmts marked for removal.
37747 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
37749 * expr.c (store_field): In the bitfield case, fetch the return value
37750 from the registers before applying a single big-endian adjustment.
37751 Always do a final load for a BLKmode value not larger than a word.
37753 2017-01-10 David Malcolm <dmalcolm@redhat.com>
37756 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
37757 that we correctly handle column numbers greater than
37758 LINE_MAP_MAX_COLUMN_NUMBER.
37760 2017-01-10 Martin Sebor <msebor@redhat.com>
37762 PR middle-end/78245
37763 * gimple-ssa-sprintf.c (get_destination_size): Call
37764 {init,fini}object_sizes.
37765 * tree-object-size.c (addr_object_size): Adjust.
37766 (pass_through_call): Adjust.
37767 (pass_object_sizes::execute): Adjust.
37768 * tree-object-size.h (fini_object_sizes): Declare.
37770 2017-01-10 Martin Sebor <msebor@redhat.com>
37772 PR tree-optimization/78775
37773 * builtins.c (get_size_range): Move...
37774 * calls.c: ...to here.
37775 (alloc_max_size): Accept zero argument.
37776 (operand_signed_p): Remove.
37777 (maybe_warn_alloc_args_overflow): Call get_size_range.
37778 * calls.h (get_size_range): Declare.
37780 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
37782 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
37783 from TI's devices.csv file as of September 2016.
37784 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
37786 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
37788 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
37789 * doc/invoke.texi: Likewise.
37790 * doc/md.texi: Likewise.
37791 * doc/objc.texi: Likewise.
37793 2017-01-10 Joshua Conner <joshconner@google.com>
37795 * config/arm/fuchsia-elf.h: New file.
37796 * config/fuchsia.h: New file.
37797 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
37798 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
37800 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
37802 2016-01-10 Richard Biener <rguenther@suse.de>
37804 PR tree-optimization/79034
37805 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
37806 Propagate out degenerate PHIs in the joiner.
37808 2017-01-10 Martin Liska <mliska@suse.cz>
37810 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
37811 (sort_congruence_classes_by_decl_uid): Likewise.
37812 (sort_congruence_class_groups_by_decl_uid): Likewise.
37813 (sem_item_optimizer::merge_classes): Sort class, groups in these
37814 classes and members in the groups by DECL_UID of declarations.
37815 This would make merge operations stable.
37817 2017-01-10 Martin Liska <mliska@suse.cz>
37819 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
37820 usage of m_classes_vec.
37821 (sem_item_optimizer::~sem_item_optimizer): Likewise.
37822 (sem_item_optimizer::get_group_by_hash): Likewise.
37823 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
37824 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
37825 (sem_item_optimizer::verify_classes): Likewise.
37826 (sem_item_optimizer::process_cong_reduction): Likewise.
37827 (sem_item_optimizer::dump_cong_classes): Likewise.
37828 (sem_item_optimizer::merge_classes): Likewise.
37829 * ipa-icf.h (congruence_class_hash): Rename from
37830 congruence_class_group_hash. Remove declaration of m_classes_vec.
37832 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
37834 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
37835 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
37836 * config.gcc: Add avx512vpopcntdqintrin.h.
37837 * config/i386/avx512vpopcntdqintrin.h: New.
37838 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
37839 * config/i386/i386-builtin-types.def: Add new types.
37840 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
37841 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
37842 __builtin_ia32_vpopcountq_v8di_mask): New.
37843 * config/i386/i386-c.c (ix86_target_macros_internal): Define
37844 __AVX512VPOPCNTDQ__.
37845 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
37846 (PTA_AVX512VPOPCNTDQ): Define.
37847 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
37848 TARGET_AVX512VPOPCNTDQ_P): Define.
37849 * config/i386/i386.opt: Add mavx512vpopcntdq.
37850 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
37851 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
37853 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37855 PR middle-end/77484
37856 * predict.def (PRED_CALL): Set to 67.
37858 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
37860 * expr.c (store_field): In the bitfield case, if the value comes from
37861 a function call and is of an aggregate type returned in registers, do
37862 not modify the field mode; extract the value in all cases if the mode
37863 is BLKmode and the size is not larger than a word.
37865 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
37868 * config/i386/cpuid.h: Fix undefined behavior.
37870 2017-01-04 Jeff Law <law@redhat.com>
37872 PR tree-optimization/79007
37873 PR tree-optimization/67955
37874 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
37875 conservative for pt.null when flag_non_call_exceptions is on.
37877 2017-01-09 Jakub Jelinek <jakub@redhat.com>
37879 PR translation/79019
37880 PR translation/79020
37881 * params.def (PARAM_INLINE_MIN_SPEEDUP,
37882 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
37883 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
37885 * config/avr/avr.opt (maccumulate-args): Likewise.
37886 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
37887 * common.opt (freport-bug): Likewise.
37888 * cif-code.def (CIF_FINAL_ERROR): Likewise.
37889 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
37890 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
37891 translatable string.
37892 * config/i386/i386.c (function_value_32): Likewise.
37893 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
37894 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
37896 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
37897 * common/config/msp430/msp430-common.c (msp430_handle_option):
37899 * symtab.c (symtab_node::verify_base): Likewise.
37900 * opts.c (set_debug_level): Likewise.
37901 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
37902 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
37903 missing whitespace to translatable strings.
37904 * config/avr/avr.md (bswapsi2): Fix typo in comment.
37905 * config/sh/superh.h: Likewise.
37906 * config/i386/xopintrin.h: Likewise.
37907 * config/i386/znver1.md: Likewise.
37908 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
37909 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
37910 * double-int.h (struct double_int): Likewise.
37911 * double-int.c (div_and_round_double): Likewise.
37912 * wide-int.cc: Likewise.
37913 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
37914 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
37915 * cfgcleanup.c (crossjumps_occured): Renamed to ...
37916 (crossjumps_occurred): ... this.
37917 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
37920 PR tree-optimization/78899
37921 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
37922 returning bool return struct loop *, NULL for failure and the new
37924 (versionable_outer_loop_p): Don't version outer loop if it has
37925 dont_vectorized bit set.
37926 (tree_if_conversion): When versioning outer loop, ensure
37927 tree_if_conversion is performed also on the inner loop of the
37928 non-vectorizable outer loop copy.
37929 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
37930 LOOP_VECTORIZED in inner loop of the scalar outer loop and
37931 prevent vectorization of it.
37932 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
37933 the outer loop vectorization of the non-scalar version is attempted
37934 before vectorization of the inner loop in scalar version. If
37935 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
37936 vectorization of its inner loop.
37937 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
37938 has 2 inner loops, rename also on edges from bb whose single pred
37939 is outer_loop->header. Fix typo in function comment.
37941 2017-01-09 Martin Sebor <msebor@redhat.com>
37944 * asan.c (asan_emit_stack_protection): Increase local buffer size
37945 to avoid snprintf truncation warning.
37947 2017-01-09 Andrew Pinski <apinski@cavium.com>
37949 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
37950 to reference thunderx2t99 for the tuning structure
37951 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
37953 (thunderx2t99_extra_costs): This.
37954 * config/aarch64/aarch64-tune.md: Regenerate.
37955 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
37956 (vulcan_addrcost_table): This.
37957 (vulcan_regmove_cost): Rename to ...
37958 (thunderx2t99_regmove_cost): This.
37959 (vulcan_vector_cost): Rename to ...
37960 (thunderx2t99_vector_cost): this.
37961 (vulcan_branch_cost): Rename to ...
37962 (thunderx2t99_branch_cost): This.
37963 (vulcan_tunings): Rename to ...
37964 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
37965 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
37967 2017-01-09 Martin Jambor <mjambor@suse.cz>
37971 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
37972 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
37973 (propagate_vr_accross_jump_function): Use the above function for all
37974 value range computations for pass-through jump functions and type
37975 converasion from explicit value range values.
37976 (ipcp_propagate_stage): Do not attempt to deduce types of formal
37977 parameters from TYPE_ARG_TYPES.
37978 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
37979 (ipa_write_node_info): Stream type of the actual argument.
37980 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
37982 2017-01-09 Martin Liska <mliska@suse.cz>
37985 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
37986 (lookup_compiler): Do not show error message with have_E.
37988 2017-01-09 Jakub Jelinek <jakub@redhat.com>
37990 PR tree-optimization/78938
37991 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
37992 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
37993 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
37994 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
37997 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37999 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
38002 2017-01-09 Richard Biener <rguenther@suse.de>
38004 PR tree-optimization/78997
38005 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
38006 name condition properly.
38008 2017-01-09 Richard Biener <rguenther@suse.de>
38011 * dwarf2out.c (is_cxx): New overload with context.
38012 (is_naming_typedef_decl): Use it.
38014 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38016 * invoke.texi (Option Summary): Correct spacing in option lists
38017 and add line breaks to fix over-long lines.
38019 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38021 PR middle-end/17660
38023 * extend.texi (Common Variable Attributes): Add xref to GCC
38024 Internals manual to explain mode attribute keywords.
38026 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38029 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
38030 and Preprocessor Options.
38031 (Options for Linking): Document -pthread here....
38032 (RS/6000 and PowerPC Options): ...not here.
38033 (Solaris 2 Options): ...or here.
38034 * doc/cppopts.texi: Document -pthread.
38036 2017-01-08 Martin Sebor <msebor@redhat.com>
38038 PR middle-end/77708
38039 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
38040 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
38041 New member functions.
38042 (format_directive): Used them.
38044 (pass_sprintf_length::handle_gimple_call): Same.
38045 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
38046 to avoid truncation for any argument.
38047 (extract_affine_mul): Same.
38048 * tree.c (get_file_function_name): Same.
38050 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38052 PR middle-end/77484
38053 * predict.def (PRED_INDIR_CALL): Set to 86.
38055 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38057 PR preprocessor/54124
38058 * doc/cppopts.texi: Reformat -d subtable to list the full name
38059 of the options. Add cross-reference to the docs for the general
38060 compiler -d options.
38061 * doc/invoke.texi (Developer Options): Add cross-reference to the
38062 preprocessor-specific -d option documentation.
38064 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38066 PR preprocessor/13498
38067 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
38068 redudant material, and reflect new command-line options.
38069 (System Headers): Likewise.
38071 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38073 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
38074 -isystem, and -idirafter. Copy-edit.
38075 * doc/cppopts.texi: Copy-edit. Remove contradiction about
38076 default for -ftrack-macro-expansion. Delete obsolete and
38077 badly-formatted implementation details about -fdebug-cpp output.
38078 * doc/cppwarnopts.texi: Copy-edit.
38080 2017-01-07 David Malcolm <dmalcolm@redhat.com>
38083 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
38084 that the transition from a max line width >= 1<<10 to narrower
38085 lines works correctly.
38087 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
38089 * doc/options.texi (PerFunction): New.
38090 * opt-functions.awk (switch_flags): Map both Optimization and
38091 PerFunction to CL_OPTIMIZATION.
38092 * opth-gen.awk: Test for PerFunction flag along with
38094 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
38095 it only when the latter is present. Skip those that don't in
38096 the hash function generator.
38097 * common.opt (fvar-tracking): Mark as PerFunction instead of
38099 (fvar-tracking-assignments): Likewise.
38100 (fvar-tracking-assignments-toggle): Likewise.
38101 (fvar-tracking-uninit): Likewise.
38103 2017-01-07 Jakub Jelinek <jakub@redhat.com>
38105 PR translation/79018
38106 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
38109 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
38112 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
38113 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
38114 TARGET_LONG_JUMP_TABLE_OFFSETS.
38115 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
38116 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
38117 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
38118 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
38119 * config/m68k/m68k.md (tablejump expander): Likewise.
38120 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
38121 TARGET_LONG_JUMP_TABLE_OFFSETS.
38122 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
38123 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
38125 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
38126 David Holsgrove <david.holsgrove@xilinx.com>
38128 * common/config/microblaze/microblaze-common.c
38129 (TARGET_EXCEPT_UNWIND_INFO): Remove.
38130 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
38132 * config/microblaze/microblaze.c (microblaze_must_save_register)
38133 (microblaze_expand_epilogue, microblaze_return_addr): Handle
38135 (microblaze_eh_return): New function.
38136 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
38137 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
38138 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
38139 * config/microblaze/microblaze.md (eh_return): New pattern.
38141 2017-01-06 Jakub Jelinek <jakub@redhat.com>
38143 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
38144 GCC_DIAGNOSTIC_STRINGIFY): Define.
38146 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
38148 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38150 * config/arm/arm.md (<mcrr>): New.
38152 * config/arm/arm.c (arm_arch5te): New.
38153 (arm_option_override): Set arm_arch5te.
38154 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
38156 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
38157 (arm_mcrr_qualifiers): ... this. New.
38158 (MRRC_QUALIFIERS): Define to...
38159 (arm_mrrc_qualifiers): ... this. New.
38160 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
38162 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
38163 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
38164 (MRRCI, mrrc, MRRC): New.
38165 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
38166 VUNSPEC_MRRC2): New.
38168 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38170 * config/arm/arm.md (<mcr>): New.
38172 * config/arm/arm.c (arm_coproc_builtin_available): Add
38173 support for mcr, mrc, mcr2 and mrc2.
38174 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
38175 (arm_mcr_qualifiers): ... this. New.
38176 (MRC_QUALIFIERS): Define to ...
38177 (arm_mrc_qualifiers): ... this. New.
38178 (MCR_QUALIFIERS): Define to ...
38179 (arm_mcr_qualifiers): ... this. New.
38180 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
38182 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
38183 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
38184 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
38185 VUNSPEC_MRC2): New.
38187 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38189 * config/arm/arm.md (*ldc): New.
38193 * config/arm/arm.c (arm_coproc_builtin_available): Add
38194 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
38195 (arm_coproc_ldc_stc_legitimate_address): New.
38196 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
38197 'qualifier_const_pointer'.
38198 (LDC_QUALIFIERS): Define to...
38199 (arm_ldc_qualifiers): ... this. New.
38200 (STC_QUALIFIERS): Define to...
38201 (arm_stc_qualifiers): ... this. New.
38202 * config/arm/arm-protos.h
38203 (arm_coproc_ldc_stc_legitimate_address): New.
38204 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
38205 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
38206 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
38207 stc2, stcl, stc2l): New.
38208 * config/arm/constraints.md (Uz): New.
38209 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
38210 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
38211 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
38212 VUNSPEC_STC2L): New.
38214 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38216 * config/arm/arm.md (<cdp>): New.
38217 * config/arm/arm.c (neon_const_bounds): Rename this ...
38218 (arm_const_bounds): ... this.
38219 (arm_coproc_builtin_available): New.
38220 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
38221 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
38222 (CDP_QUALIFIERS): Define to...
38223 (arm_cdp_qualifiers): ... this. New.
38225 (arm_expand_builtin_args): Add case for 6 arguments.
38226 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
38227 (arm_const_bounds): ... this.
38228 (arm_coproc_builtin_available): New.
38229 * config/arm/arm_acle.h (__arm_cdp): New.
38231 * config/arm/arm_acle_builtins.def (cdp): New.
38233 * config/arm/iterators.md (CDPI,CDP,cdp): New.
38234 * config/arm/neon.md: Rename all 'neon_const_bounds' to
38235 'arm_const_bounds'.
38236 * config/arm/types.md (coproc): New.
38237 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
38238 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
38239 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
38240 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
38242 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38244 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
38245 (UBINOP_QUALIFIERS): New.
38247 (acle_builtin_data): New. Change comment.
38248 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
38249 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
38250 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
38251 arm_acle_builtins.def.
38252 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
38253 (arm_init_acle_builtins): New.
38254 (CRC32_BUILTIN): Remove.
38255 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
38256 crc32cb, crc32ch and crc32cw.
38257 (arm_init_crc32_builtins): Remove.
38258 (arm_init_builtins): Use arm_init_acle_builtins rather
38259 than arm_init_crc32_builtins.
38260 (arm_expand_acle_builtin): New.
38261 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
38262 * config/arm/arm_acle_builtins.def: New.
38264 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38266 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
38267 (arm_builtin_datum): ... this.
38268 (arm_init_neon_builtin): Rename to ...
38269 (arm_init_builtin): ... this. Add a new parameters PREFIX
38270 and USE_SIG_IN_NAME.
38271 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
38272 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
38273 'arm_builtin_datum'.
38274 (arm_init_vfp_builtins): Likewise.
38275 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
38276 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
38277 (arm_expand_neon_args): Rename to ...
38278 (arm_expand_builtin_args): ... this. Rename builtin_arg
38279 enum values and differentiate between ARG_BUILTIN_MEMORY
38280 and ARG_BUILTIN_NEON_MEMORY.
38281 (arm_expand_neon_builtin_1): Rename to ...
38282 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
38283 values, arm_expand_builtin_args and add bool parameter NEON.
38284 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
38285 (arm_expand_vfp_builtin): Likewise.
38286 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
38288 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38290 PR middle-end/77484
38291 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
38292 * predict.c (tree_estimate_probability_bb): Reverse direction of
38293 polymorphic call predictor.
38295 2017-01-06 David Malcolm <dmalcolm@redhat.com>
38297 * passes.c (execute_one_pass): Split out pass-skipping logic into...
38298 (determine_pass_name_match): ...this new function and...
38299 (should_skip_pass_p): ...this new function.
38301 2017-01-06 Nathan Sidwell <nathan@acm.org>
38303 * ipa-visibility.c (function_and_variable_visibility): Reformat
38304 comments and long lines. Remove extrneous if.
38305 * symtab.c (symtab_node::make_decl_local): Fix code format.
38306 (symtab_node::set_section_for_node): Fix comment typo.
38308 2017-01-06 Martin Liska <mliska@suse.cz>
38311 * lra-constraints.c: Rename invariant to lra_invariant.
38312 * predict.c (set_even_probabilities): Initialize e to NULL.
38314 2017-01-05 Martin Sebor <msebor@redhat.com>
38316 PR tree-optimization/78910
38317 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
38318 (format_integer): Correct off-by-one error in the handling
38319 of precision with negative numbers in signed conversions..
38321 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
38323 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
38325 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38327 PR tree-optimization/71016
38328 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
38329 factor_out_conditional_conversion. Formatting fix.
38330 (factor_out_conditional_conversion): Add cond_stmt argument.
38331 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
38332 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
38335 2017-01-05 David Malcolm <dmalcolm@redhat.com>
38337 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
38338 read-rtl-function.o, and selftest-rtl.o.
38339 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
38340 (selftest::aarch64_test_loading_full_dump): New function.
38341 (selftest::aarch64_run_selftests): New function.
38342 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
38343 selftest::aarch64_run_selftests.
38344 * config/i386/i386.c
38345 (selftest::ix86_test_loading_dump_fragment_1): New function.
38346 (selftest::ix86_test_loading_call_insn): New function.
38347 (selftest::ix86_test_loading_full_dump): New function.
38348 (selftest::ix86_test_loading_unspec): New function.
38349 (selftest::ix86_run_selftests): Call the new functions.
38350 * emit-rtl.c (maybe_set_max_label_num): New function.
38351 * emit-rtl.h (maybe_set_max_label_num): New decl.
38352 * function.c (instantiate_decls): Guard call to
38353 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
38354 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
38356 * gensupport.c (gen_reader::gen_reader): Pass "false"
38357 for new "compact" param of rtx_reader.
38358 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
38359 rather than an empty string for NULL strings.
38360 * read-md.c: Potentially include config.h rather than bconfig.h.
38361 Wrap include of errors.h with #ifdef GENERATOR_FILE.
38362 (have_error): New global, copied from errors.c.
38363 (md_reader::read_name): Rename to...
38364 (md_reader::read_name_1): ...this, adding "out_loc" param,
38365 and converting "missing name or number" to returning false, rather
38367 (md_reader::read_name): Reimplement in terms of read_name_1.
38368 (md_reader::read_name_or_nil): New function.
38369 (md_reader::read_string): Handle "(nil)" by returning NULL.
38370 (md_reader::md_reader): Add new param "compact".
38371 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
38372 (md_reader::read_file): New method.
38373 * read-md.h (md_reader::md_reader): Add new param "compact".
38374 (md_reader::read_file): New method.
38375 (md_reader::is_compact): New accessor.
38376 (md_reader::read_name): Convert return type from void to file_location.
38377 (md_reader::read_name_or_nil): New decl.
38378 (md_reader::read_name_1): New decl.
38379 (md_reader::m_compact): New field.
38380 (noop_reader::noop_reader): Pass "false" for new "compact" param
38382 (rtx_reader::rtx_reader): Add new "compact" param.
38383 (rtx_reader::read_rtx_operand): Make virtual and convert return
38384 type from void to rtx.
38385 (rtx_reader::read_until): New decl.
38386 (rtx_reader::handle_any_trailing_information): New virtual function.
38387 (rtx_reader::postprocess): New virtual function.
38388 (rtx_reader::finalize_string): New virtual function.
38389 (rtx_reader::m_in_call_function_usage): New field.
38390 (rtx_reader::m_reuse_rtx_by_id): New field.
38391 * read-rtl-function.c: New file.
38392 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
38393 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
38394 (selftest::verify_three_block_rtl_cfg): New decl.
38395 * read-rtl-function.h: New file.
38396 * read-rtl.c: Potentially include config.h rather than bconfig.h.
38397 For host, include function.h, memmodel.h, and emit-rtl.h.
38398 (one_time_initialization): New function.
38399 (struct compact_insn_name): New struct.
38400 (compact_insn_names): New array.
38401 (find_code): Handle insn codes in compact dumps.
38402 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
38403 (bind_subst_iter_and_attr): Likewise.
38404 (add_condition_to_string): Likewise.
38405 (add_condition_to_rtx): Likewise.
38406 (apply_attribute_uses): Likewise.
38407 (add_current_iterators): Likewise.
38408 (apply_iterators): Likewise.
38409 (initialize_iterators): Guard usage of apply_subst_iterator with
38410 #ifdef GENERATOR_FILE.
38411 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
38412 (md_reader::read_mapping): Likewise.
38413 (add_define_attr_for_define_subst): Likewise.
38414 (add_define_subst_attr): Likewise.
38415 (read_subst_mapping): Likewise.
38416 (check_code_iterator): Likewise.
38417 (rtx_reader::read_rtx): Likewise. Move one-time initialization
38419 (one_time_initialization): New function.
38420 (rtx_reader::read_until): New method.
38421 (read_flags): New function.
38422 (parse_reg_note_name): New function.
38423 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
38424 Handle reuse_rtx ids.
38425 Wrap iterator lookup within #ifdef GENERATOR_FILE.
38426 Add parsing support for RTL dumps, mirroring the special-cases in
38427 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
38428 values, and calling handle_any_trailing_information.
38429 (rtx_reader::read_rtx_operand): Convert return type from void
38430 to rtx, returning return_rtx. Handle case 'e'. Call
38431 finalize_string on XSTR and XTMPL fields.
38432 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
38433 "(nil)" values were omitted. Call the postprocess vfunc on the
38435 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
38436 class ctor. Initialize m_in_call_function_usage. Call
38437 one_time_initialization.
38438 * rtl-tests.c (selftest::test_uncond_jump): Call
38439 set_new_first_and_last_insn.
38440 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
38441 * selftest-rtl.c: New file.
38442 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
38443 (selftest::get_insn_by_uid): New decl.
38444 * selftest-run-tests.c (selftest::run_tests): Call
38445 read_rtl_function_c_tests.
38446 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
38447 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
38450 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
38452 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
38453 operands in a special way. Assert that pos+len <= mode precision.
38455 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38457 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
38458 3 argument Alias with unlimited for the negative form.
38459 (fno-vect-cost-model): Removed.
38461 2017-01-05 Martin Liska <mliska@suse.cz>
38463 * hsa-gen.c (gen_hsa_divmod): New function.
38464 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
38466 2017-01-05 Martin Liska <mliska@suse.cz>
38469 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
38472 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38474 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
38475 small constant length operands.
38477 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38479 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
38480 between loop iterations.
38482 2017-01-05 Martin Liska <mliska@suse.cz>
38485 * gimplify.c (gimplify_decl_expr): Compare to
38486 asan_poisoned_variables instread of checking flags.
38487 (gimplify_target_expr): Likewise.
38488 (gimplify_expr): Likewise.
38489 (gimplify_function_tree): Conditionally initialize
38490 asan_poisoned_variables.
38492 2017-01-04 Jeff Law <law@redhat.com>
38494 PR tree-optimizatin/78812
38495 * rtl.h (contains_mem_rtx_p): Prototype.
38496 * ifcvt.c (containts_mem_rtx_p): Move from here to...
38497 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
38498 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
38499 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
38500 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
38502 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38504 * input.c (assert_char_at_range): Default-initialize actual_range.
38506 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38508 * df-scan.c (df_ref_create_structure): Make regno unsigned,
38509 to match the caller.
38511 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38513 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
38514 insns after final jump in test to emit dummy move.
38516 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38518 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
38519 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
38521 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38523 * multiple_target.c (create_dispatcher_calls): Init e_next.
38524 * tree-ssa-loop-split.c (split_loop): Init border.
38525 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
38528 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
38533 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
38534 (altivec_register_operand): Do not return true if the operand
38535 contains a SUBREG mixing SImode and SFmode.
38536 (vsx_register_operand): Likewise.
38537 (vsx_reg_sfsubreg_ok): New predicate.
38538 (vfloat_operand): Do not return true if the operand contains a
38539 SUBREG mixing SImode and SFmode.
38540 (vint_operand): Likewise.
38541 (vlogical_operand): Likewise.
38542 (gpc_reg_operand): Likewise.
38543 (int_reg_operand): Likewise.
38544 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
38545 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
38546 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
38548 (rs6000_emit_move_si_sf_subreg): New helper function.
38549 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
38550 fixup SUBREGs involving SImode and SFmode.
38551 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
38552 numbers for the new peephole2 optimization.
38553 (peephole2 for SFmode unions): New peephole2 to optimize cases in
38554 the GLIBC math library that do AND/IOR/XOR operations on single
38555 precision floating point.
38556 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
38557 target macros to say whether we need to avoid SUBREGs mixing
38559 (TARGET_ALLOW_SF_SUBREG): Likewise.
38560 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
38561 (UNSPEC_SI_FROM_SF): Likewise.
38562 (iorxor): Change spacing.
38563 (and_ior_xor): New iterator for AND, IOR, and XOR.
38564 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
38565 (movdi_from_sf_zero_ext): Likewise.
38566 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
38567 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
38568 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
38569 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
38570 (fms<mode>4): Likewise.
38571 (fnma<mode>4): Likewise.
38572 (fnms<mode>4): Likewise.
38573 (nfma<mode>4): Likewise.
38574 (nfms<mode>4): Likewise.
38576 2017-01-04 Marek Polacek <polacek@redhat.com>
38579 * doc/invoke.texi: Document -Wpointer-compare.
38581 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38583 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
38586 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
38587 descriptions for -gdwarf-5 and emit them as uleb128 instead of
38590 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
38593 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
38594 documentation of the powerpc_popcntb_ok attribute.
38595 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
38596 code to issue warning messages if a requested CPU configuration is
38597 not supported by the binary (assembler and loader) toolchain.
38598 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
38599 made to define a built-in function that has been disabled.
38600 (paired_init_builtins): Add assertion to prevent ICE if attempt is
38601 made to define a built-in function that has been disabled.
38602 (altivec_init_builtins): Add comment explaining why definition
38603 of the DST built-in functions is not preceded by an assertion
38604 check. Add assertions to prevent ICE if attempts are made to
38605 define an altivec predicate or an abs* built-in function that has
38607 (htm_init_builtins): Add comment explaining why definition of the
38608 htm built-in functions is not preceded by an assertion check.
38610 2017-01-04 Jeff Law <law@redhat.com>
38612 PR tree-optimizatin/67955
38613 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
38614 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
38615 the points-to solution does not include pt_null. Use DECL_PT_UID
38618 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
38620 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
38621 Use gen_int_mode instead of gen_lopwart for const_int operands.
38623 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38625 PR tree-optimization/71563
38626 * match.pd: Simplify X << Y into X if Y is known to be 0 or
38627 out of range value - has low bits known to be zero.
38629 2017-01-04 Alan Modra <amodra@gmail.com>
38631 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
38632 * configure: Regenerate.
38633 * config.in: Regenerate.
38635 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38638 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
38639 a substring of the message, but strcmp with the whole message. Ifdef
38640 ENABLE_NLS, translate the message first using dgettext.
38642 2017-01-03 Jeff Law <law@redhat.com>
38644 PR tree-optimizatin/78856
38645 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
38646 (mark_threaded_blocks): Remove code to truncate thread paths that
38647 cross multiple loop headers. Instead invalidate the cached loop
38648 iteration information and handle case of a thread path walking
38649 into an irreducible region.
38651 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
38654 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
38655 assertions. Add support for doing the signbit if the IEEE 128-bit
38656 floating point value is in a GPR.
38657 * config/rs6000/rs6000.md (Fsignbit): Delete.
38658 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
38659 Update the length attribute if the value is in a GPR.
38660 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
38661 the sign or zero extension instruction, since the value is always 0/1.
38662 (signbit<mode>2_dm2): Delete using <Fsignbit>.
38665 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
38666 extracting SImode to a GPR register so that we can generate a
38667 store, limit the vector to be in a traditional Altivec register
38668 for the vextuwrx instruction.
38670 2017-01-03 Ian Lance Taylor <iant@google.com>
38672 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
38674 2017-01-03 Martin Sebor <msebor@redhat.com>
38676 PR tree-optimization/78696
38677 * gimple-ssa-sprintf.c (format_floating): Correct handling of
38678 precision. Use MPFR for %f for greater fidelity. Correct handling
38680 (pass_sprintf_length::compute_format_length): Set width and precision
38681 specified by asrerisk to void_node for vararg functions.
38682 (try_substitute_return_value): Adjust dump output.
38684 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
38686 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
38688 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
38690 * doc/invoke.texi (SPARC options): Document -mlra as the default.
38691 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
38692 -mlra/-mno-lra was passed to the compiler.
38694 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
38696 PR rtl-optimization/65618
38697 * emit-rtl.c (try_split): Move initialization of "before" and
38698 "after" to just before the call to emit_insn_after_setloc.
38700 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
38702 * doc/md.texi (Standard Names): Remove reference to Java frontend.
38704 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
38706 * dwarf2out.c (gen_enumeration_type_die): When
38707 -gno-strict-dwarf, add a DW_AT_encoding attribute.
38709 2017-01-03 Jakub Jelinek <jakub@redhat.com>
38711 PR tree-optimization/78965
38712 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
38713 Change first argument from const call_info & to call_info &. For %n
38714 set info.nowrite to false.
38716 PR middle-end/78901
38717 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
38718 possibly throwing calls.
38720 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
38721 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
38722 and fns handling, rather than in a separate case SSA_NAME.
38724 2017-01-02 Jeff Law <law@redhat.com>
38726 * config/darwin-driver.c (darwin_driver_init): Const-correctness
38727 fixes for first_period and second_period variables.
38729 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
38732 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
38733 (*insvqi_1): New insn pattern.
38734 (*insvqi_1_mem_rex64): Ditto.
38735 (*insvqi_2): Ditto.
38736 (*insvqi_3): Rename from *insvqi.
38738 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
38740 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
38742 * doc/cfg.texi (Edges): Remove reference to Java.
38743 (Maintaining the CFG): Ditto.
38745 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38747 PR middle-end/77674
38748 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
38749 transparent aliases.
38751 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38753 PR middle-end/77484
38754 * predict.def (PRED_CALL): Update hitrate.
38755 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
38756 * predict.c (tree_estimate_probability_bb): Split CALL predictor
38757 into direct/indirect/polymorphic variants.
38759 2017-01-01 Jakub Jelinek <jakub@redhat.com>
38761 Update copyright years.
38763 * gcc.c (process_command): Update copyright notice dates.
38764 * gcov-dump.c (print_version): Ditto.
38765 * gcov.c (print_version): Ditto.
38766 * gcov-tool.c (print_version): Ditto.
38767 * gengtype.c (create_file): Ditto.
38768 * doc/cpp.texi: Bump @copying's copyright year.
38769 * doc/cppinternals.texi: Ditto.
38770 * doc/gcc.texi: Ditto.
38771 * doc/gccint.texi: Ditto.
38772 * doc/gcov.texi: Ditto.
38773 * doc/install.texi: Ditto.
38774 * doc/invoke.texi: Ditto.
38776 Copyright (C) 2017 Free Software Foundation, Inc.
38778 Copying and distribution of this file, with or without modification,
38779 are permitted in any medium without royalty provided the copyright
38780 notice and this notice are preserved.